d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
a9d4fe977a
freebsd-skq/sys/kern/kern_sig.c

4251 lines
102 KiB
C
Raw Normal View History

/* -> /*- for copyright notices, minor format tweaks as necessary
2005-01-06 23:35:40 +00:00
/*-
sys: further adoption of SPDX licensing ID tags. Mainly focus on files that use BSD 3-Clause license. The Software Package Data Exchange (SPDX) group provides a specification to make it easier for automated tools to detect and summarize well known opensource licenses. We are gradually adopting the specification, noting that the tags are considered only advisory and do not, in any way, superceed or replace the license texts. Special thanks to Wind River for providing access to "The Duke of Highlander" tool: an older (2014) run over FreeBSD tree was useful as a starting point.
2017-11-20 19:43:44 +00:00
* SPDX-License-Identifier: BSD-3-Clause
*
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* Copyright (c) 1982, 1986, 1989, 1991, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
Renumber license clauses in sys/kern to avoid skipping #3
2016-09-15 13:16:20 +00:00
* 3. Neither the name of the University nor the names of its contributors
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)kern_sig.c 8.7 (Berkeley) 4/18/94
*/
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Converted two options over to the new scheme: USER_LDT and KTRACE.
1996-01-03 21:42:35 +00:00
#include "opt_ktrace.h"
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
Sanitise the coredump file names sent to devd. While there, add a sysctl to turn this feature off as requested by kib@.
2015-02-10 04:34:39 +00:00
#include <sys/ctype.h>
Move the include of <sys/systm.h> so that KTR gets a declaration for snprintf().
2000-09-10 13:54:52 +00:00
#include <sys/systm.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/signalvar.h>
#include <sys/vnode.h>
#include <sys/acct.h>
Update kernel inclusions of capability.h to use capsicum.h instead; some further refinement is required as some device drivers intended to be portable over FreeBSD versions rely on __FreeBSD_version to decide whether to include capability.h. MFC after: 3 weeks
2014-03-16 10:55:57 +00:00
#include <sys/capsicum.h>
Generalize the gzio API. We currently use a set of subroutines in kern_gzio.c to perform compression of user and kernel core dumps. In the interest of adding support for other compression algorithms (zstd) in this role without complicating the API consumers, add a simple compressor API which can be used to select an algorithm. Also change the (non-default) GZIO kernel option to not enable compressed user cores by default. It's not clear that such a default would be desirable with support for multiple algorithms implemented, and it's inconsistent in that it isn't applied to kernel dumps. Reviewed by: cem Differential Revision: https://reviews.freebsd.org/D13632
2018-01-08 21:27:41 +00:00
#include <sys/compressor.h>
Implement condition variables.
2001-01-16 01:00:43 +00:00
#include <sys/condvar.h>
Use devctl.h instead of bus.h to reduce newbus pollution. There's no need for these parts of the kernel to know about newbus, so narrow what is included to devctl.h for device_notify_*. Suggested by: kib@
2020-08-21 00:03:24 +00:00
#include <sys/devctl.h>
Mostly sort the includes.
2003-05-05 21:26:25 +00:00
#include <sys/event.h>
#include <sys/fcntl.h>
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
#include <sys/imgact.h>
Mostly sort the includes.
2003-05-05 21:26:25 +00:00
#include <sys/kernel.h>
Major update to the way synchronization is done in the kernel. Highlights include: * Mutual exclusion is used instead of spl*(). See mutex(9). (Note: The alpha port is still in transition and currently uses both.) * Per-CPU idle processes. * Interrupts are run in their own separate kernel threads and can be preempted (i386 only). Partially contributed by: BSDi (BSD/OS) Submissions by (at least): cp, dfr, dillon, grog, jake, jhb, sheldonh
2000-09-07 01:33:02 +00:00
#include <sys/ktr.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/ktrace.h>
Avoid relying on header pollution from sys/refcount.h. MFC after: 3 days Sponsored by: The FreeBSD Foundation
2019-07-29 20:26:01 +00:00
#include <sys/limits.h>
Mostly sort the includes.
2003-05-05 21:26:25 +00:00
#include <sys/lock.h>
#include <sys/malloc.h>
#include <sys/mutex.h>
Manage struct sigacts refcnt with atomics instead of a mutex. MFC after: 1 week
2014-07-14 21:12:59 +00:00
#include <sys/refcount.h>
Mostly sort the includes.
2003-05-05 21:26:25 +00:00
#include <sys/namei.h>
#include <sys/proc.h>
Add experimental support for process descriptors A "process descriptor" file descriptor is used to manage processes without using the PID namespace. This is required for Capsicum's Capability Mode, where the PID namespace is unavailable. New system calls pdfork(2) and pdkill(2) offer the functional equivalents of fork(2) and kill(2). pdgetpid(2) allows querying the PID of the remote process for debugging purposes. The currently-unimplemented pdwait(2) will, in the future, allow querying rusage/exit status. In the interim, poll(2) may be used to check (and wait for) process termination. When a process is referenced by a process descriptor, it does not issue SIGCHLD to the parent, making it suitable for use in libraries---a common scenario when using library compartmentalisation from within large applications (such as web browsers). Some observers may note a similarity to Mach task ports; process descriptors provide a subset of this behaviour, but in a UNIX style. This feature is enabled by "options PROCDESC", but as with several other Capsicum kernel features, is not enabled by default in GENERIC 9.0. Reviewed by: jhb, kib Approved by: re (kib), mentor (rwatson) Sponsored by: Google Inc
2011-08-18 22:51:30 +00:00
#include <sys/procdesc.h>
Requested and tested by: kevans Reviewed by: kevans (previous version), markj Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D22546
2019-11-27 20:33:53 +00:00
#include <sys/ptrace.h>
Merge posix4/* into normal kernel hierarchy. Reviewed by: glanced at by jhb Approved by: silence on -arch@ and -standards@
2006-11-11 16:26:58 +00:00
#include <sys/posix4.h>
Fix support for RACCT_CORE by merging forgotten file.
2011-05-26 18:54:07 +00:00
#include <sys/racct.h>
- Resort some includes to deal with the new witness code coming in shortly. - Make sure we have Giant locked before calling coredump() in sigexit(). Spotted by: peter (2)
2001-03-28 08:41:04 +00:00
#include <sys/resourcevar.h>
Add DTrace 'proc' provider probes using the Statically Defined Trace (sdt) mechanism.
2008-05-24 06:22:16 +00:00
#include <sys/sdt.h>
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
#include <sys/sbuf.h>
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
#include <sys/sleepqueue.h>
Overhaul of the SMP code. Several portions of the SMP kernel support have been made machine independent and various other adjustments have been made to support Alpha SMP. - It splits the per-process portions of hardclock() and statclock() off into hardclock_process() and statclock_process() respectively. hardclock() and statclock() call the *_process() functions for the current process so that UP systems will run as before. For SMP systems, it is simply necessary to ensure that all other processors execute the *_process() functions when the main clock functions are triggered on one CPU by an interrupt. For the alpha 4100, clock interrupts are delievered in a staggered broadcast fashion, so we simply call hardclock/statclock on the boot CPU and call the *_process() functions on the secondaries. For x86, we call statclock and hardclock as usual and then call forward_hardclock/statclock in the MD code to send an IPI to cause the AP's to execute forwared_hardclock/statclock which then call the *_process() functions. - forward_signal() and forward_roundrobin() have been reworked to be MI and to involve less hackery. Now the cpu doing the forward sets any flags, etc. and sends a very simple IPI_AST to the other cpu(s). AST IPIs now just basically return so that they can execute ast() and don't bother with setting the astpending or needresched flags themselves. This also removes the loop in forward_signal() as sched_lock closes the race condition that the loop worked around. - need_resched(), resched_wanted() and clear_resched() have been changed to take a process to act on rather than assuming curproc so that they can be used to implement forward_roundrobin() as described above. - Various other SMP variables have been moved to a MI subr_smp.c and a new header sys/smp.h declares MI SMP variables and API's. The IPI API's from machine/ipl.h have moved to machine/smp.h which is included by sys/smp.h. - The globaldata_register() and globaldata_find() functions as well as the SLIST of globaldata structures has become MI and moved into subr_smp.c. Also, the globaldata list is only available if SMP support is compiled in. Reviewed by: jake, peter Looked over by: eivind
2001-04-27 19:28:25 +00:00
#include <sys/smp.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/stat.h>
Convert the allproc and proctree locks from lockmgr locks to sx locks.
2001-03-28 11:52:56 +00:00
#include <sys/sx.h>
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
#include <sys/syscallsubr.h>
Added a sysctl variable kern.sugid_coredump for controlling coredump behavior of setuid/setgid binaries that defaults to 0 (coredump disabled).
1998-06-28 08:37:45 +00:00
#include <sys/sysctl.h>
Mostly sort the includes.
2003-05-05 21:26:25 +00:00
#include <sys/sysent.h>
#include <sys/syslog.h>
#include <sys/sysproto.h>
Let itimer store itimerspec instead of itimerval, so I don't have to convert to or from timeval frequently. Introduce function itimer_accept() to ack a timer signal in signal acceptance code, this allows us to return more fresh overrun counter than at signal generating time. while POSIX says: "the value returned by timer_getoverrun() shall apply to the most recent expiration signal delivery or acceptance for the timer,.." I prefer returning it at acceptance time. Introduce SIGEV_THREAD_ID notification mode, it is used by thread libary to request kernel to deliver signal to a specified thread, and in turn, the thread library may use the mechanism to implement SIGEV_THREAD which is required by POSIX. Timer signal is managed by timer code, so it can not fail even if signal queue is full filled by sigqueue syscall.
2005-10-30 02:56:08 +00:00
#include <sys/timers.h>
This brings in a Yahoo coredump patch from Paul, with additional mods by me (addition of vn_rdwr_inchunks). The problem Yahoo is solving is that if you have large process images core dumping, or you have a large number of forked processes all core dumping at the same time, the original coredump code would leave the vnode locked throughout. This can cause the directory vnode to get locked up, which can cause the parent directory vnode to get locked up, and so on all the way to the root node, locking the entire machine up for extremely long periods of time. This patch solves the problem in two ways. First it uses an advisory non-blocking lock to abort multiple processes trying to core to the same file. Second (my contribution) it chunks up the writes and uses bwillwrite() to avoid holding the vnode locked while blocking in the buffer cache. Submitted by: ps Reviewed by: dillon MFC after: 2 weeks
2001-09-08 20:02:33 +00:00
#include <sys/unistd.h>
Mostly sort the includes.
2003-05-05 21:26:25 +00:00
#include <sys/wait.h>
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
#include <vm/vm.h>
#include <vm/vm_extern.h>
#include <vm/uma.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
#include <sys/jail.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <machine/cpu.h>
Audit the arguments to the kill(2) and killpg(2) system calls. Obtained from: TrustedBSD Project Approved by: rwatson (mentor)
2006-02-14 01:17:03 +00:00
#include <security/audit/audit.h>
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
#define ONSIG 32 /* NSIG for osig* syscalls. XXX. */
Add DTrace 'proc' provider probes using the Statically Defined Trace (sdt) mechanism.
2008-05-24 06:22:16 +00:00
SDT_PROVIDER_DECLARE(proc);
Fix style issues around existing SDT probes. - Use SDT_PROBE<N>() instead of SDT_PROBE(). This has no functional effect at the moment, but will be needed for some future changes. - Don't hardcode the module component of the probe identifier. This is set automatically by the SDT framework. MFC after: 1 week
2015-12-16 23:39:27 +00:00
SDT_PROBE_DEFINE3(proc, , , signal__send,
"struct thread *", "struct proc *", "int");
SDT_PROBE_DEFINE2(proc, , , signal__clear,
"int", "ksiginfo_t *");
SDT_PROBE_DEFINE3(proc, , , signal__discard,
Specify SDT probe argument types in the probe definition itself rather than using SDT_PROBE_ARGTYPE(). This will make it easy to extend the SDT(9) API to allow probes with dynamically-translated types. There is no functional change. MFC after: 2 weeks
2013-08-15 04:08:55 +00:00
"struct thread *", "struct proc *", "int");
Add DTrace 'proc' provider probes using the Statically Defined Trace (sdt) mechanism.
2008-05-24 06:22:16 +00:00
White space commit. I'm working on this file but I wanted to make the whitespece commit separatly.
2002-07-03 06:15:26 +00:00
static int coredump(struct thread *);
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
static int killpg1(struct thread *td, int sig, int pgid, int all,
ksiginfo_t *ksi);
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
static int issignal(struct thread *td);
Return reschedule_signals() to being static again. It was used after sigfastblock_setpend() call in in ast() when current thread fast-blocks signals. Add a flag to sigfastblock_setpend() to request reschedule, and remove the direct use of the function from subr_trap.c Tested by: pho Sponsored by: The FreeBSD Foundation
2020-03-10 20:04:38 +00:00
static void reschedule_signals(struct proc *p, sigset_t block, int flags);
White space commit. I'm working on this file but I wanted to make the whitespece commit separatly.
2002-07-03 06:15:26 +00:00
static int sigprop(int sig);
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
static void tdsigwakeup(struct thread *, int, sig_t, int);
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
static int sig_suspend_threads(struct thread *, struct proc *, int);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
static int filt_sigattach(struct knote *kn);
static void filt_sigdetach(struct knote *kn);
static int filt_signal(struct knote *kn, long hint);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
static struct thread *sigtd(struct proc *p, int sig, bool fast_sigblock);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
static void sigqueue_start(void);
static uma_zone_t ksiginfo_zone = NULL;
Use C99 initialization for struct filterops. Obtained from: Mac OS X Sponsored by: Apple Inc. MFC after: 3 weeks
2009-09-12 20:03:45 +00:00
struct filterops sig_filtops = {
.f_isfd = 0,
.f_attach = filt_sigattach,
.f_detach = filt_sigdetach,
.f_event = filt_signal,
};
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
- Various style and whitespace fixes. - Make sugid_coredump and kern_logsigexit private to kern_sig.c. Submitted by: bde (partially) MFC after: 1 month
2010-07-08 19:15:26 +00:00
static int kern_logsigexit = 1;
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
SYSCTL_INT(_kern, KERN_LOGSIGEXIT, logsigexit, CTLFLAG_RW,
&kern_logsigexit, 0,
Add sysctl descriptions to many SYSCTL_XXXs PR: kern/11197 Submitted by: Adrian Chadd <adrian@FreeBSD.org> Reviewed by: billf(spelling/style/minor nits) Looked at by: bde(style)
1999-05-03 23:57:32 +00:00
"Log processes quitting on abnormal signals to syslog(3)");
Make the logging of abnormally exiting processes optional by a sysctl. PR: kern/1711 Submitted by: Nick Sayer <nsayer@kfu.com>
1998-07-28 22:34:12 +00:00
Add a sysctl to force a process to sigexit if a trap signal is being hold by current thread or ignored by current process, otherwise, it is very possible the thread will enter an infinite loop and lead to an administrator's nightmare.
2005-12-09 08:29:29 +00:00
static int kern_forcesigexit = 1;
SYSCTL_INT(_kern, OID_AUTO, forcesigexit, CTLFLAG_RW,
&kern_forcesigexit, 0, "Force trap signal to be handled");
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
static SYSCTL_NODE(_kern, OID_AUTO, sigqueue, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
Mark all SYSCTL_NODEs static that have no corresponding SYSCTL_DECLs. The SYSCTL_NODE macro defines a list that stores all child-elements of that node. If there's no SYSCTL_DECL macro anywhere else, there's no reason why it shouldn't be static.
2011-11-07 15:43:11 +00:00
"POSIX real time signal");
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
static int max_pending_per_proc = 128;
SYSCTL_INT(_kern_sigqueue, OID_AUTO, max_pending_per_proc, CTLFLAG_RW,
&max_pending_per_proc, 0, "Max pending signals per proc");
static int preallocate_siginfo = 1024;
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern_sigqueue, OID_AUTO, preallocate, CTLFLAG_RDTUN,
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
&preallocate_siginfo, 0, "Preallocated signal memory size");
static int signal_overflow = 0;
Cleanup sigqueue sysctl.
2005-12-09 02:26:44 +00:00
SYSCTL_INT(_kern_sigqueue, OID_AUTO, overflow, CTLFLAG_RD,
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
&signal_overflow, 0, "Number of signals overflew");
static int signal_alloc_fail = 0;
Cleanup sigqueue sysctl.
2005-12-09 02:26:44 +00:00
SYSCTL_INT(_kern_sigqueue, OID_AUTO, alloc_fail, CTLFLAG_RD,
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
&signal_alloc_fail, 0, "signals failed to be allocated");
Print unimplemented syscall number to the ctty on SIGSYS, if enabled by the knob kern.lognosys. Discussed with: imp Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 3 weeks X-Differential revision: https://reviews.freebsd.org/D11080
2017-06-12 21:11:11 +00:00
static int kern_lognosys = 0;
SYSCTL_INT(_kern, OID_AUTO, lognosys, CTLFLAG_RWTUN, &kern_lognosys, 0,
"Log invalid syscalls");
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
__read_frequently bool sigfastblock_fetch_always = false;
SYSCTL_BOOL(_kern, OID_AUTO, sigfastblock_fetch_always, CTLFLAG_RWTUN,
&sigfastblock_fetch_always, 0,
"Fetch sigfastblock word on each syscall entry for proper "
"blocking semantic");
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
SYSINIT(signal, SI_SUB_P1003_1B, SI_ORDER_FIRST+3, sigqueue_start, NULL);
o Revert kern_sig.c#1.143, as cr_cansignal() doesn't currently permit a number of desirable cases in which SIGIO/SIGURG are delivered. We'll keep tweaking. Reported by: Alexander Kabaev <ak03@gte.com>
2002-01-10 01:25:35 +00:00
/*
* Policy -- Can ucred cr1 send SIGIO to process cr2?
* Should use cr_cansignal() once cr_cansignal() allows SIGIO and SIGURG
* in the right situations.
*/
#define CANSIGIO(cr1, cr2) \
((cr1)->cr_uid == 0 || \
(cr1)->cr_ruid == (cr2)->cr_ruid || \
(cr1)->cr_uid == (cr2)->cr_ruid || \
(cr1)->cr_ruid == (cr2)->cr_uid || \
(cr1)->cr_uid == (cr2)->cr_uid)
- Various style and whitespace fixes. - Make sugid_coredump and kern_logsigexit private to kern_sig.c. Submitted by: bde (partially) MFC after: 1 month
2010-07-08 19:15:26 +00:00
static int sugid_coredump;
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern, OID_AUTO, sugid_coredump, CTLFLAG_RWTUN,
Improve the description of sysctl "kern.sugid_coredump". Submitted by: Mel Flynn <mel.flynn+fbsd.hackers at mailing.thruhere.net> on -hackers
2009-10-12 15:49:48 +00:00
&sugid_coredump, 0, "Allow setuid and setgid processes to dump core");
Added a sysctl variable kern.sugid_coredump for controlling coredump behavior of setuid/setgid binaries that defaults to 0 (coredump disabled).
1998-06-28 08:37:45 +00:00
Add kern.capmode_coredump sysctl/tunable to allow processes in capability mode to dump core. Reviewed by: rwatson Obtained from: WHEEL Systems MFC after: 2 weeks
2012-11-27 10:38:11 +00:00
static int capmode_coredump;
Pull in r267961 and r267973 again. Fix for issues reported will follow.
2014-06-28 03:56:17 +00:00
SYSCTL_INT(_kern, OID_AUTO, capmode_coredump, CTLFLAG_RWTUN,
Add kern.capmode_coredump sysctl/tunable to allow processes in capability mode to dump core. Reviewed by: rwatson Obtained from: WHEEL Systems MFC after: 2 weeks
2012-11-27 10:38:11 +00:00
&capmode_coredump, 0, "Allow processes in capability mode to dump core");
Add sysctl kern.coredump to enable/disable core dumps system wide.
2000-03-21 07:10:42 +00:00
static int do_coredump = 1;
SYSCTL_INT(_kern, OID_AUTO, coredump, CTLFLAG_RW,
&do_coredump, 0, "Enable/Disable coredumps");
add option to automaticly mark core dumps with the nodump flag PR: 57065 Submitted by: Walter C. Pelissero
2004-08-09 05:46:46 +00:00
static int set_core_nodump_flag = 0;
SYSCTL_INT(_kern, OID_AUTO, nodump_coredump, CTLFLAG_RW, &set_core_nodump_flag,
0, "Enable setting the NODUMP flag on coredump files");
Disable coredump_devctl because it could lead to leaking paths to jails.
2015-03-24 02:17:17 +00:00
static int coredump_devctl = 0;
Sanitise the coredump file names sent to devd. While there, add a sysctl to turn this feature off as requested by kib@.
2015-02-10 04:34:39 +00:00
SYSCTL_INT(_kern, OID_AUTO, coredump_devctl, CTLFLAG_RW, &coredump_devctl,
0, "Generate a devctl notification when processes coredump");
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
/*
* Signal properties and actions.
* The array below categorizes the signals and their default actions
* according to the following properties:
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
#define SIGPROP_KILL 0x01 /* terminates process by default */
#define SIGPROP_CORE 0x02 /* ditto and coredumps */
#define SIGPROP_STOP 0x04 /* suspend process */
#define SIGPROP_TTYSTOP 0x08 /* ditto, from tty */
#define SIGPROP_IGNORE 0x10 /* ignore by default */
#define SIGPROP_CONT 0x20 /* continue if suspended */
#define SIGPROP_CANTMASK 0x40 /* non-maskable, catchable */
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
static int sigproptbl[NSIG] = {
Modernize the initalization of sigproptbl. Use C99 designators to set the value of each slot and the nitems macro to check for valid entries. In the process, switch to indexing by signal number rather than signal-1 for improved clarity. Obtained from: CheriBSD (a6053c5abf03a5f53bbfcdd3a26429383f67e09f) Sponsored by: DARPA, AFRL Reviewed by: kib
2016-09-06 22:03:53 +00:00
[SIGHUP] = SIGPROP_KILL,
[SIGINT] = SIGPROP_KILL,
[SIGQUIT] = SIGPROP_KILL | SIGPROP_CORE,
[SIGILL] = SIGPROP_KILL | SIGPROP_CORE,
[SIGTRAP] = SIGPROP_KILL | SIGPROP_CORE,
[SIGABRT] = SIGPROP_KILL | SIGPROP_CORE,
[SIGEMT] = SIGPROP_KILL | SIGPROP_CORE,
[SIGFPE] = SIGPROP_KILL | SIGPROP_CORE,
[SIGKILL] = SIGPROP_KILL,
[SIGBUS] = SIGPROP_KILL | SIGPROP_CORE,
[SIGSEGV] = SIGPROP_KILL | SIGPROP_CORE,
[SIGSYS] = SIGPROP_KILL | SIGPROP_CORE,
[SIGPIPE] = SIGPROP_KILL,
[SIGALRM] = SIGPROP_KILL,
[SIGTERM] = SIGPROP_KILL,
[SIGURG] = SIGPROP_IGNORE,
[SIGSTOP] = SIGPROP_STOP,
[SIGTSTP] = SIGPROP_STOP | SIGPROP_TTYSTOP,
[SIGCONT] = SIGPROP_IGNORE | SIGPROP_CONT,
[SIGCHLD] = SIGPROP_IGNORE,
[SIGTTIN] = SIGPROP_STOP | SIGPROP_TTYSTOP,
[SIGTTOU] = SIGPROP_STOP | SIGPROP_TTYSTOP,
[SIGIO] = SIGPROP_IGNORE,
[SIGXCPU] = SIGPROP_KILL,
[SIGXFSZ] = SIGPROP_KILL,
[SIGVTALRM] = SIGPROP_KILL,
[SIGPROF] = SIGPROP_KILL,
[SIGWINCH] = SIGPROP_IGNORE,
[SIGINFO] = SIGPROP_IGNORE,
[SIGUSR1] = SIGPROP_KILL,
[SIGUSR2] = SIGPROP_KILL,
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
};
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
sigset_t fastblock_mask;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
static void
sigqueue_start(void)
{
ksiginfo_zone = uma_zcreate("ksiginfo", sizeof(ksiginfo_t),
NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0);
uma_prealloc(ksiginfo_zone, preallocate_siginfo);
set signal queue values for sysconf().
2005-12-01 00:25:50 +00:00
p31b_setcfg(CTL_P1003_1B_REALTIME_SIGNALS, _POSIX_REALTIME_SIGNALS);
p31b_setcfg(CTL_P1003_1B_RTSIG_MAX, SIGRTMAX - SIGRTMIN + 1);
p31b_setcfg(CTL_P1003_1B_SIGQUEUE_MAX, max_pending_per_proc);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
SIGFILLSET(fastblock_mask);
SIG_CANTMASK(fastblock_mask);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
}
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksiginfo_t *
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
ksiginfo_alloc(int wait)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
int flags;
flags = M_ZERO;
if (! wait)
flags |= M_NOWAIT;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
if (ksiginfo_zone != NULL)
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
return ((ksiginfo_t *)uma_zalloc(ksiginfo_zone, flags));
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (NULL);
}
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
void
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_free(ksiginfo_t *ksi)
{
uma_zfree(ksiginfo_zone, ksi);
}
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
static __inline int
ksiginfo_tryfree(ksiginfo_t *ksi)
{
if (!(ksi->ksi_flags & KSI_EXT)) {
uma_zfree(ksiginfo_zone, ksi);
return (1);
}
return (0);
}
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
void
sigqueue_init(sigqueue_t *list, struct proc *p)
{
SIGEMPTYSET(list->sq_signals);
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
SIGEMPTYSET(list->sq_kill);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
SIGEMPTYSET(list->sq_ptrace);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
TAILQ_INIT(&list->sq_list);
list->sq_proc = p;
list->sq_flags = SQ_INIT;
}
/*
* Get a signal's ksiginfo.
* Return:
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
* 0 - signal not found
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
* others - signal number
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
*/
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static int
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_get(sigqueue_t *sq, int signo, ksiginfo_t *si)
{
struct proc *p = sq->sq_proc;
struct ksiginfo *ksi, *next;
int count = 0;
KASSERT(sq->sq_flags & SQ_INIT, ("sigqueue not inited"));
if (!SIGISMEMBER(sq->sq_signals, signo))
return (0);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if (SIGISMEMBER(sq->sq_ptrace, signo)) {
count++;
SIGDELSET(sq->sq_ptrace, signo);
si->ksi_flags |= KSI_PTRACE;
}
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
if (SIGISMEMBER(sq->sq_kill, signo)) {
count++;
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if (count == 1)
SIGDELSET(sq->sq_kill, signo);
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
}
Use macro TAILQ_FOREACH_SAFE instead of expanding it.
2006-10-22 00:09:41 +00:00
TAILQ_FOREACH_SAFE(ksi, &sq->sq_list, ksi_link, next) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
if (ksi->ksi_signo == signo) {
if (count == 0) {
TAILQ_REMOVE(&sq->sq_list, ksi, ksi_link);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksi->ksi_sigq = NULL;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_copy(ksi, si);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
if (ksiginfo_tryfree(ksi) && p != NULL)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
p->p_pendingcnt--;
}
break loop early if we know that there are at least two signals.
2006-12-25 03:00:15 +00:00
if (++count > 1)
break;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
}
}
if (count <= 1)
SIGDELSET(sq->sq_signals, signo);
si->ksi_signo = signo;
return (signo);
}
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
void
sigqueue_take(ksiginfo_t *ksi)
{
struct ksiginfo *kp;
struct proc *p;
sigqueue_t *sq;
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
if (ksi == NULL || (sq = ksi->ksi_sigq) == NULL)
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
return;
p = sq->sq_proc;
TAILQ_REMOVE(&sq->sq_list, ksi, ksi_link);
ksi->ksi_sigq = NULL;
if (!(ksi->ksi_flags & KSI_EXT) && p != NULL)
p->p_pendingcnt--;
for (kp = TAILQ_FIRST(&sq->sq_list); kp != NULL;
kp = TAILQ_NEXT(kp, ksi_link)) {
if (kp->ksi_signo == ksi->ksi_signo)
break;
}
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if (kp == NULL && !SIGISMEMBER(sq->sq_kill, ksi->ksi_signo) &&
!SIGISMEMBER(sq->sq_ptrace, ksi->ksi_signo))
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
SIGDELSET(sq->sq_signals, ksi->ksi_signo);
}
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static int
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_add(sigqueue_t *sq, int signo, ksiginfo_t *si)
{
struct proc *p = sq->sq_proc;
struct ksiginfo *ksi;
int ret = 0;
KASSERT(sq->sq_flags & SQ_INIT, ("sigqueue not inited"));
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
/*
* SIGKILL/SIGSTOP cannot be caught or masked, so take the fast path
* for these signals.
*/
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
if (signo == SIGKILL || signo == SIGSTOP || si == NULL) {
SIGADDSET(sq->sq_kill, signo);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
goto out_set_bit;
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
}
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
/* directly insert the ksi, don't copy it */
if (si->ksi_flags & KSI_INS) {
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
if (si->ksi_flags & KSI_HEAD)
TAILQ_INSERT_HEAD(&sq->sq_list, si, ksi_link);
else
TAILQ_INSERT_TAIL(&sq->sq_list, si, ksi_link);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
si->ksi_sigq = sq;
goto out_set_bit;
}
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
if (__predict_false(ksiginfo_zone == NULL)) {
SIGADDSET(sq->sq_kill, signo);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
goto out_set_bit;
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
}
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
if (p != NULL && p->p_pendingcnt >= max_pending_per_proc) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
signal_overflow++;
ret = EAGAIN;
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
} else if ((ksi = ksiginfo_alloc(0)) == NULL) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
signal_alloc_fail++;
ret = EAGAIN;
} else {
if (p != NULL)
p->p_pendingcnt++;
ksiginfo_copy(si, ksi);
ksi->ksi_signo = signo;
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
if (si->ksi_flags & KSI_HEAD)
TAILQ_INSERT_HEAD(&sq->sq_list, ksi, ksi_link);
else
TAILQ_INSERT_TAIL(&sq->sq_list, ksi, ksi_link);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksi->ksi_sigq = sq;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
}
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if (ret != 0) {
if ((si->ksi_flags & KSI_PTRACE) != 0) {
SIGADDSET(sq->sq_ptrace, signo);
ret = 0;
goto out_set_bit;
} else if ((si->ksi_flags & KSI_TRAP) != 0 ||
(si->ksi_flags & KSI_SIGQ) == 0) {
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
SIGADDSET(sq->sq_kill, signo);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
ret = 0;
goto out_set_bit;
}
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (ret);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
}
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
out_set_bit:
SIGADDSET(sq->sq_signals, signo);
return (ret);
}
void
sigqueue_flush(sigqueue_t *sq)
{
struct proc *p = sq->sq_proc;
ksiginfo_t *ksi;
KASSERT(sq->sq_flags & SQ_INIT, ("sigqueue not inited"));
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
if (p != NULL)
PROC_LOCK_ASSERT(p, MA_OWNED);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
while ((ksi = TAILQ_FIRST(&sq->sq_list)) != NULL) {
TAILQ_REMOVE(&sq->sq_list, ksi, ksi_link);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksi->ksi_sigq = NULL;
if (ksiginfo_tryfree(ksi) && p != NULL)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
p->p_pendingcnt--;
}
SIGEMPTYSET(sq->sq_signals);
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
SIGEMPTYSET(sq->sq_kill);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
SIGEMPTYSET(sq->sq_ptrace);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
}
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static void
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
sigqueue_move_set(sigqueue_t *src, sigqueue_t *dst, const sigset_t *set)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
sigset_t tmp;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
struct proc *p1, *p2;
ksiginfo_t *ksi, *next;
KASSERT(src->sq_flags & SQ_INIT, ("src sigqueue not inited"));
KASSERT(dst->sq_flags & SQ_INIT, ("dst sigqueue not inited"));
p1 = src->sq_proc;
p2 = dst->sq_proc;
/* Move siginfo to target list */
Use macro TAILQ_FOREACH_SAFE instead of expanding it.
2006-10-22 00:09:41 +00:00
TAILQ_FOREACH_SAFE(ksi, &src->sq_list, ksi_link, next) {
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
if (SIGISMEMBER(*set, ksi->ksi_signo)) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
TAILQ_REMOVE(&src->sq_list, ksi, ksi_link);
if (p1 != NULL)
p1->p_pendingcnt--;
TAILQ_INSERT_TAIL(&dst->sq_list, ksi, ksi_link);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksi->ksi_sigq = dst;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
if (p2 != NULL)
p2->p_pendingcnt++;
}
}
/* Move pending bits to target list */
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
tmp = src->sq_kill;
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
SIGSETAND(tmp, *set);
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
SIGSETOR(dst->sq_kill, tmp);
SIGSETNAND(src->sq_kill, tmp);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
tmp = src->sq_ptrace;
SIGSETAND(tmp, *set);
SIGSETOR(dst->sq_ptrace, tmp);
SIGSETNAND(src->sq_ptrace, tmp);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
tmp = src->sq_signals;
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
SIGSETAND(tmp, *set);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
SIGSETOR(dst->sq_signals, tmp);
SIGSETNAND(src->sq_signals, tmp);
}
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
#if 0
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static void
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_move(sigqueue_t *src, sigqueue_t *dst, int signo)
{
sigset_t set;
SIGEMPTYSET(set);
SIGADDSET(set, signo);
sigqueue_move_set(src, dst, &set);
}
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
#endif
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static void
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
sigqueue_delete_set(sigqueue_t *sq, const sigset_t *set)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
struct proc *p = sq->sq_proc;
ksiginfo_t *ksi, *next;
KASSERT(sq->sq_flags & SQ_INIT, ("src sigqueue not inited"));
/* Remove siginfo queue */
Use macro TAILQ_FOREACH_SAFE instead of expanding it.
2006-10-22 00:09:41 +00:00
TAILQ_FOREACH_SAFE(ksi, &sq->sq_list, ksi_link, next) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
if (SIGISMEMBER(*set, ksi->ksi_signo)) {
TAILQ_REMOVE(&sq->sq_list, ksi, ksi_link);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksi->ksi_sigq = NULL;
if (ksiginfo_tryfree(ksi) && p != NULL)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
p->p_pendingcnt--;
}
}
Add signal set sq_kill to sigqueue structure, the member saves all signals sent by kill() syscall, without this, a signal sent by sigqueue() can cause a signal sent by kill() to be lost.
2006-03-02 14:06:40 +00:00
SIGSETNAND(sq->sq_kill, *set);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
SIGSETNAND(sq->sq_ptrace, *set);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
SIGSETNAND(sq->sq_signals, *set);
}
void
sigqueue_delete(sigqueue_t *sq, int signo)
{
sigset_t set;
SIGEMPTYSET(set);
SIGADDSET(set, signo);
sigqueue_delete_set(sq, &set);
}
/* Remove a set of signals for a process */
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static void
sigqueue_collect_set() is no longer needed because other functions maintain pending set correctly.
2010-10-13 06:28:40 +00:00
sigqueue_delete_set_proc(struct proc *p, const sigset_t *set)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
sigqueue_t worklist;
struct thread *td0;
PROC_LOCK_ASSERT(p, MA_OWNED);
sigqueue_init(&worklist, NULL);
sigqueue_move_set(&p->p_sigqueue, &worklist, set);
FOREACH_THREAD_IN_PROC(p, td0)
sigqueue_move_set(&td0->td_sigqueue, &worklist, set);
sigqueue_flush(&worklist);
}
void
sigqueue_delete_proc(struct proc *p, int signo)
{
sigset_t set;
SIGEMPTYSET(set);
SIGADDSET(set, signo);
sigqueue_delete_set_proc(p, &set);
}
Staticise sigqueue manipulation functions used only in kern_sig.c. MFC after: 1 week
2010-01-23 11:43:30 +00:00
static void
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete_stopmask_proc(struct proc *p)
{
sigset_t set;
SIGEMPTYSET(set);
SIGADDSET(set, SIGSTOP);
SIGADDSET(set, SIGTSTP);
SIGADDSET(set, SIGTTIN);
SIGADDSET(set, SIGTTOU);
sigqueue_delete_set_proc(p, &set);
}
Uninlined CURSIG() and unpolluted <sys/signalvar.h>. CURSIG() had become very bloated, first with 128-bit sigset_t's, then with locking in the SMP case, then with locking in all cases. The space bloat was probably also time bloat, partly because the fast path through CURSIG() was pessimized by the sigset_t changes. This change speeds up lmbench's pipe-based latency benchmark by 4% on a Celeron. <sys/signalvar.h> had become very polluted to support the bloat.
2000-09-17 14:28:33 +00:00
/*
Tweak some comments.
2013-03-18 18:04:09 +00:00
* Determine signal that should be delivered to thread td, the current
* thread, 0 if none. If there is a pending stop signal with default
Uninlined CURSIG() and unpolluted <sys/signalvar.h>. CURSIG() had become very bloated, first with 128-bit sigset_t's, then with locking in the SMP case, then with locking in all cases. The space bloat was probably also time bloat, partly because the fast path through CURSIG() was pessimized by the sigset_t changes. This change speeds up lmbench's pipe-based latency benchmark by 4% on a Celeron. <sys/signalvar.h> had become very polluted to support the bloat.
2000-09-17 14:28:33 +00:00
* action, the process stops in issignal().
*/
int
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
cursig(struct thread *td)
Uninlined CURSIG() and unpolluted <sys/signalvar.h>. CURSIG() had become very bloated, first with 128-bit sigset_t's, then with locking in the SMP case, then with locking in all cases. The space bloat was probably also time bloat, partly because the fast path through CURSIG() was pessimized by the sigset_t changes. This change speeds up lmbench's pipe-based latency benchmark by 4% on a Celeron. <sys/signalvar.h> had become very polluted to support the bloat.
2000-09-17 14:28:33 +00:00
{
- p will be unused in cursig() if INVARIANTS is not defined. Access it through td->td_proc to avoid the unused variable. Spotted by: Maxim Konovalov <maxim@macomnet.ru>
2003-04-01 09:07:36 +00:00
PROC_LOCK_ASSERT(td->td_proc, MA_OWNED);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_assert(&td->td_proc->p_sigacts->ps_mtx, MA_OWNED);
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
THREAD_LOCK_ASSERT(td, MA_NOTOWNED);
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
return (SIGPENDING(td) ? issignal(td) : 0);
Uninlined CURSIG() and unpolluted <sys/signalvar.h>. CURSIG() had become very bloated, first with 128-bit sigset_t's, then with locking in the SMP case, then with locking in all cases. The space bloat was probably also time bloat, partly because the fast path through CURSIG() was pessimized by the sigset_t changes. This change speeds up lmbench's pipe-based latency benchmark by 4% on a Celeron. <sys/signalvar.h> had become very polluted to support the bloat.
2000-09-17 14:28:33 +00:00
}
Moved signal handling and rescheduling from userret() to ast() so that they aren't in the usual path of execution for syscalls and traps. The main complication for this is that we have to set flags to control ast() everywhere that changes the signal mask. Avoid locking in userret() in most of the remaining cases. Submitted by: luoqi (first part only, long ago, reorganized by me) Reminded by: dillon
2002-04-04 17:49:48 +00:00
/*
* Arrange for ast() to handle unmasked pending signals on return to user
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
* mode. This must be called whenever a signal is added to td_sigqueue or
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
* unmasked in td_sigmask.
Moved signal handling and rescheduling from userret() to ast() so that they aren't in the usual path of execution for syscalls and traps. The main complication for this is that we have to set flags to control ast() everywhere that changes the signal mask. Avoid locking in userret() in most of the remaining cases. Submitted by: luoqi (first part only, long ago, reorganized by me) Reminded by: dillon
2002-04-04 17:49:48 +00:00
*/
void
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
signotify(struct thread *td)
Moved signal handling and rescheduling from userret() to ast() so that they aren't in the usual path of execution for syscalls and traps. The main complication for this is that we have to set flags to control ast() everywhere that changes the signal mask. Avoid locking in userret() in most of the remaining cases. Submitted by: luoqi (first part only, long ago, reorganized by me) Reminded by: dillon
2002-04-04 17:49:48 +00:00
{
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
signotify: don't create a stack local that isn't used on non-debug builds
2018-05-19 03:57:41 +00:00
PROC_LOCK_ASSERT(td->td_proc, MA_OWNED);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
- Make sigonstack() a regular function instead of an inline and add a proc lock assertion to it. - SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify(). - Add a proc lock assertion to tdsigwakeup(). - Since we always set TDF_OLDMASK while holding the proc lock, the proc lock is sufficient protection to check its state in postsig() and we only need sched_lock when clearing the actual flag.
2003-04-18 20:59:05 +00:00
if (SIGPENDING(td)) {
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_lock(td);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
td->td_flags |= TDF_NEEDSIGCHK | TDF_ASTPENDING;
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_unlock(td);
- Make sigonstack() a regular function instead of an inline and add a proc lock assertion to it. - SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify(). - Add a proc lock assertion to tdsigwakeup(). - Since we always set TDF_OLDMASK while holding the proc lock, the proc lock is sufficient protection to check its state in postsig() and we only need sched_lock when clearing the actual flag.
2003-04-18 20:59:05 +00:00
}
}
Improve sigonstack(). Avoid relying on unsigned overflow for the test. Simplify expressions to avoid duplicate check for the range. Style. Add herald comment. Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D18361
2018-11-27 19:50:58 +00:00
/*
* Returns 1 (true) if altstack is configured for the thread, and the
* passed stack bottom address falls into the altstack range. Handles
* the 43 compat special case where the alt stack size is zero.
*/
- Make sigonstack() a regular function instead of an inline and add a proc lock assertion to it. - SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify(). - Add a proc lock assertion to tdsigwakeup(). - Since we always set TDF_OLDMASK while holding the proc lock, the proc lock is sufficient protection to check its state in postsig() and we only need sched_lock when clearing the actual flag.
2003-04-18 20:59:05 +00:00
int
sigonstack(size_t sp)
{
Improve sigonstack(). Avoid relying on unsigned overflow for the test. Simplify expressions to avoid duplicate check for the range. Style. Add herald comment. Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D18361
2018-11-27 19:50:58 +00:00
struct thread *td;
- Make sigonstack() a regular function instead of an inline and add a proc lock assertion to it. - SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify(). - Add a proc lock assertion to tdsigwakeup(). - Since we always set TDF_OLDMASK while holding the proc lock, the proc lock is sufficient protection to check its state in postsig() and we only need sched_lock when clearing the actual flag.
2003-04-18 20:59:05 +00:00
Improve sigonstack(). Avoid relying on unsigned overflow for the test. Simplify expressions to avoid duplicate check for the range. Style. Add herald comment. Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D18361
2018-11-27 19:50:58 +00:00
td = curthread;
if ((td->td_pflags & TDP_ALTSTACK) == 0)
return (0);
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#if defined(COMPAT_43)
Only enable COMPAT_43 changes for syscalls ABI for a.out processes. Reviewed by: imp, jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D21200
2019-08-11 19:16:07 +00:00
if (SV_PROC_FLAG(td->td_proc, SV_AOUT) && td->td_sigstk.ss_size == 0)
Improve sigonstack(). Avoid relying on unsigned overflow for the test. Simplify expressions to avoid duplicate check for the range. Style. Add herald comment. Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D18361
2018-11-27 19:50:58 +00:00
return ((td->td_sigstk.ss_flags & SS_ONSTACK) != 0);
- Make sigonstack() a regular function instead of an inline and add a proc lock assertion to it. - SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify(). - Add a proc lock assertion to tdsigwakeup(). - Since we always set TDF_OLDMASK while holding the proc lock, the proc lock is sufficient protection to check its state in postsig() and we only need sched_lock when clearing the actual flag.
2003-04-18 20:59:05 +00:00
#endif
Improve sigonstack(). Avoid relying on unsigned overflow for the test. Simplify expressions to avoid duplicate check for the range. Style. Add herald comment. Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D18361
2018-11-27 19:50:58 +00:00
return (sp >= (size_t)td->td_sigstk.ss_sp &&
sp < td->td_sigstk.ss_size + (size_t)td->td_sigstk.ss_sp);
Moved signal handling and rescheduling from userret() to ast() so that they aren't in the usual path of execution for syscalls and traps. The main complication for this is that we have to set flags to control ast() everywhere that changes the signal mask. Avoid locking in userret() in most of the remaining cases. Submitted by: luoqi (first part only, long ago, reorganized by me) Reminded by: dillon
2002-04-04 17:49:48 +00:00
}
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
static __inline int
sigprop(int sig)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
Modernize the initalization of sigproptbl. Use C99 designators to set the value of each slot and the nitems macro to check for valid entries. In the process, switch to indexing by signal number rather than signal-1 for improved clarity. Obtained from: CheriBSD (a6053c5abf03a5f53bbfcdd3a26429383f67e09f) Sponsored by: DARPA, AFRL Reviewed by: kib
2016-09-06 22:03:53 +00:00
if (sig > 0 && sig < nitems(sigproptbl))
return (sigproptbl[sig]);
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
return (0);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
int
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
sig_ffs(sigset_t *set)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
int i;
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
for (i = 0; i < _SIG_WORDS; i++)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (set->__bits[i])
return (ffs(set->__bits[i]) + (i * 32));
return (0);
}
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
static bool
Add missing const keyword to kern_sigaction()'s 'act' parameter. This structure is not modified by the function. Also add const to sigact_flag_test(), as it is called by kern_sigaction().
2015-07-10 14:39:46 +00:00
sigact_flag_test(const struct sigaction *act, int flag)
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
{
Revert the handling of all siginfo sa_flags except SA_SIGINFO to the pre-r270321. Namely, the flags are preserved for SIG_DFL and SIG_IGN dispositions. Requested and reviewed by: jilles Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-24 16:37:50 +00:00
/*
* SA_SIGINFO is reset when signal disposition is set to
* ignore or default. Other flags are kept according to user
* settings.
*/
return ((act->sa_flags & flag) != 0 && (flag != SA_SIGINFO ||
((__sighandler_t *)act->sa_sigaction != SIG_IGN &&
(__sighandler_t *)act->sa_sigaction != SIG_DFL)));
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
}
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
/*
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
* kern_sigaction
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
* sigaction
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
* freebsd4_sigaction
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
* osigaction
*/
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
int
Add missing const keyword to kern_sigaction()'s 'act' parameter. This structure is not modified by the function. Also add const to sigact_flag_test(), as it is called by kern_sigaction().
2015-07-10 14:39:46 +00:00
kern_sigaction(struct thread *td, int sig, const struct sigaction *act,
struct sigaction *oact, int flags)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
struct sigacts *ps;
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
struct proc *p = td->td_proc;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
We have a _SIG_VALID() macro, so use it instead of duplicating the test all over the place. Also replace a printf() + panic() with a KASSERT(). Reviewed by: jhb
2001-11-02 23:50:00 +00:00
if (!_SIG_VALID(sig))
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (EINVAL);
For sigaction(2), ignore possible garbage in sa_flags for sa_handler == SIG_DFL or SIG_IGN. Sloppy code does not fully initialize struct sigaction for such cases, and being too demanding in the case of default handler does not catch anything. Reported and tested by: Alex Tutubalin <lexa@lexa.ru> Sponsored by: The FreeBSD Foundation MFC after: 1 week
2015-01-16 07:06:58 +00:00
if (act != NULL && act->sa_handler != SIG_DFL &&
act->sa_handler != SIG_IGN && (act->sa_flags & ~(SA_ONSTACK |
SA_RESTART | SA_RESETHAND | SA_NOCLDSTOP | SA_NODEFER |
SA_NOCLDWAIT | SA_SIGINFO)) != 0)
Check the validity of struct sigaction sa_flags value, reject unknown flags. Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 07:52:47 +00:00
return (EINVAL);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
ps = p->p_sigacts;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_lock(&ps->ps_mtx);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (oact) {
Fix several leaks of kernel stack data through paddings. It is random collection of fixes for issues not yet corrected, reported at https://tsyrklevi.ch/clang_analyzer/freebsd_013017/. Many issues from that list were already corrected. Most of them are for compat32, old compat32 or affect both primary host ABI and compat32. The freebsd32_kldstat(), for instance, was already fixed by using malloc(M_ZERO). Patch includes correction to report the supplied version back, which is just pedantic. Reviewed by: brooks, emaste (previous version) Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D14868
2018-03-27 18:05:51 +00:00
memset(oact, 0, sizeof(*oact));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
oact->sa_mask = ps->ps_catchmask[_SIG_IDX(sig)];
if (SIGISMEMBER(ps->ps_sigonstack, sig))
oact->sa_flags |= SA_ONSTACK;
if (!SIGISMEMBER(ps->ps_sigintr, sig))
oact->sa_flags |= SA_RESTART;
if (SIGISMEMBER(ps->ps_sigreset, sig))
oact->sa_flags |= SA_RESETHAND;
if (SIGISMEMBER(ps->ps_signodefer, sig))
oact->sa_flags |= SA_NODEFER;
Be more exact with sigaction SA_SIGINFO handling. Reviewed by: marcel
2007-12-18 20:39:13 +00:00
if (SIGISMEMBER(ps->ps_siginfo, sig)) {
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
oact->sa_flags |= SA_SIGINFO;
Be more exact with sigaction SA_SIGINFO handling. Reviewed by: marcel
2007-12-18 20:39:13 +00:00
oact->sa_sigaction =
(__siginfohandler_t *)ps->ps_sigact[_SIG_IDX(sig)];
} else
oact->sa_handler = ps->ps_sigact[_SIG_IDX(sig)];
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
if (sig == SIGCHLD && ps->ps_flag & PS_NOCLDSTOP)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
oact->sa_flags |= SA_NOCLDSTOP;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
if (sig == SIGCHLD && ps->ps_flag & PS_NOCLDWAIT)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
oact->sa_flags |= SA_NOCLDWAIT;
}
if (act) {
if ((sig == SIGKILL || sig == SIGSTOP) &&
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
act->sa_handler != SIG_DFL) {
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (EINVAL);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
}
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
/*
* Change setting atomically.
*/
ps->ps_catchmask[_SIG_IDX(sig)] = act->sa_mask;
SIG_CANTMASK(ps->ps_catchmask[_SIG_IDX(sig)]);
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
if (sigact_flag_test(act, SA_SIGINFO)) {
Temporarily back out kern_sig.c rev 1.125 and kern_exit.c rev 1.131. This paniced my one of my machines one time too many :-( and there is no sign of a solution in the pipeline. The deltas are still easily available in cvs. The problem is that if the parent has been swapped out, the child process cannot grope around in the parent's UPAGES to see the sigact[] array or it will fault. This probably is a showstopper for this implementation anyway.
2001-08-01 20:35:24 +00:00
ps->ps_sigact[_SIG_IDX(sig)] =
(__sighandler_t *)act->sa_sigaction;
Fix a typo in do_sigaction() where sa_sigaction and sa_handler were confused. Since sa_sigaction and sa_handler alias each other in a union, the bug was completely harmless. This had been fixed as part of the SIGCHLD changes in revision 1.125, but it was reverted when they were backed out in revision 1.126.
2001-10-07 16:11:37 +00:00
SIGADDSET(ps->ps_siginfo, sig);
} else {
ps->ps_sigact[_SIG_IDX(sig)] = act->sa_handler;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
SIGDELSET(ps->ps_siginfo, sig);
}
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
if (!sigact_flag_test(act, SA_RESTART))
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
SIGADDSET(ps->ps_sigintr, sig);
else
SIGDELSET(ps->ps_sigintr, sig);
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
if (sigact_flag_test(act, SA_ONSTACK))
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
SIGADDSET(ps->ps_sigonstack, sig);
else
SIGDELSET(ps->ps_sigonstack, sig);
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
if (sigact_flag_test(act, SA_RESETHAND))
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
SIGADDSET(ps->ps_sigreset, sig);
else
SIGDELSET(ps->ps_sigreset, sig);
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
if (sigact_flag_test(act, SA_NODEFER))
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
SIGADDSET(ps->ps_signodefer, sig);
else
SIGDELSET(ps->ps_signodefer, sig);
if (sig == SIGCHLD) {
if (act->sa_flags & SA_NOCLDSTOP)
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag |= PS_NOCLDSTOP;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
else
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag &= ~PS_NOCLDSTOP;
Avoid the user-visible effect of setting SA_NOCLDWAIT when the SIGCHLD handler is SIG_IGN. This is a reimplementation of the problematic revision 1.131 of kern_exit.c. To avoid accessing process UPAGES, we set a new procsig flag when the SIGCHLD handler is SIG_IGN and use that instead.
2002-04-27 22:41:41 +00:00
if (act->sa_flags & SA_NOCLDWAIT) {
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
/*
* Paranoia: since SA_NOCLDWAIT is implemented
* by reparenting the dying child to PID 1 (and
* trust it to reap the zombie), PID 1 itself
* is forbidden to set SA_NOCLDWAIT.
*/
if (p->p_pid == 1)
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag &= ~PS_NOCLDWAIT;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
else
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag |= PS_NOCLDWAIT;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
} else
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag &= ~PS_NOCLDWAIT;
Avoid the user-visible effect of setting SA_NOCLDWAIT when the SIGCHLD handler is SIG_IGN. This is a reimplementation of the problematic revision 1.131 of kern_exit.c. To avoid accessing process UPAGES, we set a new procsig flag when the SIGCHLD handler is SIG_IGN and use that instead.
2002-04-27 22:41:41 +00:00
if (ps->ps_sigact[_SIG_IDX(SIGCHLD)] == SIG_IGN)
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag |= PS_CLDSIGIGN;
Avoid the user-visible effect of setting SA_NOCLDWAIT when the SIGCHLD handler is SIG_IGN. This is a reimplementation of the problematic revision 1.131 of kern_exit.c. To avoid accessing process UPAGES, we set a new procsig flag when the SIGCHLD handler is SIG_IGN and use that instead.
2002-04-27 22:41:41 +00:00
else
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag &= ~PS_CLDSIGIGN;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
/*
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
* Set bit in ps_sigignore for signals that are set to SIG_IGN,
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
* and for signals set to SIG_DFL where the default is to
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
* ignore. However, don't put SIGCONT in ps_sigignore, as we
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
* have to restart the process.
*/
if (ps->ps_sigact[_SIG_IDX(sig)] == SIG_IGN ||
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
(sigprop(sig) & SIGPROP_IGNORE &&
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
ps->ps_sigact[_SIG_IDX(sig)] == SIG_DFL)) {
/* never to be seen again */
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete_proc(p, sig);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (sig != SIGCONT)
/* easier in psignal */
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
SIGADDSET(ps->ps_sigignore, sig);
SIGDELSET(ps->ps_sigcatch, sig);
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
} else {
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
SIGDELSET(ps->ps_sigignore, sig);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (ps->ps_sigact[_SIG_IDX(sig)] == SIG_DFL)
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
SIGDELSET(ps->ps_sigcatch, sig);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
else
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
SIGADDSET(ps->ps_sigcatch, sig);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
#ifdef COMPAT_FREEBSD4
if (ps->ps_sigact[_SIG_IDX(sig)] == SIG_IGN ||
ps->ps_sigact[_SIG_IDX(sig)] == SIG_DFL ||
(flags & KSA_FREEBSD4) == 0)
SIGDELSET(ps->ps_freebsd4, sig);
else
SIGADDSET(ps->ps_freebsd4, sig);
#endif
Make COMPAT_43 optional again. XXX we need COMPAT_FBSD3 etc for this stuff.
2001-08-21 02:32:59 +00:00
#ifdef COMPAT_43
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
if (ps->ps_sigact[_SIG_IDX(sig)] == SIG_IGN ||
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
ps->ps_sigact[_SIG_IDX(sig)] == SIG_DFL ||
(flags & KSA_OSIGSET) == 0)
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
SIGDELSET(ps->ps_osigset, sig);
else
SIGADDSET(ps->ps_osigset, sig);
Make COMPAT_43 optional again. XXX we need COMPAT_FBSD3 etc for this stuff.
2001-08-21 02:32:59 +00:00
#endif
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (0);
}
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct sigaction_args {
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
int sig;
struct sigaction *act;
struct sigaction *oact;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sys_sigaction(struct thread *td, struct sigaction_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
struct sigaction act, oact;
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
struct sigaction *actp, *oactp;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
actp = (uap->act != NULL) ? &act : NULL;
oactp = (uap->oact != NULL) ? &oact : NULL;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (actp) {
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyin(uap->act, actp, sizeof(act));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (error)
Acquire Giant around calls to kern_sigaction() in sigaction(), freebsd4_sigaction() and osigaction() instead of around the whole body of those functions. They now no longer hold Giant around calls to copyin() and copyout(), and it is slightly more obvious what Giant is protecting.
2003-02-15 09:56:09 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
error = kern_sigaction(td, uap->sig, actp, oactp, 0);
Push Giant down into kern_sigaction() instead of locking it around calls to kern_sigaction() in the various callers of the function.
2003-04-25 20:01:19 +00:00
if (oactp && !error)
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyout(oactp, uap->oact, sizeof(oact));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
#ifdef COMPAT_FREEBSD4
#ifndef _SYS_SYSPROTO_H_
struct freebsd4_sigaction_args {
int sig;
struct sigaction *act;
struct sigaction *oact;
};
#endif
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
freebsd4_sigaction(struct thread *td, struct freebsd4_sigaction_args *uap)
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
{
struct sigaction act, oact;
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
struct sigaction *actp, *oactp;
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
int error;
actp = (uap->act != NULL) ? &act : NULL;
oactp = (uap->oact != NULL) ? &oact : NULL;
if (actp) {
error = copyin(uap->act, actp, sizeof(act));
if (error)
Acquire Giant around calls to kern_sigaction() in sigaction(), freebsd4_sigaction() and osigaction() instead of around the whole body of those functions. They now no longer hold Giant around calls to copyin() and copyout(), and it is slightly more obvious what Giant is protecting.
2003-02-15 09:56:09 +00:00
return (error);
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
}
error = kern_sigaction(td, uap->sig, actp, oactp, KSA_FREEBSD4);
Forgot to remove Giant around call to kern_sigaction() in freebsd4_sigaction() in revision 1.232.
2003-04-30 19:45:13 +00:00
if (oactp && !error)
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
error = copyout(oactp, uap->oact, sizeof(oact));
return (error);
}
#endif /* COMAPT_FREEBSD4 */
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#ifdef COMPAT_43 /* XXX - COMPAT_FBSD3 */
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
#ifndef _SYS_SYSPROTO_H_
struct osigaction_args {
int signum;
struct osigaction *nsa;
struct osigaction *osa;
};
#endif
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigaction(struct thread *td, struct osigaction_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
struct osigaction sa;
struct sigaction nsa, osa;
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
struct sigaction *nsap, *osap;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
if (uap->signum <= 0 || uap->signum >= ONSIG)
return (EINVAL);
Giant Pushdown clock_gettime() clock_settime() nanosleep() settimeofday() adjtime() getitimer() setitimer() __sysctl() ogetkerninfo() sigaction() osigaction() sigpending() osigpending() osigvec() osigblock() osigsetmask() sigsuspend() osigsuspend() osigstack() sigaltstack() kill() okillpg() trapsignal() nosys()
2001-09-01 18:19:21 +00:00
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
nsap = (uap->nsa != NULL) ? &nsa : NULL;
osap = (uap->osa != NULL) ? &osa : NULL;
Giant Pushdown clock_gettime() clock_settime() nanosleep() settimeofday() adjtime() getitimer() setitimer() __sysctl() ogetkerninfo() sigaction() osigaction() sigpending() osigpending() osigvec() osigblock() osigsetmask() sigsuspend() osigsuspend() osigstack() sigaltstack() kill() okillpg() trapsignal() nosys()
2001-09-01 18:19:21 +00:00
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (nsap) {
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyin(uap->nsa, &sa, sizeof(sa));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (error)
Acquire Giant around calls to kern_sigaction() in sigaction(), freebsd4_sigaction() and osigaction() instead of around the whole body of those functions. They now no longer hold Giant around calls to copyin() and copyout(), and it is slightly more obvious what Giant is protecting.
2003-02-15 09:56:09 +00:00
return (error);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
nsap->sa_handler = sa.sa_handler;
nsap->sa_flags = sa.sa_flags;
OSIG2SIG(sa.sa_mask, nsap->sa_mask);
Implement SA_SIGINFO for i386. Thanks to Bruce Evans for much more than a review, this was a nice puzzle. This is supposed to be binary and source compatible with older applications that access the old FreeBSD-style three arguments to a signal handler. Except those applications that access hidden signal handler arguments bejond the documented third one. If you have applications that do, please let me know so that we take the opportunity to provide the functionality they need in a documented manner. Also except application that use 'struct sigframe' directly. You need to recompile gdb and doscmd. `make world` is recommended. Example program that demonstrates how SA_SIGINFO and old-style FreeBSD handlers (with their three args) may be used in the same process is at http://www3.cons.org/tmp/fbsd-siginfo.c Programs that use the old FreeBSD-style three arguments are easy to change to SA_SIGINFO (although they don't need to, since the old style will still work): Old args to signal handler: void handler_sn(int sig, int code, struct sigcontext *scp) New args: void handler_si(int sig, siginfo_t *si, void *third) where: old:code == new:second->si_code old:scp == &(new:si->si_scp) /* Passed by value! */ The latter is also pointed to by new:third, but accessing via si->si_scp is preferred because it is type-save. FreeBSD implementation notes: - This is just the framework to make the interface POSIX compatible. For now, no additional functionality is provided. This is supposed to happen now, starting with floating point values. - We don't use 'sigcontext_t.si_value' for now (POSIX meant it for realtime-related values). - Documentation will be updated when new functionality is added and the exact arguments passed are determined. The comments in sys/signal.h are meant to be useful. Reviewed by: BDE
1999-07-06 07:13:48 +00:00
}
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
error = kern_sigaction(td, uap->signum, nsap, osap, KSA_OSIGSET);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (osap && !error) {
sa.sa_handler = osap->sa_handler;
sa.sa_flags = osap->sa_flags;
SIG2OSIG(osap->sa_mask, sa.sa_mask);
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyout(&sa, uap->osa, sizeof(sa));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
Remove various bits of conditional Alpha code and fixup a few comments.
2006-05-12 05:04:46 +00:00
#if !defined(__i386__)
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
/* Avoid replicating the same stub everywhere */
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigreturn(struct thread *td, struct osigreturn_args *uap)
Split 4.x and 5.x signal handling so that we can keep 4.x signal handling clean and functional as 5.x evolves. This allows some of the nasty bandaids in the 5.x codepaths to be unwound. Encapsulate 4.x signal handling under COMPAT_FREEBSD4 (there is an anti-foot-shooting measure in place, 5.x folks need this for a while) and finish encapsulating the older stuff under COMPAT_43. Since the ancient stuff is required on alpha (longjmp(3) passes a 'struct osigcontext *' to the current sigreturn(2), instead of the 'ucontext_t *' that sigreturn is supposed to take), add a compile time check to prevent foot shooting there too. Add uniform COMPAT_43 stubs for ia64/sparc64/powerpc. Tested on: i386, alpha, ia64. Compiled on sparc64 (a few days ago). Approved by: re
2002-10-25 19:10:58 +00:00
{
return (nosys(td, (struct nosys_args *)uap));
}
#endif
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Initialize signal state for process 0;
* set to ignore signals that are ignored by default.
*/
void
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
siginit(struct proc *p)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Remove register keyword from sys/ and ANSIfy prototypes A long long time ago the register keyword told the compiler to store the corresponding variable in a CPU register, but it is not relevant for any compiler used in the FreeBSD world today. ANSIfy related prototypes while here. Reviewed by: cem, jhb Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D10193
2017-05-17 00:34:34 +00:00
int i;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
struct sigacts *ps;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps = p->p_sigacts;
mtx_lock(&ps->ps_mtx);
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
for (i = 1; i <= NSIG; i++) {
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (sigprop(i) & SIGPROP_IGNORE && i != SIGCONT) {
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
SIGADDSET(ps->ps_sigignore, i);
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
}
}
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
/*
* Reset specified signal to the default disposition.
*/
static void
sigdflt(struct sigacts *ps, int sig)
{
mtx_assert(&ps->ps_mtx, MA_OWNED);
SIGDELSET(ps->ps_sigcatch, sig);
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if ((sigprop(sig) & SIGPROP_IGNORE) != 0 && sig != SIGCONT)
Ensure that sigaction flags for signal, which disposition is reset to ignored or default, are not leaking. Apparently, there exists code which relies on SA_SIGINFO not reported for SIG_DFL or SIG_IGN. In kern_sigaction, ignore flags when resetting. Encapsulate the flag and disposition testing into helper sigact_flag_test(). On exec, and when delivering signal with SA_RESETHAND flag set, signals are reset automatically. Use new helper sigdflt(), which removes duplicated code and corrects all flag bits for the signal. For proc0, set sigintr bit for all ignored signals. Ignored signals are consumed in tdsendsignal() and not delivered to the victim thread at all. Reported and tested by: royger Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-08-22 08:19:08 +00:00
SIGADDSET(ps->ps_sigignore, sig);
ps->ps_sigact[_SIG_IDX(sig)] = SIG_DFL;
SIGDELSET(ps->ps_siginfo, sig);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Reset signals for an exec of the specified process.
*/
void
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
execsigs(struct proc *p)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Unignore signals when starting CloudABI processes. As CloudABI processes cannot adjust their signal handlers, we need to make sure that we start up CloudABI processes with consistent signal masks. Though the POSIx standard signal behavior is all right, we do need to make sure that we ignore SIGPIPE, as it would otherwise be hard to interact with pipes and sockets. Extend execsigs() to iterate over ps_sigignore and call sigdflt() for each of the ignored signals. Reviewed by: kib Obtained from: https://github.com/NuxiNL/freebsd Differential Revision: https://reviews.freebsd.org/D3365
2015-08-12 11:30:31 +00:00
sigset_t osigignore;
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
struct sigacts *ps;
int sig;
struct thread *td;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Reset caught signals. Held signals remain held
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
* through td_sigmask (unless they were caught,
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* and are now ignored by default).
*/
- Reorder execve() so that it performs blocking operations before it locks the process. - Defer other blocking operations such as vrele()'s until after we release locks. - execsigs() now requires the proc lock to be held when it is called rather than locking the process internally.
2002-05-02 15:00:14 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
ps = p->p_sigacts;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_lock(&ps->ps_mtx);
rfork(2): add RFSPAWN flag When RFSPAWN is passed, rfork exhibits vfork(2) semantics but also resets signal handlers in the child during creation to avoid a point of corruption of parent state from the child. This flag will be used by posix_spawn(3) to handle potential signal issues. Reviewed by: jilles, kib Differential Revision: https://reviews.freebsd.org/D19058
2019-09-25 19:20:41 +00:00
sig_drop_caught(p);
Unignore signals when starting CloudABI processes. As CloudABI processes cannot adjust their signal handlers, we need to make sure that we start up CloudABI processes with consistent signal masks. Though the POSIx standard signal behavior is all right, we do need to make sure that we ignore SIGPIPE, as it would otherwise be hard to interact with pipes and sockets. Extend execsigs() to iterate over ps_sigignore and call sigdflt() for each of the ignored signals. Reviewed by: kib Obtained from: https://github.com/NuxiNL/freebsd Differential Revision: https://reviews.freebsd.org/D3365
2015-08-12 11:30:31 +00:00
/*
* As CloudABI processes cannot modify signal handlers, fully
* reset all signals to their default behavior. Do ignore
* SIGPIPE, as it would otherwise be impossible to recover from
* writes to broken pipes and sockets.
*/
if (SV_PROC_ABI(p) == SV_ABI_CLOUDABI) {
osigignore = ps->ps_sigignore;
while (SIGNOTEMPTY(osigignore)) {
sig = sig_ffs(&osigignore);
SIGDELSET(osigignore, sig);
if (sig != SIGPIPE)
sigdflt(ps, sig);
}
SIGADDSET(ps->ps_sigignore, SIGPIPE);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Reset stack state to the user stack.
* Clear set of signals caught on the signal stack.
*/
When clearing altsigstack settings on exec, do it to the right thread. Diagnosed by: smh Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-03-17 13:37:37 +00:00
td = curthread;
MPASS(td->td_proc == p);
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
td->td_sigstk.ss_flags = SS_DISABLE;
td->td_sigstk.ss_size = 0;
td->td_sigstk.ss_sp = 0;
td->td_pflags &= ~TDP_ALTSTACK;
Reset SA_NOCLDWAIT on exec(). PR: kern/12669 Submitted by: Doug Ambrisko <ambrisko@whistle.com>
1999-07-18 13:40:11 +00:00
/*
* Reset no zombies if child dies flag as Solaris does.
*/
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps->ps_flag &= ~(PS_NOCLDWAIT | PS_CLDSIGIGN);
Try to make the setting of the SIGCHLD handler the same as setting of the NOCLDWAI flag. Susv2 seems to require this. Submitted by: Cejka Rudolf <cejkar@dcse.fee.vutbr.cz> Reviewed by: dillon
2001-06-11 09:15:41 +00:00
if (ps->ps_sigact[_SIG_IDX(SIGCHLD)] == SIG_IGN)
ps->ps_sigact[_SIG_IDX(SIGCHLD)] = SIG_DFL;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
Rename do_sigprocmask() to kern_sigprocmask() and make it a global symbol so that it can be used by binary emulators.
2003-04-18 20:18:44 +00:00
* kern_sigprocmask()
Make the sigprocmask() and geteuid() system calls MP SAFE. Expand commentary for copyin/copyout to indicate that they are MP SAFE as well. Reviewed by: msmith
2000-04-02 17:52:43 +00:00
*
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
* Manipulate signal mask.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Rename do_sigprocmask() to kern_sigprocmask() and make it a global symbol so that it can be used by binary emulators.
2003-04-18 20:18:44 +00:00
int
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
kern_sigprocmask(struct thread *td, int how, sigset_t *set, sigset_t *oset,
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
int flags)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
sigset_t new_block, oset1;
struct proc *p;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
int error;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
p = td->td_proc;
Assert the state of the process lock and sigact mutex in kern_sigprocmask() and reschedule_signals(). Discussed with: rea Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-11-28 10:20:00 +00:00
if ((flags & SIGPROCMASK_PROC_LOCKED) != 0)
PROC_LOCK_ASSERT(p, MA_OWNED);
else
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
PROC_LOCK(p);
Assert the state of the process lock and sigact mutex in kern_sigprocmask() and reschedule_signals(). Discussed with: rea Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-11-28 10:20:00 +00:00
mtx_assert(&p->p_sigacts->ps_mtx, (flags & SIGPROCMASK_PS_LOCKED) != 0
? MA_OWNED : MA_NOTOWNED);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (oset != NULL)
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
*oset = td->td_sigmask;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
error = 0;
if (set != NULL) {
switch (how) {
case SIG_BLOCK:
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
SIG_CANTMASK(*set);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
oset1 = td->td_sigmask;
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
SIGSETOR(td->td_sigmask, *set);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
new_block = td->td_sigmask;
SIGSETNAND(new_block, oset1);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
break;
case SIG_UNBLOCK:
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
SIGSETNAND(td->td_sigmask, *set);
signotify(td);
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
goto out;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
case SIG_SETMASK:
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
SIG_CANTMASK(*set);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
oset1 = td->td_sigmask;
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
if (flags & SIGPROCMASK_OLD)
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
SIGSETLO(td->td_sigmask, *set);
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
else
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
td->td_sigmask = *set;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
new_block = td->td_sigmask;
SIGSETNAND(new_block, oset1);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
signotify(td);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
break;
default:
error = EINVAL;
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
goto out;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
/*
* The new_block set contains signals that were not previously
* blocked, but are blocked now.
*
* In case we block any signal that was not previously blocked
* for td, and process has the signal pending, try to schedule
* signal delivery to some thread that does not block the
* signal, possibly waking it up.
*/
if (p->p_numthreads != 1)
reschedule_signals(p, new_block, flags);
}
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
out:
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
if (!(flags & SIGPROCMASK_PROC_LOCKED))
PROC_UNLOCK(p);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (error);
}
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct sigprocmask_args {
int how;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
const sigset_t *set;
sigset_t *oset;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sys_sigprocmask(struct thread *td, struct sigprocmask_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
sigset_t set, oset;
sigset_t *setp, *osetp;
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
setp = (uap->set != NULL) ? &set : NULL;
osetp = (uap->oset != NULL) ? &oset : NULL;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (setp) {
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyin(uap->set, setp, sizeof(set));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (error)
return (error);
}
Rename do_sigprocmask() to kern_sigprocmask() and make it a global symbol so that it can be used by binary emulators.
2003-04-18 20:18:44 +00:00
error = kern_sigprocmask(td, uap->how, setp, osetp, 0);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (osetp && !error) {
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyout(osetp, uap->oset, sizeof(oset));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
return (error);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#ifdef COMPAT_43 /* XXX - COMPAT_FBSD3 */
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
#ifndef _SYS_SYSPROTO_H_
struct osigprocmask_args {
int how;
osigset_t mask;
};
#endif
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigprocmask(struct thread *td, struct osigprocmask_args *uap)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
sigset_t set, oset;
int error;
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
OSIG2SIG(uap->mask, set);
Rename do_sigprocmask() to kern_sigprocmask() and make it a global symbol so that it can be used by binary emulators.
2003-04-18 20:18:44 +00:00
error = kern_sigprocmask(td, uap->how, &set, &oset, 1);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
SIG2OSIG(oset, td->td_retval[0]);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_sigwait(struct thread *td, struct sigwait_args *uap)
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
{
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_t ksi;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
sigset_t set;
int error;
error = copyin(uap->set, &set, sizeof(set));
According to SUSv3, sigwait is different with sigwaitinfo, sigwait returns error code in return value, not in errno.
2004-06-07 13:35:02 +00:00
if (error) {
td->td_retval[0] = error;
return (0);
}
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
error = kern_sigtimedwait(td, set, &ksi, NULL);
According to SUSv3, sigwait is different with sigwaitinfo, sigwait returns error code in return value, not in errno.
2004-06-07 13:35:02 +00:00
if (error) {
The sigwait(3) function shall not return EINTR, according to the POSIX/SUSvN. The sigwait(2) syscall does return EINTR, and libc.so.7 contains the wrapper sigwait(3) which hides EINTR from callers. The EINTR return is used by libthr to handle required cancellation point in the sigwait(3). To help the binaries linked against pre-libc.so.7, i.e. RELENG_6 and earlier, to have right ABI for sigwait(3), transform EINTR return from sigwait(2) into ERESTART. Discussed with: davidxu MFC after: 1 week
2011-10-01 10:18:55 +00:00
if (error == EINTR && td->td_proc->p_osrel < P_OSREL_SIGWAIT)
error = ERESTART;
According to SUSv3, sigwait is different with sigwaitinfo, sigwait returns error code in return value, not in errno.
2004-06-07 13:35:02 +00:00
if (error == ERESTART)
return (error);
td->td_retval[0] = error;
return (0);
}
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
error = copyout(&ksi.ksi_signo, uap->sig, sizeof(ksi.ksi_signo));
According to SUSv3, sigwait is different with sigwaitinfo, sigwait returns error code in return value, not in errno.
2004-06-07 13:35:02 +00:00
td->td_retval[0] = error;
return (0);
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
}
Remove 'MPSAFE' annotations from the comments above most system calls: all system calls now enter without Giant held, and then in some cases, acquire Giant explicitly. Remove a number of other MPSAFE annotations in the credential code and tweak one or two other adjacent comments.
2007-03-04 22:36:48 +00:00
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_sigtimedwait(struct thread *td, struct sigtimedwait_args *uap)
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
{
struct timespec ts;
struct timespec *timeout;
sigset_t set;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_t ksi;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
int error;
if (uap->timeout) {
error = copyin(uap->timeout, &ts, sizeof(ts));
if (error)
return (error);
timeout = &ts;
} else
timeout = NULL;
error = copyin(uap->set, &set, sizeof(set));
if (error)
return (error);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
error = kern_sigtimedwait(td, set, &ksi, timeout);
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
if (error)
return (error);
o Change kse_thr_interrupt to allow send a signal to a specified thread, or unblock a thread in kernel, and allow UTS to specify whether syscall should be restarted. o Add ability for UTS to monitor signal comes in and removed from process, the flag PS_SIGEVENT is used to indicate the events. o Add a KMF_WAITSIGEVENT for KSE mailbox flag, UTS call kse_release with this flag set to wait for above signal event. o For SA based thread, kernel masks all signal in its signal mask, let UTS to use kse_thr_interrupt interrupt a thread, and install a signal frame in userland for the thread. o Add a tm_syncsig in thread mailbox, when a hardware trap occurs, it is used to deliver synchronous signal to userland, and upcall is schedule, so UTS can process the synchronous signal for the thread. Reviewed by: julian (mentor)
2003-06-28 08:29:05 +00:00
Fix POSIX compatible bug for sigwaitinfo and sigtimedwait. POSIX says siginfo pointer parameter can be NULL and if the function success, it should return signal number but not zero. The waitset it past should be negatived before it can be used as thread signal mask.
2003-06-28 08:03:28 +00:00
if (uap->info)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
error = copyout(&ksi.ksi_info, uap->info, sizeof(siginfo_t));
if (error == 0)
td->td_retval[0] = ksi.ksi_signo;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
return (error);
}
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_sigwaitinfo(struct thread *td, struct sigwaitinfo_args *uap)
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
{
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_t ksi;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
sigset_t set;
int error;
error = copyin(uap->set, &set, sizeof(set));
if (error)
return (error);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
error = kern_sigtimedwait(td, set, &ksi, NULL);
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
if (error)
return (error);
Fix POSIX compatible bug for sigwaitinfo and sigtimedwait. POSIX says siginfo pointer parameter can be NULL and if the function success, it should return signal number but not zero. The waitset it past should be negatived before it can be used as thread signal mask.
2003-06-28 08:03:28 +00:00
if (uap->info)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
error = copyout(&ksi.ksi_info, uap->info, sizeof(siginfo_t));
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
if (error == 0)
td->td_retval[0] = ksi.ksi_signo;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
return (error);
}
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
static void
proc_td_siginfo_capture(struct thread *td, siginfo_t *si)
{
struct thread *thr;
FOREACH_THREAD_IN_PROC(td->td_proc, thr) {
if (thr == td)
thr->td_si = *si;
else
thr->td_si.si_signo = 0;
}
}
Move some declaration of 32-bit signal structures into file freebsd32-signal.h, implement sigtimedwait and sigwaitinfo system calls.
2006-10-05 01:56:11 +00:00
int
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
kern_sigtimedwait(struct thread *td, sigset_t waitset, ksiginfo_t *ksi,
struct timespec *timeout)
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
{
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
struct sigacts *ps;
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
sigset_t saved_mask, new_block;
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
struct proc *p;
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
int error, sig, timo, timevalid = 0;
Don't restart a timeout wait in kern_sigtimedwait, also allow it to wait longer than a single integer can represent.
2005-02-19 06:05:49 +00:00
struct timespec rts, ets, ts;
struct timeval tv;
Requested and tested by: kevans Reviewed by: kevans (previous version), markj Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D22546
2019-11-27 20:33:53 +00:00
bool traced;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
p = td->td_proc;
error = 0;
Initialized ets to zero. This is arguably a gcc bug in that ets is always set to rts when timeout is non-NULL and then timevalid is set and ets is only checked later when timervalid is set.
2007-06-10 01:43:11 +00:00
ets.tv_sec = 0;
ets.tv_nsec = 0;
Requested and tested by: kevans Reviewed by: kevans (previous version), markj Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D22546
2019-11-27 20:33:53 +00:00
traced = false;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
if (timeout != NULL) {
Don't restart a timeout wait in kern_sigtimedwait, also allow it to wait longer than a single integer can represent.
2005-02-19 06:05:49 +00:00
if (timeout->tv_nsec >= 0 && timeout->tv_nsec < 1000000000) {
timevalid = 1;
getnanouptime(&rts);
Make timespecadd(3) and friends public The timespecadd(3) family of macros were imported from NetBSD back in r35029. However, they were initially guarded by #ifdef _KERNEL. In the meantime, we have grown at least 28 syscalls that use timespecs in some way, leading many programs both inside and outside of the base system to redefine those macros. It's better just to make the definitions public. Our kernel currently defines two-argument versions of timespecadd and timespecsub. NetBSD, OpenBSD, and FreeDesktop.org's libbsd, however, define three-argument versions. Solaris also defines a three-argument version, but only in its kernel. This revision changes our definition to match the common three-argument version. Bump _FreeBSD_version due to the breaking KPI change. Discussed with: cem, jilles, ian, bde Differential Revision: https://reviews.freebsd.org/D14725
2018-07-30 15:46:40 +00:00
timespecadd(&rts, timeout, &ets);
Don't restart a timeout wait in kern_sigtimedwait, also allow it to wait longer than a single integer can represent.
2005-02-19 06:05:49 +00:00
}
}
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
ksiginfo_init(ksi);
/* Some signals can not be waited for. */
SIG_CANTMASK(waitset);
ps = p->p_sigacts;
PROC_LOCK(p);
saved_mask = td->td_sigmask;
SIGSETNAND(td->td_sigmask, waitset);
for (;;) {
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
mtx_lock(&ps->ps_mtx);
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
sig = cursig(td);
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
mtx_unlock(&ps->ps_mtx);
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
KASSERT(sig >= 0, ("sig %d", sig));
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
if (sig != 0 && SIGISMEMBER(waitset, sig)) {
if (sigqueue_get(&td->td_sigqueue, sig, ksi) != 0 ||
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
sigqueue_get(&p->p_sigqueue, sig, ksi) != 0) {
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
error = 0;
break;
}
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
}
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
if (error != 0)
break;
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
/*
* POSIX says this must be checked after looking for pending
* signals.
*/
if (timeout != NULL) {
if (!timevalid) {
error = EINVAL;
break;
}
getnanouptime(&rts);
if (timespeccmp(&rts, &ets, >=)) {
error = EAGAIN;
break;
}
Make timespecadd(3) and friends public The timespecadd(3) family of macros were imported from NetBSD back in r35029. However, they were initially guarded by #ifdef _KERNEL. In the meantime, we have grown at least 28 syscalls that use timespecs in some way, leading many programs both inside and outside of the base system to redefine those macros. It's better just to make the definitions public. Our kernel currently defines two-argument versions of timespecadd and timespecsub. NetBSD, OpenBSD, and FreeDesktop.org's libbsd, however, define three-argument versions. Solaris also defines a three-argument version, but only in its kernel. This revision changes our definition to match the common three-argument version. Bump _FreeBSD_version due to the breaking KPI change. Discussed with: cem, jilles, ian, bde Differential Revision: https://reviews.freebsd.org/D14725
2018-07-30 15:46:40 +00:00
timespecsub(&ets, &rts, &ts);
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
TIMESPEC_TO_TIMEVAL(&tv, &ts);
timo = tvtohz(&tv);
} else {
timo = 0;
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
}
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
Requested and tested by: kevans Reviewed by: kevans (previous version), markj Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D22546
2019-11-27 20:33:53 +00:00
if (traced) {
error = EINTR;
break;
}
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
error = msleep(ps, &p->p_mtx, PPAUSE|PCATCH, "sigwait", timo);
if (timeout != NULL) {
if (error == ERESTART) {
/* Timeout can not be restarted. */
error = EINTR;
} else if (error == EAGAIN) {
/* We will calculate timeout by ourself. */
error = 0;
}
Don't restart a timeout wait in kern_sigtimedwait, also allow it to wait longer than a single integer can represent.
2005-02-19 06:05:49 +00:00
}
Requested and tested by: kevans Reviewed by: kevans (previous version), markj Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D22546
2019-11-27 20:33:53 +00:00
/*
* If PTRACE_SCE or PTRACE_SCX were set after
* userspace entered the syscall, return spurious
* EINTR after wait was done. Only do this as last
* resort after rechecking for possible queued signals
* and expired timeouts.
*/
if (error == 0 && (p->p_ptevents & PTRACE_SYSCALL) != 0)
traced = true;
Don't restart a timeout wait in kern_sigtimedwait, also allow it to wait longer than a single integer can represent.
2005-02-19 06:05:49 +00:00
}
o Change kse_thr_interrupt to allow send a signal to a specified thread, or unblock a thread in kernel, and allow UTS to specify whether syscall should be restarted. o Add ability for UTS to monitor signal comes in and removed from process, the flag PS_SIGEVENT is used to indicate the events. o Add a KMF_WAITSIGEVENT for KSE mailbox flag, UTS call kse_release with this flag set to wait for above signal event. o For SA based thread, kernel masks all signal in its signal mask, let UTS to use kse_thr_interrupt interrupt a thread, and install a signal frame in userland for the thread. o Add a tm_syncsig in thread mailbox, when a hardware trap occurs, it is used to deliver synchronous signal to userland, and upcall is schedule, so UTS can process the synchronous signal for the thread. Reviewed by: julian (mentor)
2003-06-28 08:29:05 +00:00
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
new_block = saved_mask;
SIGSETNAND(new_block, td->td_sigmask);
td->td_sigmask = saved_mask;
/*
* Fewer signals can be delivered to us, reschedule signal
* notification.
*/
if (p->p_numthreads != 1)
reschedule_signals(p, new_block, 0);
Add DTrace 'proc' provider probes using the Statically Defined Trace (sdt) mechanism.
2008-05-24 06:22:16 +00:00
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
if (error == 0) {
Fix style issues around existing SDT probes. - Use SDT_PROBE<N>() instead of SDT_PROBE(). This has no functional effect at the moment, but will be needed for some future changes. - Don't hardcode the module component of the probe identifier. This is set automatically by the SDT framework. MFC after: 1 week
2015-12-16 23:39:27 +00:00
SDT_PROBE2(proc, , , signal__clear, sig, ksi);
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
Let itimer store itimerspec instead of itimerval, so I don't have to convert to or from timeval frequently. Introduce function itimer_accept() to ack a timer signal in signal acceptance code, this allows us to return more fresh overrun counter than at signal generating time. while POSIX says: "the value returned by timer_getoverrun() shall apply to the most recent expiration signal delivery or acceptance for the timer,.." I prefer returning it at acceptance time. Introduce SIGEV_THREAD_ID notification mode, it is used by thread libary to request kernel to deliver signal to a specified thread, and in turn, the thread library may use the mechanism to implement SIGEV_THREAD which is required by POSIX. Timer signal is managed by timer code, so it can not fail even if signal queue is full filled by sigqueue syscall.
2005-10-30 02:56:08 +00:00
if (ksi->ksi_code == SI_TIMER)
itimer_accept(p, ksi->ksi_timerid, ksi);
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
#ifdef KTRACE
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
if (KTRPOINT(td, KTR_PSIG)) {
sig_t action;
mtx_lock(&ps->ps_mtx);
action = ps->ps_sigact[_SIG_IDX(sig)];
mtx_unlock(&ps->ps_mtx);
ktrace: Log the code for all signals (PSIG events). The code provides information on how the signal was generated. Formerly, the code was only logged for traps, much like only signal handlers for traps received a meaningful si_code before FreeBSD 7.0. In rare cases, no information is available and 0 is still logged. MFC after: 1 week
2011-04-17 14:38:11 +00:00
ktrpsig(sig, action, &td->td_sigmask, ksi->ksi_code);
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
}
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
#endif
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
if (sig == SIGKILL) {
proc_td_siginfo_capture(td, &ksi->ksi_info);
1. Refine kern_sigtimedwait() to remove redundant code. 2. Fix a bug, if thread got a SIGKILL signal, call sigexit() to kill its process. MFC after: 3 days
2006-02-23 09:24:19 +00:00
sigexit(td, sig);
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
}
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
}
- Define sigwait, sigtimedwait, and sigwaitinfo in terms of kern_sigtimedwait() which is capable of supporting all of their semantics. - These should be POSIX compliant but more careful review is needed before we announce this.
2003-03-31 23:30:41 +00:00
PROC_UNLOCK(p);
return (error);
}
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
#ifndef _SYS_SYSPROTO_H_
struct sigpending_args {
sigset_t *set;
};
#endif
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sys_sigpending(struct thread *td, struct sigpending_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct proc *p = td->td_proc;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigset_t pending;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
pending = p->p_sigqueue.sq_signals;
SIGSETOR(pending, td->td_sigqueue.sq_signals);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (copyout(&pending, uap->set, sizeof(sigset_t)));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#ifdef COMPAT_43 /* XXX - COMPAT_FBSD3 */
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
#ifndef _SYS_SYSPROTO_H_
struct osigpending_args {
int dummy;
};
#endif
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigpending(struct thread *td, struct osigpending_args *uap)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct proc *p = td->td_proc;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigset_t pending;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
pending = p->p_sigqueue.sq_signals;
SIGSETOR(pending, td->td_sigqueue.sq_signals);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
SIG2OSIG(pending, td->td_retval[0]);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (0);
}
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#if defined(COMPAT_43)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Generalized interface signal handler, 4.3-compatible.
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct osigvec_args {
int signum;
struct sigvec *nsv;
struct sigvec *osv;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigvec(struct thread *td, struct osigvec_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
struct sigvec vec;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
struct sigaction nsa, osa;
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
struct sigaction *nsap, *osap;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
if (uap->signum <= 0 || uap->signum >= ONSIG)
return (EINVAL);
nsap = (uap->nsv != NULL) ? &nsa : NULL;
osap = (uap->osv != NULL) ? &osa : NULL;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (nsap) {
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyin(uap->nsv, &vec, sizeof(vec));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (error)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
nsap->sa_handler = vec.sv_handler;
OSIG2SIG(vec.sv_mask, nsap->sa_mask);
nsap->sa_flags = vec.sv_flags;
nsap->sa_flags ^= SA_RESTART; /* opposite of SV_INTERRUPT */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Remove garbage SIGSETOR() that snuck into struct sigpending_args definition. - Use the proper constant for the last arg to kern_sigaction() in osigvec() instead of a magic value.
2003-04-17 22:06:43 +00:00
error = kern_sigaction(td, uap->signum, nsap, osap, KSA_OSIGSET);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (osap && !error) {
vec.sv_handler = osap->sa_handler;
SIG2OSIG(osap->sa_mask, vec.sv_mask);
vec.sv_flags = osap->sa_flags;
vec.sv_flags &= ~SA_NOCLDWAIT;
vec.sv_flags ^= SA_RESTART;
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyout(&vec, uap->osv, sizeof(vec));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct osigblock_args {
int mask;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigblock(struct thread *td, struct osigblock_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
In r197963, a race with thread being selected for signal delivery while in kernel mode, and later changing signal mask to block the signal, was fixed for sigprocmask(2) and ptread_exit(3). The same race exists for sigreturn(2), setcontext(2) and swapcontext(2) syscalls. Use kern_sigprocmask() instead of direct manipulation of td_sigmask to reschedule newly blocked signals, closing the race. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:47:58 +00:00
sigset_t set, oset;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
OSIG2SIG(uap->mask, set);
In r197963, a race with thread being selected for signal delivery while in kernel mode, and later changing signal mask to block the signal, was fixed for sigprocmask(2) and ptread_exit(3). The same race exists for sigreturn(2), setcontext(2) and swapcontext(2) syscalls. Use kern_sigprocmask() instead of direct manipulation of td_sigmask to reschedule newly blocked signals, closing the race. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:47:58 +00:00
kern_sigprocmask(td, SIG_BLOCK, &set, &oset, 0);
SIG2OSIG(oset, td->td_retval[0]);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (0);
}
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct osigsetmask_args {
int mask;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigsetmask(struct thread *td, struct osigsetmask_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
In r197963, a race with thread being selected for signal delivery while in kernel mode, and later changing signal mask to block the signal, was fixed for sigprocmask(2) and ptread_exit(3). The same race exists for sigreturn(2), setcontext(2) and swapcontext(2) syscalls. Use kern_sigprocmask() instead of direct manipulation of td_sigmask to reschedule newly blocked signals, closing the race. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:47:58 +00:00
sigset_t set, oset;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
OSIG2SIG(uap->mask, set);
In r197963, a race with thread being selected for signal delivery while in kernel mode, and later changing signal mask to block the signal, was fixed for sigprocmask(2) and ptread_exit(3). The same race exists for sigreturn(2), setcontext(2) and swapcontext(2) syscalls. Use kern_sigprocmask() instead of direct manipulation of td_sigmask to reschedule newly blocked signals, closing the race. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:47:58 +00:00
kern_sigprocmask(td, SIG_SETMASK, &set, &oset, 0);
SIG2OSIG(oset, td->td_retval[0]);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (0);
}
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Further system call comment cleanup: - Remove also "MP SAFE" after prior "MPSAFE" pass. (suggested by bde) - Remove extra blank lines in some cases. - Add extra blank lines in some cases. - Remove no-op comments consisting solely of the function name, the word "syscall", or the system call name. - Add punctuation. - Re-wrap some comments.
2007-03-05 13:10:58 +00:00
* Suspend calling thread until signal, providing mask to be set in the
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
* meantime.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct sigsuspend_args {
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
const sigset_t *sigmask;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sys_sigsuspend(struct thread *td, struct sigsuspend_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
sigset_t mask;
int error;
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
error = copyin(uap->sigmask, &mask, sizeof(mask));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
if (error)
return (error);
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
return (kern_sigsuspend(td, mask));
}
int
kern_sigsuspend(struct thread *td, sigset_t mask)
{
struct proc *p = td->td_proc;
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
int has_sig, sig;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
* When returning from sigsuspend, we want
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* the old mask to be restored after the
* signal handler has finished. Thus, we
* save it here and mark the sigacts structure
* to indicate this.
*/
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
kern_sigprocmask(td, SIG_SETMASK, &mask, &td->td_oldsigmask,
SIGPROCMASK_PROC_LOCKED);
- Add a td_pflags field to struct thread for private flags accessed only by curthread. Unlike td_flags, this field does not need any locking. - Replace the td_inktr and td_inktrace variables with equivalent private thread flags. - Move TDF_OLDMASK over to the private flags field so it no longer requires sched_lock.
2003-06-09 17:38:32 +00:00
td->td_pflags |= TDP_OLDMASK;
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
/*
* Process signals now. Otherwise, we can get spurious wakeup
* due to signal entered process queue, but delivered to other
* thread. But sigsuspend should return only on signal
* delivery.
*/
Reorganize syscall entry and leave handling. Extend struct sysvec with three new elements: sv_fetch_syscall_args - the method to fetch syscall arguments from usermode into struct syscall_args. The structure is machine-depended (this might be reconsidered after all architectures are converted). sv_set_syscall_retval - the method to set a return value for usermode from the syscall. It is a generalization of cpu_set_syscall_retval(9) to allow ABIs to override the way to set a return value. sv_syscallnames - the table of syscall names. Use sv_set_syscall_retval in kern_sigsuspend() instead of hardcoding the call to cpu_set_syscall_retval(). The new functions syscallenter(9) and syscallret(9) are provided that use sv_*syscall* pointers and contain the common repeated code from the syscall() implementations for the architecture-specific syscall trap handlers. Syscallenter() fetches arguments, calls syscall implementation from ABI sysent table, and set up return frame. The end of syscall bookkeeping is done by syscallret(). Take advantage of single place for MI syscall handling code and implement ptrace_lwpinfo pl_flags PL_FLAG_SCE, PL_FLAG_SCX and PL_FLAG_EXEC. The SCE and SCX flags notify the debugger that the thread is stopped at syscall entry or return point respectively. The EXEC flag augments SCX and notifies debugger that the process address space was changed by one of exec(2)-family syscalls. The i386, amd64, sparc64, sun4v, powerpc and ia64 syscall()s are changed to use syscallenter()/syscallret(). MIPS and arm are not converted and use the mostly unchanged syscall() implementation. Reviewed by: jhb, marcel, marius, nwhitehorn, stas Tested by: marcel (ia64), marius (sparc64), nwhitehorn (powerpc), stas (mips) MFC after: 1 month
2010-05-23 18:32:02 +00:00
(p->p_sysent->sv_set_syscall_retval)(td, EINTR);
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
for (has_sig = 0; !has_sig;) {
while (msleep(&p->p_sigacts, &p->p_mtx, PPAUSE|PCATCH, "pause",
0) == 0)
/* void */;
thread_suspend_check(0);
mtx_lock(&p->p_sigacts->ps_mtx);
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
while ((sig = cursig(td)) != 0) {
KASSERT(sig >= 0, ("sig %d", sig));
In r198506, kern_sigsuspend() started doing cursig/postsig loop to make sure that a signal was delivered to the thread before returning from syscall. Signal delivery puts new return frame on the user stack, and modifies trap frame to enter signal handler. As a consequence, syscall return code sets EINTR as error return for signal frame, instead of the syscall return. Also, for ia64, due to different registers layout for those two kind of frames, usermode sigsegfaulted when returned from signal handler. Use newly-introduced cpu_set_syscall_retval(9) to set syscall result, and return EJUSTRETURN from kern_sigsuspend() to prevent syscall return code from modifying this frame [1]. Another issue is that pending SIGCONT might be cancelled by SIGSTOP, causing postsig() not to deliver any catched signal [2]. Modify postsig() to return 1 if signal was posted, and 0 otherwise, and use this in the kern_sigsuspend loop. Proposed by: marcel [1] Noted by: davidxu [2] Reviewed by: marcel, davidxu MFC after: 1 month
2009-11-10 11:46:53 +00:00
has_sig += postsig(sig);
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
}
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
mtx_unlock(&p->p_sigacts->ps_mtx);
Requested and tested by: kevans Reviewed by: kevans (previous version), markj Sponsored by: The FreeBSD Foundation MFC after: 1 week Differential revision: https://reviews.freebsd.org/D22546
2019-11-27 20:33:53 +00:00
/*
* If PTRACE_SCE or PTRACE_SCX were set after
* userspace entered the syscall, return spurious
* EINTR.
*/
if ((p->p_ptevents & PTRACE_SYSCALL) != 0)
has_sig += 1;
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
}
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
Add thread-private flag to indicate that error value is already placed in td_errno. Flag is supposed to be used by syscalls returning EJUSTRETURN because errno was already placed into the usermode frame by a call to set_syscall_retval(9). Both ktrace and dtrace get errno value from td_errno if the flag is set. Use the flag to fix sigsuspend(2) error return ktrace records. Requested by: bde MFC after: 1 week
2012-04-12 10:48:43 +00:00
td->td_errno = EINTR;
td->td_pflags |= TDP_NERRNO;
In r198506, kern_sigsuspend() started doing cursig/postsig loop to make sure that a signal was delivered to the thread before returning from syscall. Signal delivery puts new return frame on the user stack, and modifies trap frame to enter signal handler. As a consequence, syscall return code sets EINTR as error return for signal frame, instead of the syscall return. Also, for ia64, due to different registers layout for those two kind of frames, usermode sigsegfaulted when returned from signal handler. Use newly-introduced cpu_set_syscall_retval(9) to set syscall result, and return EJUSTRETURN from kern_sigsuspend() to prevent syscall return code from modifying this frame [1]. Another issue is that pending SIGCONT might be cancelled by SIGSTOP, causing postsig() not to deliver any catched signal [2]. Modify postsig() to return 1 if signal was posted, and 0 otherwise, and use this in the kern_sigsuspend loop. Proposed by: marcel [1] Noted by: davidxu [2] Reviewed by: marcel, davidxu MFC after: 1 month
2009-11-10 11:46:53 +00:00
return (EJUSTRETURN);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
}
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#ifdef COMPAT_43 /* XXX - COMPAT_FBSD3 */
Move the part of the comment which applies to osigsuspend where it belongs. The current sigsuspend syscall does expect a pointer to the mask as argument. Submitted by: Igor Sysoev <is at rambler-co dot ru>
2004-01-28 06:06:04 +00:00
/*
* Compatibility sigsuspend call for old binaries. Note nonstandard calling
* convention: libc stub passes mask, not pointer, to save a copyin.
*/
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
#ifndef _SYS_SYSPROTO_H_
struct osigsuspend_args {
osigset_t mask;
};
#endif
/* ARGSUSED */
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigsuspend(struct thread *td, struct osigsuspend_args *uap)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
{
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
sigset_t mask;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
OSIG2SIG(uap->mask, mask);
In kern_sigsuspend(), better manipulate thread signal mask using kern_sigprocmask() to properly notify other possible candidate threads for signal delivery. Since sigsuspend() shall only return to usermode after a signal was delivered, do cursig/postsig loop immediately after waiting for signal, repeating the wait if wakeup was spurious due to race with other thread fetching signal from the process queue before us. Add thread_suspend_check() call to allow the thread to be stopped or killed while in loop. Modify last argument of kern_sigprocmask() from boolean to flags, allowing the function to be called with locked proc. Convertion of the callers that supplied 1 to the old argument will be done in the next commit, and due to SIGPROCMASK_OLD value equial to 1, code is formally correct in between. Reviewed by: davidxu Tested by: pho MFC after: 1 month
2009-10-27 10:42:24 +00:00
return (kern_sigsuspend(td, mask));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Make this file compile again when COMPAT_43 has not been defined. This boils down to conditionally compile the old signal syscalls. We might want to extend the types in syscalls.master to make these syscalls conditionally on something more appropriate than COMPAT_43.
2000-08-26 02:27:01 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#if defined(COMPAT_43)
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct osigstack_args {
struct sigstack *nss;
struct sigstack *oss;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
osigstack(struct thread *td, struct osigstack_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
- Reorganize osigstack() to do the copyin first, grab the proc lock once, do all the various sigstack dances, unlock the proc lock, and finally do the copyout. This more closely resembles the behavior of kern_sigaltstack() and closes a small race. - Remove Giant from osigstack as it is no longer needed.
2003-04-23 18:50:25 +00:00
struct sigstack nss, oss;
Giant Pushdown clock_gettime() clock_settime() nanosleep() settimeofday() adjtime() getitimer() setitimer() __sysctl() ogetkerninfo() sigaction() osigaction() sigpending() osigpending() osigvec() osigblock() osigsetmask() sigsuspend() osigsuspend() osigstack() sigaltstack() kill() okillpg() trapsignal() nosys()
2001-09-01 18:19:21 +00:00
int error = 0;
- Reorganize osigstack() to do the copyin first, grab the proc lock once, do all the various sigstack dances, unlock the proc lock, and finally do the copyout. This more closely resembles the behavior of kern_sigaltstack() and closes a small race. - Remove Giant from osigstack as it is no longer needed.
2003-04-23 18:50:25 +00:00
if (uap->nss != NULL) {
error = copyin(uap->nss, &nss, sizeof(nss));
Don't use p->p_sigstk.ss_flags to keep state of whether the process is on the alternate stack or not. For compatibility with sigstack(2) state is being updated if such is needed. We now determine whether the process is on the alternate stack by looking at its stack pointer. This allows a process to siglongjmp from a signal handler on the alternate stack to the place of the sigsetjmp on the normal stack. When maintaining state, this would have invalidated the state information and causing a subsequent signal to be delivered on the normal stack instead of the alternate stack. PR: 22286
2000-11-30 05:23:49 +00:00
if (error)
- Reorganize osigstack() to do the copyin first, grab the proc lock once, do all the various sigstack dances, unlock the proc lock, and finally do the copyout. This more closely resembles the behavior of kern_sigaltstack() and closes a small race. - Remove Giant from osigstack as it is no longer needed.
2003-04-23 18:50:25 +00:00
return (error);
Don't use p->p_sigstk.ss_flags to keep state of whether the process is on the alternate stack or not. For compatibility with sigstack(2) state is being updated if such is needed. We now determine whether the process is on the alternate stack by looking at its stack pointer. This allows a process to siglongjmp from a signal handler on the alternate stack to the place of the sigsetjmp on the normal stack. When maintaining state, this would have invalidated the state information and causing a subsequent signal to be delivered on the normal stack instead of the alternate stack. PR: 22286
2000-11-30 05:23:49 +00:00
}
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
oss.ss_sp = td->td_sigstk.ss_sp;
- Reorganize osigstack() to do the copyin first, grab the proc lock once, do all the various sigstack dances, unlock the proc lock, and finally do the copyout. This more closely resembles the behavior of kern_sigaltstack() and closes a small race. - Remove Giant from osigstack as it is no longer needed.
2003-04-23 18:50:25 +00:00
oss.ss_onstack = sigonstack(cpu_getstack(td));
Don't use p->p_sigstk.ss_flags to keep state of whether the process is on the alternate stack or not. For compatibility with sigstack(2) state is being updated if such is needed. We now determine whether the process is on the alternate stack by looking at its stack pointer. This allows a process to siglongjmp from a signal handler on the alternate stack to the place of the sigsetjmp on the normal stack. When maintaining state, this would have invalidated the state information and causing a subsequent signal to be delivered on the normal stack instead of the alternate stack. PR: 22286
2000-11-30 05:23:49 +00:00
if (uap->nss != NULL) {
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
td->td_sigstk.ss_sp = nss.ss_sp;
td->td_sigstk.ss_size = 0;
td->td_sigstk.ss_flags |= nss.ss_onstack & SS_ONSTACK;
td->td_pflags |= TDP_ALTSTACK;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Reorganize osigstack() to do the copyin first, grab the proc lock once, do all the various sigstack dances, unlock the proc lock, and finally do the copyout. This more closely resembles the behavior of kern_sigaltstack() and closes a small race. - Remove Giant from osigstack as it is no longer needed.
2003-04-23 18:50:25 +00:00
if (uap->oss != NULL)
error = copyout(&oss, uap->oss, sizeof(oss));
Giant Pushdown clock_gettime() clock_settime() nanosleep() settimeofday() adjtime() getitimer() setitimer() __sysctl() ogetkerninfo() sigaction() osigaction() sigpending() osigpending() osigvec() osigblock() osigsetmask() sigsuspend() osigsuspend() osigstack() sigaltstack() kill() okillpg() trapsignal() nosys()
2001-09-01 18:19:21 +00:00
return (error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct sigaltstack_args {
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
stack_t *ss;
stack_t *oss;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sys_sigaltstack(struct thread *td, struct sigaltstack_args *uap)
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
{
stack_t ss, oss;
int error;
if (uap->ss != NULL) {
error = copyin(uap->ss, &ss, sizeof(ss));
if (error)
return (error);
}
error = kern_sigaltstack(td, (uap->ss != NULL) ? &ss : NULL,
(uap->oss != NULL) ? &oss : NULL);
if (error)
return (error);
if (uap->oss != NULL)
error = copyout(&oss, uap->oss, sizeof(stack_t));
return (error);
}
int
kern_sigaltstack(struct thread *td, stack_t *ss, stack_t *oss)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct proc *p = td->td_proc;
Giant Pushdown clock_gettime() clock_settime() nanosleep() settimeofday() adjtime() getitimer() setitimer() __sysctl() ogetkerninfo() sigaction() osigaction() sigpending() osigpending() osigvec() osigblock() osigsetmask() sigsuspend() osigsuspend() osigstack() sigaltstack() kill() okillpg() trapsignal() nosys()
2001-09-01 18:19:21 +00:00
int oonstack;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
oonstack = sigonstack(cpu_getstack(td));
Don't use p->p_sigstk.ss_flags to keep state of whether the process is on the alternate stack or not. For compatibility with sigstack(2) state is being updated if such is needed. We now determine whether the process is on the alternate stack by looking at its stack pointer. This allows a process to siglongjmp from a signal handler on the alternate stack to the place of the sigsetjmp on the normal stack. When maintaining state, this would have invalidated the state information and causing a subsequent signal to be delivered on the normal stack instead of the alternate stack. PR: 22286
2000-11-30 05:23:49 +00:00
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (oss != NULL) {
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
*oss = td->td_sigstk;
oss->ss_flags = (td->td_pflags & TDP_ALTSTACK)
Don't use p->p_sigstk.ss_flags to keep state of whether the process is on the alternate stack or not. For compatibility with sigstack(2) state is being updated if such is needed. We now determine whether the process is on the alternate stack by looking at its stack pointer. This allows a process to siglongjmp from a signal handler on the alternate stack to the place of the sigsetjmp on the normal stack. When maintaining state, this would have invalidated the state information and causing a subsequent signal to be delivered on the normal stack instead of the alternate stack. PR: 22286
2000-11-30 05:23:49 +00:00
? ((oonstack) ? SS_ONSTACK : 0) : SS_DISABLE;
}
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (ss != NULL) {
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
if (oonstack)
Remove Giant from osigblock(), osigsetmask(), and kern_sigaltstack().
2003-04-23 19:49:18 +00:00
return (EPERM);
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
if ((ss->ss_flags & ~SS_DISABLE) != 0)
Remove Giant from osigblock(), osigsetmask(), and kern_sigaltstack().
2003-04-23 19:49:18 +00:00
return (EINVAL);
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (!(ss->ss_flags & SS_DISABLE)) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
if (ss->ss_size < p->p_sysent->sv_minsigstksz)
Remove Giant from osigblock(), osigsetmask(), and kern_sigaltstack().
2003-04-23 19:49:18 +00:00
return (ENOMEM);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
td->td_sigstk = *ss;
td->td_pflags |= TDP_ALTSTACK;
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
} else {
Make sigaltstack as per-threaded, because per-process sigaltstack state is useless for threaded programs, multiple threads can not share same stack. The alternative signal stack is private for thread, no lock is needed, the orignal P_ALTSTACK is now moved into td_pflags and renamed to TDP_ALTSTACK. For single thread or Linux clone() based threaded program, there is no semantic changed, because those programs only have one kernel thread in every process. Reviewed by: deischen, dfr
2004-01-03 02:02:26 +00:00
td->td_pflags &= ~TDP_ALTSTACK;
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Remove Giant from osigblock(), osigsetmask(), and kern_sigaltstack().
2003-04-23 19:49:18 +00:00
return (0);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Only return EPERM from kill(-pid) when no process was signalled. As mandated by POSIX. Also clarify the kill(2) manpage. While there, restructure the code in killpg1() to use helper which keeps overall state of the process list iteration in the killpg1_ctx structued, later used to infer the error returned. Reported by: amdmi3 Reviewed by: jilles Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D22621
2019-12-07 18:07:49 +00:00
struct killpg1_ctx {
struct thread *td;
ksiginfo_t *ksi;
int sig;
bool sent;
bool found;
int ret;
};
static void
killpg1_sendsig(struct proc *p, bool notself, struct killpg1_ctx *arg)
{
int err;
if (p->p_pid <= 1 || (p->p_flag & P_SYSTEM) != 0 ||
(notself && p == arg->td->td_proc) || p->p_state == PRS_NEW)
return;
PROC_LOCK(p);
err = p_cansignal(arg->td, p, arg->sig);
if (err == 0 && arg->sig != 0)
pksignal(p, arg->sig, arg->ksi);
PROC_UNLOCK(p);
if (err != ESRCH)
arg->found = true;
if (err == 0)
arg->sent = true;
else if (arg->ret == 0 && err != ESRCH && err != EPERM)
arg->ret = err;
}
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
/*
* Common code for kill process group/broadcast kill.
* cp is calling process.
*/
Be consistent about "static" functions: if the function is marked static in its prototype, mark it static at the definition too. Inspired by: FlexeLint warning #512
2002-09-28 17:15:38 +00:00
static int
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
killpg1(struct thread *td, int sig, int pgid, int all, ksiginfo_t *ksi)
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
{
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
struct proc *p;
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
struct pgrp *pgrp;
Only return EPERM from kill(-pid) when no process was signalled. As mandated by POSIX. Also clarify the kill(2) manpage. While there, restructure the code in killpg1() to use helper which keeps overall state of the process list iteration in the killpg1_ctx structued, later used to infer the error returned. Reported by: amdmi3 Reviewed by: jilles Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D22621
2019-12-07 18:07:49 +00:00
struct killpg1_ctx arg;
arg.td = td;
arg.ksi = ksi;
arg.sig = sig;
arg.sent = false;
arg.found = false;
arg.ret = 0;
Protect the following with a lockmgr lock: allproc zombproc pidhashtbl proc.p_list proc.p_hash nextpid Reviewed by: jhb Obtained from: BSD/OS and netbsd
2000-11-22 07:42:04 +00:00
if (all) {
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
/*
* broadcast
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
*/
Convert the allproc and proctree locks from lockmgr locks to sx locks.
2001-03-28 11:52:56 +00:00
sx_slock(&allproc_lock);
Use FOREACH_PROC_IN_SYSTEM instead of using its unrolled form.
2007-01-17 14:58:53 +00:00
FOREACH_PROC_IN_SYSTEM(p) {
Only return EPERM from kill(-pid) when no process was signalled. As mandated by POSIX. Also clarify the kill(2) manpage. While there, restructure the code in killpg1() to use helper which keeps overall state of the process list iteration in the killpg1_ctx structued, later used to infer the error returned. Reported by: amdmi3 Reviewed by: jilles Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D22621
2019-12-07 18:07:49 +00:00
killpg1_sendsig(p, true, &arg);
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
}
Convert the allproc and proctree locks from lockmgr locks to sx locks.
2001-03-28 11:52:56 +00:00
sx_sunlock(&allproc_lock);
Protect the following with a lockmgr lock: allproc zombproc pidhashtbl proc.p_list proc.p_hash nextpid Reviewed by: jhb Obtained from: BSD/OS and netbsd
2000-11-22 07:42:04 +00:00
} else {
Lock proctree_lock instead of pgrpsess_lock.
2002-04-16 17:11:34 +00:00
sx_slock(&proctree_lock);
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
if (pgid == 0) {
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
/*
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
* zero pgid means send to my process group.
*/
- Change killpg1()'s first argument to be a thread instead of a process so we can use td_ucred. - In killpg1(), the proc lock is sufficient to check if p_stat is SZOMB or not. We don't need sched_lock. - Close some races in psignal(). In psignal() there is a big switch statement based on p_stat. All the different cases are assuming that the process (or thread) isn't going to change state out from under it. To ensure this is true, just lock sched_lock for the entire switch. We practically held it the entire time already anyways. This also simplifies the locking somewhat and actually results in fewer lock operations. - Allow signotify() to be called with the sched_lock held since psignal() now does that. - Use td_ucred in a couple of places.
2002-04-13 23:33:36 +00:00
pgrp = td->td_proc->p_pgrp;
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
PGRP_LOCK(pgrp);
} else {
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
pgrp = pgfind(pgid);
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
if (pgrp == NULL) {
Lock proctree_lock instead of pgrpsess_lock.
2002-04-16 17:11:34 +00:00
sx_sunlock(&proctree_lock);
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
return (ESRCH);
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
}
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
}
Lock proctree_lock instead of pgrpsess_lock.
2002-04-16 17:11:34 +00:00
sx_sunlock(&proctree_lock);
This is a partial commit of the patch from PR 14914: Alot of the code in sys/kern directly accesses the *Q_HEAD and *Q_ENTRY structures for list operations. This patch makes all list operations in sys/kern use the queue(3) macros, rather than directly accessing the *Q_{HEAD,ENTRY} structures. This batch of changes compile to the same object files. Reviewed by: phk Submitted by: Jake Burkholder <jake@checker.org> PR: 14914
1999-11-16 10:56:05 +00:00
LIST_FOREACH(p, &pgrp->pg_members, p_pglist) {
Only return EPERM from kill(-pid) when no process was signalled. As mandated by POSIX. Also clarify the kill(2) manpage. While there, restructure the code in killpg1() to use helper which keeps overall state of the process list iteration in the killpg1_ctx structued, later used to infer the error returned. Reported by: amdmi3 Reviewed by: jilles Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D22621
2019-12-07 18:07:49 +00:00
killpg1_sendsig(p, false, &arg);
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
}
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
PGRP_UNLOCK(pgrp);
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
}
Only return EPERM from kill(-pid) when no process was signalled. As mandated by POSIX. Also clarify the kill(2) manpage. While there, restructure the code in killpg1() to use helper which keeps overall state of the process list iteration in the killpg1_ctx structued, later used to infer the error returned. Reported by: amdmi3 Reviewed by: jilles Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D22621
2019-12-07 18:07:49 +00:00
MPASS(arg.ret != 0 || arg.found || !arg.sent);
if (arg.ret == 0 && !arg.sent)
arg.ret = arg.found ? EPERM : ESRCH;
return (arg.ret);
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
}
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct kill_args {
int pid;
int signum;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_kill(struct thread *td, struct kill_args *uap)
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
{
return (kern_kill(td, uap->pid, uap->signum));
}
int
kern_kill(struct thread *td, pid_t pid, int signum)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_t ksi;
struct proc *p;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Allow to use kill(2) in capability mode, but process can send a signal only to himself. For example abort(3) at first tries to do kill(getpid(), SIGABRT) which was failing in capability mode, so the code was failing back to exit(1). Reviewed by: rwatson Obtained from: WHEEL Systems MFC after: 2 weeks
2012-11-27 10:22:40 +00:00
/*
* A process in capability mode can send signals only to himself.
* The main rationale behind this is that abort(3) is implemented as
* kill(getpid(), SIGABRT).
*/
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
if (IN_CAPABILITY_MODE(td) && pid != td->td_proc->p_pid)
Allow to use kill(2) in capability mode, but process can send a signal only to himself. For example abort(3) at first tries to do kill(getpid(), SIGABRT) which was failing in capability mode, so the code was failing back to exit(1). Reviewed by: rwatson Obtained from: WHEEL Systems MFC after: 2 weeks
2012-11-27 10:22:40 +00:00
return (ECAPMODE);
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
AUDIT_ARG_SIGNUM(signum);
AUDIT_ARG_PID(pid);
if ((u_int)signum > _SIG_MAXSIG)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (EINVAL);
Giant Pushdown clock_gettime() clock_settime() nanosleep() settimeofday() adjtime() getitimer() setitimer() __sysctl() ogetkerninfo() sigaction() osigaction() sigpending() osigpending() osigvec() osigblock() osigsetmask() sigsuspend() osigsuspend() osigstack() sigaltstack() kill() okillpg() trapsignal() nosys()
2001-09-01 18:19:21 +00:00
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_init(&ksi);
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
ksi.ksi_signo = signum;
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksi.ksi_code = SI_USER;
ksi.ksi_pid = td->td_proc->p_pid;
ksi.ksi_uid = td->td_ucred->cr_ruid;
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
if (pid > 0) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* kill single process */
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
if ((p = pfind_any(pid)) == NULL)
Add pfind_any It looks for both regular and zombie processes. This avoids allproc relocking previously seen with pfind -> zpfind calls.
2017-11-11 18:04:39 +00:00
return (ESRCH);
Replace AUDIT_ARG() with variable argument macros with a set more more specific macros for each audit argument type. This makes it easier to follow call-graphs, especially for automated analysis tools (such as fxr). In MFC, we should leave the existing AUDIT_ARG() macros as they may be used by third-party kernel modules. Suggested by: brooks Approved by: re (kib) Obtained from: TrustedBSD Project MFC after: 1 week
2009-06-27 13:58:44 +00:00
AUDIT_ARG_PROCESS(p);
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
error = p_cansignal(td, p, signum);
if (error == 0 && signum)
pksignal(p, signum, &ksi);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
PROC_UNLOCK(p);
return (error);
}
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
switch (pid) {
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
case -1: /* broadcast signal */
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
return (killpg1(td, signum, 0, 1, &ksi));
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
case 0: /* signal own process group */
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
return (killpg1(td, signum, 0, 0, &ksi));
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
default: /* negative explicit process group */
Add kern_kill() and use it in Linuxulator. It's just a cleanup, no functional changes. Reviewed by: kib MFC after: 2 weeks Sponsored by: The FreeBSD Foundation Differential Revision: https://reviews.freebsd.org/D22645
2019-12-13 18:44:02 +00:00
return (killpg1(td, signum, -pid, 0, &ksi));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
/* NOTREACHED */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Add experimental support for process descriptors A "process descriptor" file descriptor is used to manage processes without using the PID namespace. This is required for Capsicum's Capability Mode, where the PID namespace is unavailable. New system calls pdfork(2) and pdkill(2) offer the functional equivalents of fork(2) and kill(2). pdgetpid(2) allows querying the PID of the remote process for debugging purposes. The currently-unimplemented pdwait(2) will, in the future, allow querying rusage/exit status. In the interim, poll(2) may be used to check (and wait for) process termination. When a process is referenced by a process descriptor, it does not issue SIGCHLD to the parent, making it suitable for use in libraries---a common scenario when using library compartmentalisation from within large applications (such as web browsers). Some observers may note a similarity to Mach task ports; process descriptors provide a subset of this behaviour, but in a UNIX style. This feature is enabled by "options PROCDESC", but as with several other Capsicum kernel features, is not enabled by default in GENERIC 9.0. Reviewed by: jhb, kib Approved by: re (kib), mentor (rwatson) Sponsored by: Google Inc
2011-08-18 22:51:30 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sys_pdkill(struct thread *td, struct pdkill_args *uap)
Add experimental support for process descriptors A "process descriptor" file descriptor is used to manage processes without using the PID namespace. This is required for Capsicum's Capability Mode, where the PID namespace is unavailable. New system calls pdfork(2) and pdkill(2) offer the functional equivalents of fork(2) and kill(2). pdgetpid(2) allows querying the PID of the remote process for debugging purposes. The currently-unimplemented pdwait(2) will, in the future, allow querying rusage/exit status. In the interim, poll(2) may be used to check (and wait for) process termination. When a process is referenced by a process descriptor, it does not issue SIGCHLD to the parent, making it suitable for use in libraries---a common scenario when using library compartmentalisation from within large applications (such as web browsers). Some observers may note a similarity to Mach task ports; process descriptors provide a subset of this behaviour, but in a UNIX style. This feature is enabled by "options PROCDESC", but as with several other Capsicum kernel features, is not enabled by default in GENERIC 9.0. Reviewed by: jhb, kib Approved by: re (kib), mentor (rwatson) Sponsored by: Google Inc
2011-08-18 22:51:30 +00:00
{
struct proc *p;
int error;
AUDIT_ARG_SIGNUM(uap->signum);
AUDIT_ARG_FD(uap->fd);
if ((u_int)uap->signum > _SIG_MAXSIG)
return (EINVAL);
Eliminate the overhead of gratuitous repeated reinitialization of cap_rights - Add macros to allow preinitialization of cap_rights_t. - Convert most commonly used code paths to use preinitialized cap_rights_t. A 3.6% speedup in fstat was measured with this change. Reported by: mjg Reviewed by: oshogbo Approved by: sbruno MFC after: 1 month
2018-05-09 18:47:24 +00:00
error = procdesc_find(td, uap->fd, &cap_pdkill_rights, &p);
Add experimental support for process descriptors A "process descriptor" file descriptor is used to manage processes without using the PID namespace. This is required for Capsicum's Capability Mode, where the PID namespace is unavailable. New system calls pdfork(2) and pdkill(2) offer the functional equivalents of fork(2) and kill(2). pdgetpid(2) allows querying the PID of the remote process for debugging purposes. The currently-unimplemented pdwait(2) will, in the future, allow querying rusage/exit status. In the interim, poll(2) may be used to check (and wait for) process termination. When a process is referenced by a process descriptor, it does not issue SIGCHLD to the parent, making it suitable for use in libraries---a common scenario when using library compartmentalisation from within large applications (such as web browsers). Some observers may note a similarity to Mach task ports; process descriptors provide a subset of this behaviour, but in a UNIX style. This feature is enabled by "options PROCDESC", but as with several other Capsicum kernel features, is not enabled by default in GENERIC 9.0. Reviewed by: jhb, kib Approved by: re (kib), mentor (rwatson) Sponsored by: Google Inc
2011-08-18 22:51:30 +00:00
if (error)
return (error);
AUDIT_ARG_PROCESS(p);
error = p_cansignal(td, p, uap->signum);
if (error == 0 && uap->signum)
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
kern_psignal(p, uap->signum);
Add experimental support for process descriptors A "process descriptor" file descriptor is used to manage processes without using the PID namespace. This is required for Capsicum's Capability Mode, where the PID namespace is unavailable. New system calls pdfork(2) and pdkill(2) offer the functional equivalents of fork(2) and kill(2). pdgetpid(2) allows querying the PID of the remote process for debugging purposes. The currently-unimplemented pdwait(2) will, in the future, allow querying rusage/exit status. In the interim, poll(2) may be used to check (and wait for) process termination. When a process is referenced by a process descriptor, it does not issue SIGCHLD to the parent, making it suitable for use in libraries---a common scenario when using library compartmentalisation from within large applications (such as web browsers). Some observers may note a similarity to Mach task ports; process descriptors provide a subset of this behaviour, but in a UNIX style. This feature is enabled by "options PROCDESC", but as with several other Capsicum kernel features, is not enabled by default in GENERIC 9.0. Reviewed by: jhb, kib Approved by: re (kib), mentor (rwatson) Sponsored by: Google Inc
2011-08-18 22:51:30 +00:00
PROC_UNLOCK(p);
return (error);
}
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#if defined(COMPAT_43)
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct okillpg_args {
int pgid;
int signum;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
okillpg(struct thread *td, struct okillpg_args *uap)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_t ksi;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Replace AUDIT_ARG() with variable argument macros with a set more more specific macros for each audit argument type. This makes it easier to follow call-graphs, especially for automated analysis tools (such as fxr). In MFC, we should leave the existing AUDIT_ARG() macros as they may be used by third-party kernel modules. Suggested by: brooks Approved by: re (kib) Obtained from: TrustedBSD Project MFC after: 1 week
2009-06-27 13:58:44 +00:00
AUDIT_ARG_SIGNUM(uap->signum);
AUDIT_ARG_PID(uap->pgid);
_SIG_MAXSIG (128) is the highest legal signal. The arrays are offset by one - see _SIG_IDX(). Revert part of my mis-correction in kern_sig.c (but signal 0 still has to be allowed) and fix _SIG_VALID() (it was rejecting ignal 128).
2001-11-03 13:26:15 +00:00
if ((u_int)uap->signum > _SIG_MAXSIG)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (EINVAL);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_init(&ksi);
ksi.ksi_signo = uap->signum;
ksi.ksi_code = SI_USER;
ksi.ksi_pid = td->td_proc->p_pid;
ksi.ksi_uid = td->td_ucred->cr_ruid;
return (killpg1(td, uap->signum, uap->pgid, 0, &ksi));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Deorbit COMPAT_SUNOS. We inherited this from the sparc32 port of BSD4.4-Lite1. We have neither a sparc32 port nor a SunOS4.x compatibility desire these days.
2004-06-11 11:16:26 +00:00
#endif /* COMPAT_43 */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
#ifndef _SYS_SYSPROTO_H_
struct sigqueue_args {
pid_t pid;
int signum;
/* union sigval */ void *value;
};
#endif
int
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
sys_sigqueue(struct thread *td, struct sigqueue_args *uap)
Provide a freebsd32 implementation of sigqueue() The previous misuse of sys_sigqueue() was sending random register or stack garbage to 64-bit targets. The freebsd32 implementation preserves the sival_int member of value when signaling a 64-bit process. Document the mixed ABI implementation of union sigval and the incompability of sival_ptr with pointer integrity schemes. Reviewed by: kib, wblock MFC after: 1 week Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D10605
2017-05-05 18:49:39 +00:00
{
union sigval sv;
sv.sival_ptr = uap->value;
return (kern_sigqueue(td, uap->pid, uap->signum, &sv));
}
int
kern_sigqueue(struct thread *td, pid_t pid, int signum, union sigval *value)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
ksiginfo_t ksi;
struct proc *p;
int error;
Provide a freebsd32 implementation of sigqueue() The previous misuse of sys_sigqueue() was sending random register or stack garbage to 64-bit targets. The freebsd32 implementation preserves the sival_int member of value when signaling a 64-bit process. Document the mixed ABI implementation of union sigval and the incompability of sival_ptr with pointer integrity schemes. Reviewed by: kib, wblock MFC after: 1 week Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D10605
2017-05-05 18:49:39 +00:00
if ((u_int)signum > _SIG_MAXSIG)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (EINVAL);
/*
* Specification says sigqueue can only send signal to
* single process.
*/
Provide a freebsd32 implementation of sigqueue() The previous misuse of sys_sigqueue() was sending random register or stack garbage to 64-bit targets. The freebsd32 implementation preserves the sival_int member of value when signaling a 64-bit process. Document the mixed ABI implementation of union sigval and the incompability of sival_ptr with pointer integrity schemes. Reviewed by: kib, wblock MFC after: 1 week Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D10605
2017-05-05 18:49:39 +00:00
if (pid <= 0)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (EINVAL);
Use pfind_any in linux_rt_sigqueueinfo and kern_sigqueue
2017-11-11 18:10:09 +00:00
if ((p = pfind_any(pid)) == NULL)
return (ESRCH);
Provide a freebsd32 implementation of sigqueue() The previous misuse of sys_sigqueue() was sending random register or stack garbage to 64-bit targets. The freebsd32 implementation preserves the sival_int member of value when signaling a 64-bit process. Document the mixed ABI implementation of union sigval and the incompability of sival_ptr with pointer integrity schemes. Reviewed by: kib, wblock MFC after: 1 week Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D10605
2017-05-05 18:49:39 +00:00
error = p_cansignal(td, p, signum);
if (error == 0 && signum != 0) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_init(&ksi);
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksi.ksi_flags = KSI_SIGQ;
Provide a freebsd32 implementation of sigqueue() The previous misuse of sys_sigqueue() was sending random register or stack garbage to 64-bit targets. The freebsd32 implementation preserves the sival_int member of value when signaling a 64-bit process. Document the mixed ABI implementation of union sigval and the incompability of sival_ptr with pointer integrity schemes. Reviewed by: kib, wblock MFC after: 1 week Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D10605
2017-05-05 18:49:39 +00:00
ksi.ksi_signo = signum;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksi.ksi_code = SI_QUEUE;
ksi.ksi_pid = td->td_proc->p_pid;
ksi.ksi_uid = td->td_ucred->cr_ruid;
Provide a freebsd32 implementation of sigqueue() The previous misuse of sys_sigqueue() was sending random register or stack garbage to 64-bit targets. The freebsd32 implementation preserves the sival_int member of value when signaling a 64-bit process. Document the mixed ABI implementation of union sigval and the incompability of sival_ptr with pointer integrity schemes. Reviewed by: kib, wblock MFC after: 1 week Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D10605
2017-05-05 18:49:39 +00:00
ksi.ksi_value = *value;
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
error = pksignal(p, ksi.ksi_signo, &ksi);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
}
PROC_UNLOCK(p);
return (error);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Send a signal to a process group.
*/
void
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
gsignal(int pgid, int sig, ksiginfo_t *ksi)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
struct pgrp *pgrp;
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
if (pgid != 0) {
Lock proctree_lock instead of pgrpsess_lock.
2002-04-16 17:11:34 +00:00
sx_slock(&proctree_lock);
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
pgrp = pgfind(pgid);
Lock proctree_lock instead of pgrpsess_lock.
2002-04-16 17:11:34 +00:00
sx_sunlock(&proctree_lock);
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
if (pgrp != NULL) {
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
pgsignal(pgrp, sig, 0, ksi);
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
PGRP_UNLOCK(pgrp);
}
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
This is the kernel Lite/2 commit. There are some requisite userland changes, so don't expect to be able to run the kernel as-is (very well) without the appropriate Lite/2 userland changes. The system boots and can mount UFS filesystems. Untested: ext2fs, msdosfs, NFS Known problems: Incorrect Berkeley ID strings in some files. Mount_std mounts will not work until the getfsent library routine is changed. Reviewed by: various people Submitted by: Jeffery Hsu <hsu@freebsd.org>
1997-02-10 02:22:35 +00:00
* Send a signal to a process group. If checktty is 1,
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* limit to members which have a controlling terminal.
*/
void
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
pgsignal(struct pgrp *pgrp, int sig, int checkctty, ksiginfo_t *ksi)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
struct proc *p;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
if (pgrp) {
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
PGRP_LOCK_ASSERT(pgrp, MA_OWNED);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
LIST_FOREACH(p, &pgrp->pg_members, p_pglist) {
PROC_LOCK(p);
Fix several places to ignore processes that are not yet fully constructed. MFC after: 1 week
2011-04-06 17:47:22 +00:00
if (p->p_state == PRS_NORMAL &&
(checkctty == 0 || p->p_flag & P_CONTROLT))
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
pksignal(p, sig, ksi);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
}
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fix SA_SIGINFO | SA_RESETHAND handling. The sysent' sv_sendsig() method needs pre-reset state of the ps_siginfo to correctly construct signal frame. Move sigdflt() call after the sv_sendsig() invocation in postsig(). Simultaneously extract common code from trapsignal() and postsig() into new helper postsig_done(). Submitted by: rea MFC after: 1 week
2014-11-26 14:09:04 +00:00
/*
* Recalculate the signal mask and reset the signal disposition after
* usermode frame for delivery is formed. Should be called after
* mach-specific routine, because sysent->sv_sendsig() needs correct
* ps_siginfo and signal mask.
*/
static void
postsig_done(int sig, struct thread *td, struct sigacts *ps)
{
sigset_t mask;
mtx_assert(&ps->ps_mtx, MA_OWNED);
td->td_ru.ru_nsignals++;
mask = ps->ps_catchmask[_SIG_IDX(sig)];
if (!SIGISMEMBER(ps->ps_signodefer, sig))
SIGADDSET(mask, sig);
kern_sigprocmask(td, SIG_BLOCK, &mask, NULL,
SIGPROCMASK_PROC_LOCKED | SIGPROCMASK_PS_LOCKED);
if (SIGISMEMBER(ps->ps_sigreset, sig))
sigdflt(ps, sig);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Remove 'MPSAFE' annotations from the comments above most system calls: all system calls now enter without Giant held, and then in some cases, acquire Giant explicitly. Remove a number of other MPSAFE annotations in the credential code and tweak one or two other adjacent comments.
2007-03-04 22:36:48 +00:00
* Send a signal caused by a trap to the current thread. If it will be
* caught immediately, deliver it with correct code. Otherwise, post it
* normally.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
void
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
trapsignal(struct thread *td, ksiginfo_t *ksi)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
- Change trapsignal() to accept a thread and not a proc. - Change all consumers to pass in a thread. Right now this does not cause any functional changes but it will be important later when signals can be delivered to specific threads.
2003-03-31 22:02:38 +00:00
struct sigacts *ps;
struct proc *p;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
sigset_t sigmask;
int code, sig;
- Change trapsignal() to accept a thread and not a proc. - Change all consumers to pass in a thread. Right now this does not cause any functional changes but it will be important later when signals can be delivered to specific threads.
2003-03-31 22:02:38 +00:00
p = td->td_proc;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sig = ksi->ksi_signo;
code = ksi->ksi_code;
KASSERT(_SIG_VALID(sig), ("invalid signal"));
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
sigfastblock_fetch(td);
Make KSE a kernel option, turned on by default in all GENERIC kernel configs except sun4v (which doesn't process signals properly with KSE). Reviewed by: davidxu@
2006-10-26 21:42:22 +00:00
PROC_LOCK(p);
Hold the proc lock while accessing p_procsig in trapsignal().
2003-03-09 01:40:55 +00:00
ps = p->p_sigacts;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_lock(&ps->ps_mtx);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
sigmask = td->td_sigmask;
if (td->td_sigblock_val != 0)
SIGSETOR(sigmask, fastblock_mask);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
if ((p->p_flag & P_TRACED) == 0 && SIGISMEMBER(ps->ps_sigcatch, sig) &&
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
!SIGISMEMBER(sigmask, sig)) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#ifdef KTRACE
- trapsignal() no longer needs to acquire Giant for ktrpsig(). - Catch up to new ktrace API.
2002-06-07 05:43:02 +00:00
if (KTRPOINT(curthread, KTR_PSIG))
ktrpsig(sig, ps->ps_sigact[_SIG_IDX(sig)],
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
&td->td_sigmask, code);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#endif
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
(*p->p_sysent->sv_sendsig)(ps->ps_sigact[_SIG_IDX(sig)],
Make KSE a kernel option, turned on by default in all GENERIC kernel configs except sun4v (which doesn't process signals properly with KSE). Reviewed by: davidxu@
2006-10-26 21:42:22 +00:00
ksi, &td->td_sigmask);
Fix SA_SIGINFO | SA_RESETHAND handling. The sysent' sv_sendsig() method needs pre-reset state of the ps_siginfo to correctly construct signal frame. Move sigdflt() call after the sv_sendsig() invocation in postsig(). Simultaneously extract common code from trapsignal() and postsig() into new helper postsig_done(). Submitted by: rea MFC after: 1 week
2014-11-26 14:09:04 +00:00
postsig_done(sig, td, ps);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
Don't let osigaction and osigvec accept the new signal numbers. Fix style bugs caused by the sigset_t in general while I'm here. Submitted by: bde
1999-10-12 13:14:18 +00:00
} else {
Add a sysctl to force a process to sigexit if a trap signal is being hold by current thread or ignored by current process, otherwise, it is very possible the thread will enter an infinite loop and lead to an administrator's nightmare.
2005-12-09 08:29:29 +00:00
/*
* Avoid a possible infinite loop if the thread
* masking the signal or process is ignoring the
* signal.
*/
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
if (kern_forcesigexit && (SIGISMEMBER(sigmask, sig) ||
ps->ps_sigact[_SIG_IDX(sig)] == SIG_IGN)) {
Add a sysctl to force a process to sigexit if a trap signal is being hold by current thread or ignored by current process, otherwise, it is very possible the thread will enter an infinite loop and lead to an administrator's nightmare.
2005-12-09 08:29:29 +00:00
SIGDELSET(td->td_sigmask, sig);
SIGDELSET(ps->ps_sigcatch, sig);
SIGDELSET(ps->ps_sigignore, sig);
ps->ps_sigact[_SIG_IDX(sig)] = SIG_DFL;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
td->td_pflags &= ~TDP_SIGFASTBLOCK;
td->td_sigblock_val = 0;
Add a sysctl to force a process to sigexit if a trap signal is being hold by current thread or ignored by current process, otherwise, it is very possible the thread will enter an infinite loop and lead to an administrator's nightmare.
2005-12-09 08:29:29 +00:00
}
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
p->p_sig = sig; /* XXX to verify code */
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
tdsendsignal(p, td, sig, ksi);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
static struct thread *
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
sigtd(struct proc *p, int sig, bool fast_sigblock)
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
{
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
struct thread *td, *signal_td;
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
MPASS(!fast_sigblock || p == curproc);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
/*
The td_waitset is pointing to a stack address when thread is waiting for a signal, because kernel stack is swappable, this causes page fault in kernel under heavy swapping case. Fix this bug by eliminating unneeded code.
2005-03-04 22:46:31 +00:00
* Check if current thread can handle the signal without
Fix typo. MFC after: 3 days
2009-10-01 12:46:58 +00:00
* switching context to another thread.
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
*/
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
if (curproc == p && !SIGISMEMBER(curthread->td_sigmask, sig) &&
(!fast_sigblock || curthread->td_sigblock_val == 0))
The td_waitset is pointing to a stack address when thread is waiting for a signal, because kernel stack is swappable, this causes page fault in kernel under heavy swapping case. Fix this bug by eliminating unneeded code.
2005-03-04 22:46:31 +00:00
return (curthread);
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
signal_td = NULL;
FOREACH_THREAD_IN_PROC(p, td) {
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
if (!SIGISMEMBER(td->td_sigmask, sig) && (!fast_sigblock ||
td != curthread || td->td_sigblock_val == 0)) {
The td_waitset is pointing to a stack address when thread is waiting for a signal, because kernel stack is swappable, this causes page fault in kernel under heavy swapping case. Fix this bug by eliminating unneeded code.
2005-03-04 22:46:31 +00:00
signal_td = td;
break;
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
}
}
if (signal_td == NULL)
signal_td = FIRST_THREAD_IN_PROC(p);
return (signal_td);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Send the signal to the process. If the signal has an action, the action
* is usually performed by the target process rather than the caller; we add
* the signal to the set of pending signals for the process.
*
* Exceptions:
* o When a stop signal is sent to a sleeping process that takes the
* default action, the process is stopped without awakening it.
* o SIGCONT restarts stopped processes (or puts them back to sleep)
* regardless of the signal action (eg, blocked or ignored).
*
* Other ignored signals are discarded immediately.
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
*
Comment that tdsignal() may be entered from the debugger.
2007-05-23 17:27:42 +00:00
* NB: This function may be entered from the debugger via the "kill" DDB
* command. There is little that can be done to mitigate the possibly messy
* side effects of this unwise possibility.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
void
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
kern_psignal(struct proc *p, int sig)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_t ksi;
ksiginfo_init(&ksi);
ksi.ksi_signo = sig;
ksi.ksi_code = SI_KERNEL;
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
(void) tdsendsignal(p, NULL, sig, &ksi);
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
}
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
int
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
pksignal(struct proc *p, int sig, ksiginfo_t *ksi)
{
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
return (tdsendsignal(p, NULL, sig, ksi));
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
}
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
/* Utility function for finding a thread to send signal event to. */
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
int
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
sigev_findtd(struct proc *p ,struct sigevent *sigev, struct thread **ttd)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
{
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
struct thread *td;
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
if (sigev->sigev_notify == SIGEV_THREAD_ID) {
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
td = tdfind(sigev->sigev_notify_thread_id, p->p_pid);
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
if (td == NULL)
return (ESRCH);
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
*ttd = td;
} else {
*ttd = NULL;
PROC_LOCK(p);
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
}
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
return (0);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
}
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
void
tdsignal(struct thread *td, int sig)
{
ksiginfo_t ksi;
ksiginfo_init(&ksi);
ksi.ksi_signo = sig;
ksi.ksi_code = SI_KERNEL;
(void) tdsendsignal(td->td_proc, td, sig, &ksi);
}
void
tdksignal(struct thread *td, int sig, ksiginfo_t *ksi)
{
(void) tdsendsignal(td->td_proc, td, sig, ksi);
}
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
int
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
tdsendsignal(struct proc *p, struct thread *td, int sig, ksiginfo_t *ksi)
o Change kse_thr_interrupt to allow send a signal to a specified thread, or unblock a thread in kernel, and allow UTS to specify whether syscall should be restarted. o Add ability for UTS to monitor signal comes in and removed from process, the flag PS_SIGEVENT is used to indicate the events. o Add a KMF_WAITSIGEVENT for KSE mailbox flag, UTS call kse_release with this flag set to wait for above signal event. o For SA based thread, kernel masks all signal in its signal mask, let UTS to use kse_thr_interrupt interrupt a thread, and install a signal frame in userland for the thread. o Add a tm_syncsig in thread mailbox, when a hardware trap occurs, it is used to deliver synchronous signal to userland, and upcall is schedule, so UTS can process the synchronous signal for the thread. Reviewed by: julian (mentor)
2003-06-28 08:29:05 +00:00
{
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sig_t action;
sigqueue_t *sigqueue;
int prop;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
struct sigacts *ps;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
int intrval;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
int ret = 0;
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
int wakeup_swapper;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Create a global thread hash table to speed up thread lookup, use rwlock to protect the table. In old code, thread lookup is done with process lock held, to find a thread, kernel has to iterate through process and thread list, this is quite inefficient. With this change, test shows in extreme case performance is dramatically improved. Earlier patch was reviewed by: jhb, julian
2010-10-09 02:50:23 +00:00
MPASS(td == NULL || p == td->td_proc);
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
panic() if we try to handle an out-of-range signal number in psignal()/tdsignal(). The test was historically in psignal(). It was changed into a KASSERT, and then later moved to tdsignal() when the latter was introduced. Reviewed by: iedowse, jhb
2003-08-10 23:05:37 +00:00
if (!_SIG_VALID(sig))
Fix function name in error messages.
2010-07-20 02:23:12 +00:00
panic("%s(): invalid signal %d", __func__, sig);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
Fix function name in error messages.
2010-07-20 02:23:12 +00:00
KASSERT(ksi == NULL || !KSI_ONQ(ksi), ("%s: ksi on queue", __func__));
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
/*
* IEEE Std 1003.1-2001: return success when killing a zombie.
*/
if (p->p_state == PRS_ZOMBIE) {
if (ksi && (ksi->ksi_flags & KSI_INS))
ksiginfo_tryfree(ksi);
return (ret);
}
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
ps = p->p_sigacts;
When filt_proc() removes event from the knlist due to the process exiting (NOTE_EXIT->knlist_remove_inevent()), two things happen: - knote kn_knlist pointer is reset - INFLUX knote is removed from the process knlist. And, there are two consequences: - KN_LIST_UNLOCK() on such knote is nop - there is nothing which would block exit1() from processing past the knlist_destroy() (and knlist_destroy() resets knlist lock pointers). Both consequences result either in leaked process lock, or dereferencing NULL function pointers for locking. Handle this by stopping embedding the process knlist into struct proc. Instead, the knlist is allocated together with struct proc, but marked as autodestroy on the zombie reap, by knlist_detach() function. The knlist is freed when last kevent is removed from the list, in particular, at the zombie reap time if the list is empty. As result, the knlist_remove_inevent() is no longer needed and removed. Other changes: In filt_procattach(), clear NOTE_EXEC and NOTE_FORK desired events from kn_sfflags for knote registered by kernel to only get NOTE_CHILD notifications. The flags leak resulted in excessive NOTE_EXEC/NOTE_FORK reports. Fix immediate note activation in filt_procattach(). Condition should be either the immediate CHILD_NOTE activation, or immediate NOTE_EXIT report for the exiting process. In knote_fork(), do not perform racy check for KN_INFLUX before kq lock is taken. Besides being racy, it did not accounted for notes just added by scan (KN_SCAN). Some minor and incomplete style fixes. Analyzed and tested by: Eric Badger <eric@badgerio.us> Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb) Differential revision: https://reviews.freebsd.org/D6859
2016-06-27 21:52:17 +00:00
KNOTE_LOCKED(p->p_klist, NOTE_SIGNAL | sig);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
prop = sigprop(sig);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
if (td == NULL) {
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
td = sigtd(p, sig, false);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
sigqueue = &p->p_sigqueue;
Remove duplicate assertion from tdsendsignal. MFC after: 2 weeks
2013-07-22 00:44:37 +00:00
} else
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
sigqueue = &td->td_sigqueue;
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
Fix style issues around existing SDT probes. - Use SDT_PROBE<N>() instead of SDT_PROBE(). This has no functional effect at the moment, but will be needed for some future changes. - Don't hardcode the module component of the probe identifier. This is set automatically by the SDT framework. MFC after: 1 week
2015-12-16 23:39:27 +00:00
SDT_PROBE3(proc, , , signal__send, td, p, sig);
Add DTrace 'proc' provider probes using the Statically Defined Trace (sdt) mechanism.
2008-05-24 06:22:16 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
* If the signal is being ignored,
* then we forget about it immediately.
* (Note: we don't set SIGCONT in ps_sigignore,
* and if it is set to SIG_IGN,
* action will be SIG_DFL here.)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
mtx_lock(&ps->ps_mtx);
Remove the check that prevented signals from being delivered to exiting processes. It was originally added back when support for Linux threads (and thus shared sigacts objects) was added, but no one knows why. My guess is that at some point during the Linux threads patches, the sigacts object was torn down during exit1(), so this check was added to prevent a panic for that race. However, the stuff that was actually committed to the tree doesn't teardown sigacts until wait() making the above race moot. Re-allowing signals here lets one interrupt a NFS request during process teardown (such as closing descriptors) on an interruptible mount. Requested by: kib (long time ago) MFC after: 1 week
2006-10-20 16:19:21 +00:00
if (SIGISMEMBER(ps->ps_sigignore, sig)) {
Fix style issues around existing SDT probes. - Use SDT_PROBE<N>() instead of SDT_PROBE(). This has no functional effect at the moment, but will be needed for some future changes. - Don't hardcode the module component of the probe identifier. This is set automatically by the SDT framework. MFC after: 1 week
2015-12-16 23:39:27 +00:00
SDT_PROBE3(proc, , , signal__discard, td, p, sig);
Add DTrace 'proc' provider probes using the Statically Defined Trace (sdt) mechanism.
2008-05-24 06:22:16 +00:00
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
if (ksi && (ksi->ksi_flags & KSI_INS))
ksiginfo_tryfree(ksi);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (ret);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
if (SIGISMEMBER(td->td_sigmask, sig))
action = SIG_HOLD;
else if (SIGISMEMBER(ps->ps_sigcatch, sig))
action = SIG_CATCH;
else
action = SIG_DFL;
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
if (SIGISMEMBER(ps->ps_sigintr, sig))
intrval = EINTR;
else
intrval = ERESTART;
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
mtx_unlock(&ps->ps_mtx);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (prop & SIGPROP_CONT)
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete_stopmask_proc(p);
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
else if (prop & SIGPROP_STOP) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* If sending a tty stop signal to a member of an orphaned
* process group, discard the signal here if the action
* is default; don't stop the process below if sleeping,
* and don't clear any pending SIGCONT.
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if ((prop & SIGPROP_TTYSTOP) &&
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
(p->p_pgrp->pg_jobc == 0) &&
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
(action == SIG_DFL)) {
if (ksi && (ksi->ksi_flags & KSI_INS))
ksiginfo_tryfree(ksi);
return (ret);
}
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete_proc(p, SIGCONT);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
if (p->p_flag & P_CONTINUED) {
p->p_flag &= ~P_CONTINUED;
PROC_LOCK(p->p_pptr);
sigqueue_take(p->p_ksi);
PROC_UNLOCK(p->p_pptr);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fix sigwait to conform to POSIX. When a signal is being delivered to process, first find a sigwait thread to deliver, POSIX's argument is speed of delivering signal to sigwait thread is faster than other ways. A signal in its wait set will cause sigwait to return the signal number, a signal not in its wait set but in not blocked by the thread also causes sigwait to return, but sigwait returns EINTR, sigwait is oneshot operation, only one signal can be delivered to its wait set, when a signal is delivered to the sigwait thread, the thread's sigwait state is canceled.
2003-07-17 22:52:55 +00:00
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ret = sigqueue_add(sigqueue, sig, ksi);
if (ret != 0)
return (ret);
Cleanup some signal interfaces. Now the tdsignal function accepts both proc pointer and thread pointer, if thread pointer is NULL, tdsignal automatically finds a thread, otherwise it sends signal to given thread. Add utility function psignal_event to send a realtime sigevent to a process according to the delivery requirement specified in struct sigevent.
2005-11-03 04:49:16 +00:00
signotify(td);
Check SIG_HOLD action ealier to avoid missing test it in later code.
2003-04-12 00:38:47 +00:00
/*
* Defer further processing for signals which are held,
* except that stopped processes must be continued by SIGCONT.
*/
if (action == SIG_HOLD &&
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
!((prop & SIGPROP_CONT) && (p->p_flag & P_STOPPED_SIG)))
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (ret);
Don't clear p_ptevents on normal SIGKILL delivery The ptrace() user has the option of discarding the signal. In such a case, p_ptevents should not be modified. If the ptrace() user decides to send a SIGKILL, ptevents will be cleared in ptracestop(). procfs events do not have the capability to discard the signal, so continue to clear the mask in that case. Reviewed by: jhb (initial revision) MFC after: 1 week Sponsored by: Dell EMC Differential Revision: https://reviews.freebsd.org/D9939
2017-03-16 13:03:31 +00:00
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
wakeup_swapper = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* Some signals have a process-wide effect and a per-thread
* component. Most processing occurs when the process next
* tries to cross the user boundary, however there are some
Tweak some comments.
2013-03-18 18:04:09 +00:00
* times when processing needs to be done immediately, such as
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* waking up threads so that they can cross the user boundary.
Tweak some comments.
2013-03-18 18:04:09 +00:00
* We try to do the per-process part here.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
if (P_SHOULDSTOP(p)) {
Ignore stop and continue signals sent to an exiting process. Stop signals set p_xstat to the signal that triggered the stop, but p_xstat is also used to hold the exit status of an exiting process. Without this change, a stop signal that arrived after a process was marked P_WEXIT but before it was marked a zombie would overwrite the exit status with the stop signal number. Reviewed by: kib MFC after: 1 week
2012-09-13 15:51:18 +00:00
KASSERT(!(p->p_flag & P_WEXIT),
("signal to stopped but exiting process"));
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
if (sig == SIGKILL) {
rescure comments from RELENG_4.
2010-09-01 01:26:07 +00:00
/*
* If traced process is already stopped,
* then no further action is necessary.
*/
If a process is being debugged, skips job control caused by SIGSTOP/SIGCONT signals, because it is managed by debugger, however a normal signal sent to a interruptibly sleeping thread wakes up the thread so it will handle the signal when the process leaves the stopped state. PR: 150138 MFC after: 1 week
2010-08-31 07:15:50 +00:00
if (p->p_flag & P_TRACED)
goto out;
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
* SIGKILL sets process running.
* It will die elsewhere.
* All threads must be restarted.
*/
Don't bother to turn off other P_STOPPED bits for SIGKILL, doing so would cause kernel to produce an unkillable process in some cases, especially, P_STOPPED_SINGLE has a singling thread, turning off the bit would mess the state.
2004-10-03 13:23:49 +00:00
p->p_flag &= ~P_STOPPED_SIG;
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
goto runfast;
}
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (prop & SIGPROP_CONT) {
rescure comments from RELENG_4.
2010-09-01 01:26:07 +00:00
/*
* If traced process is already stopped,
* then no further action is necessary.
*/
If a process is being debugged, skips job control caused by SIGSTOP/SIGCONT signals, because it is managed by debugger, however a normal signal sent to a interruptibly sleeping thread wakes up the thread so it will handle the signal when the process leaves the stopped state. PR: 150138 MFC after: 1 week
2010-08-31 07:15:50 +00:00
if (p->p_flag & P_TRACED)
goto out;
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
* If SIGCONT is default (or ignored), we continue the
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
* process but don't leave the signal in sigqueue as
Back our kernel support for reliable signal queues. Requested by: rwatson, phk, and many others
2002-10-01 17:15:53 +00:00
* it has no further action. If SIGCONT is held, we
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* continue the process and leave the signal in
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
* sigqueue. If the process catches SIGCONT, let it
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* handle the signal itself. If it isn't waiting on
* an event, it goes back to run state.
* Otherwise, process goes back to sleep state.
*/
s/SGNL/SIG/ s/SNGL/SINGLE/ s/SNGLE/SINGLE/ Fix abbreviation for P_STOPPED_* etc flags, in original code they were inconsistent and difficult to distinguish between them. Approved by: julian (mentor)
2002-09-05 07:30:18 +00:00
p->p_flag &= ~P_STOPPED_SIG;
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
if (p->p_numthreads == p->p_suspcount) {
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
p->p_flag |= P_CONTINUED;
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
p->p_xsig = SIGCONT;
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
PROC_LOCK(p->p_pptr);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
childproc_continued(p);
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
PROC_UNLOCK(p->p_pptr);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
}
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
if (action == SIG_DFL) {
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_unsuspend(p);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete(sigqueue, sig);
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
goto out;
}
if (action == SIG_CATCH) {
Make KSE a kernel option, turned on by default in all GENERIC kernel configs except sun4v (which doesn't process signals properly with KSE). Reviewed by: davidxu@
2006-10-26 21:42:22 +00:00
/*
* The process wants to catch it so it needs
* to run at least one thread, but which one?
*/
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
goto runfast;
}
/*
* The signal is not ignored or caught.
*/
Slight cleanup of some comments/whitespace. Make idle process state more consistant. Add an assert on thread state. Clean up idleproc/mi_switch() interaction. Use a local instead of referencing curthread 7 times in a row (I've been told curthread can be expensive on some architectures) Remove some commented out code. Add a little commented out code (completion coming soon) Reviewed by: jhb@freebsd.org
2002-08-01 18:45:10 +00:00
thread_unsuspend(p);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
}
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (prop & SIGPROP_STOP) {
rescure comments from RELENG_4.
2010-09-01 01:26:07 +00:00
/*
* If traced process is already stopped,
* then no further action is necessary.
*/
If a process is being debugged, skips job control caused by SIGSTOP/SIGCONT signals, because it is managed by debugger, however a normal signal sent to a interruptibly sleeping thread wakes up the thread so it will handle the signal when the process leaves the stopped state. PR: 150138 MFC after: 1 week
2010-08-31 07:15:50 +00:00
if (p->p_flag & P_TRACED)
goto out;
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
* Already stopped, don't need to stop again
* (If we did the shell could get confused).
Slight cleanup of some comments/whitespace. Make idle process state more consistant. Add an assert on thread state. Clean up idleproc/mi_switch() interaction. Use a local instead of referencing curthread 7 times in a row (I've been told curthread can be expensive on some architectures) Remove some commented out code. Add a little commented out code (completion coming soon) Reviewed by: jhb@freebsd.org
2002-08-01 18:45:10 +00:00
* Just make sure the signal STOP bit set.
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
*/
s/SGNL/SIG/ s/SNGL/SINGLE/ s/SNGLE/SINGLE/ Fix abbreviation for P_STOPPED_* etc flags, in original code they were inconsistent and difficult to distinguish between them. Approved by: julian (mentor)
2002-09-05 07:30:18 +00:00
p->p_flag |= P_STOPPED_SIG;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete(sigqueue, sig);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
goto out;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* All other kinds of signals:
* If a thread is sleeping interruptibly, simulate a
* wakeup so that when it is continued it will be made
* runnable and can look at the signal. However, don't make
Slight cleanup of some comments/whitespace. Make idle process state more consistant. Add an assert on thread state. Clean up idleproc/mi_switch() interaction. Use a local instead of referencing curthread 7 times in a row (I've been told curthread can be expensive on some architectures) Remove some commented out code. Add a little commented out code (completion coming soon) Reviewed by: jhb@freebsd.org
2002-08-01 18:45:10 +00:00
* the PROCESS runnable, leave it stopped.
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* It may run a bit until it hits a thread_suspend_check().
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_lock(td);
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
if (TD_CAN_ABORT(td))
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = sleepq_abort(td, intrval);
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
else
thread_unlock(td);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
goto out;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Shorten some XXXKSE commentry
2004-03-29 22:46:54 +00:00
* Mutexes are short lived. Threads waiting on them will
* hit thread_suspend_check() soon.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
} else if (p->p_state == PRS_NORMAL) {
if (p->p_flag & P_TRACED || action == SIG_CATCH) {
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
tdsigwakeup(td, sig, action, intrval);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto out;
Revert some suspension/sleep/signal code from KSE-III We need to rethink a bit of this and it doesn't matter if we break the KSE test program for now as long as non-KSE programs act as expected. Submitted by: David Xu <bsddiy@yahoo.com> (this guy's just asking to get hit with a commit bit..)
2002-08-21 20:03:55 +00:00
}
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
MPASS(action == SIG_DFL);
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (prop & SIGPROP_STOP) {
Ignore stop and continue signals sent to an exiting process. Stop signals set p_xstat to the signal that triggered the stop, but p_xstat is also used to hold the exit status of an exiting process. Without this change, a stop signal that arrived after a process was marked P_WEXIT but before it was marked a zombie would overwrite the exit status with the stop signal number. Reviewed by: kib MFC after: 1 week
2012-09-13 15:51:18 +00:00
if (p->p_flag & (P_PPWAIT|P_WEXIT))
In the kernel code, we have the tsleep() call with the PCATCH argument. PCATCH means 'if we get a signal, interrupt me!" and tsleep returns either EINTR or ERESTART depending on the circumstances. ERESTART is "special" because it causes the system call to fail, but right as it returns back to userland it tells the trap handler to move %eip back a bit so that userland will immediately re-run the syscall. This is a syscall restart. It only works for things like read() etc where nothing has changed yet. Note that *userland* is tricked into restarting the syscall by the kernel. The kernel doesn't actually do the restart. It is deadly for things like select, poll, nanosleep etc where it might cause the elapsed time to be reset and start again from scratch. So those syscalls do this to prevent userland rerunning the syscall: if (error == ERESTART) error = EINTR; Fake "signals" like SIGTSTP from ^Z etc do not normally invoke userland signal handlers. But, in -current, the PCATCH *is* being triggered and tsleep is returning ERESTART, and the syscall is aborted even though no userland signal handler was run. That is the fault here. We're triggering the PCATCH in cases that we shouldn't. ie: it is being triggered on *any* signal processing, rather than the case where the signal is posted to userland. --- Peter The work of psignal() is a patchwork of special case required by the process debugging and job-control facilities... --- Kirk McKusick "The design and impelementation of the 4.4BSD Operating system" Page 105 in STABLE source, when psignal is posting a STOP signal to sleeping process and the signal action of the process is SIG_DFL, system will directly change the process state from SSLEEP to SSTOP, and when SIGCONT is posted to the stopped process, if it finds that the process is still on sleep queue, the process state will be restored to SSLEEP, and won't wakeup the process. this commit mimics the behaviour in STABLE source tree. Reviewed by: Jon Mini, Tim Robbins, Peter Wemm Approved by: julian@freebsd.org (mentor)
2002-09-03 12:56:01 +00:00
goto out;
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
p->p_flag |= P_STOPPED_SIG;
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
p->p_xsig = sig;
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
wakeup_swapper = sig_suspend_threads(td, p, 1);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
if (p->p_numthreads == p->p_suspcount) {
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
/*
* only thread sending signal to another
* process can reach here, if thread is sending
* signal to its process, because thread does
* not suspend itself here, p_numthreads
* should never be equal to p_suspcount.
*/
thread_stopped(p);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
sigqueue_delete_proc(p, p->p_xsig);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
} else
PROC_SUNLOCK(p);
In the kernel code, we have the tsleep() call with the PCATCH argument. PCATCH means 'if we get a signal, interrupt me!" and tsleep returns either EINTR or ERESTART depending on the circumstances. ERESTART is "special" because it causes the system call to fail, but right as it returns back to userland it tells the trap handler to move %eip back a bit so that userland will immediately re-run the syscall. This is a syscall restart. It only works for things like read() etc where nothing has changed yet. Note that *userland* is tricked into restarting the syscall by the kernel. The kernel doesn't actually do the restart. It is deadly for things like select, poll, nanosleep etc where it might cause the elapsed time to be reset and start again from scratch. So those syscalls do this to prevent userland rerunning the syscall: if (error == ERESTART) error = EINTR; Fake "signals" like SIGTSTP from ^Z etc do not normally invoke userland signal handlers. But, in -current, the PCATCH *is* being triggered and tsleep is returning ERESTART, and the syscall is aborted even though no userland signal handler was run. That is the fault here. We're triggering the PCATCH in cases that we shouldn't. ie: it is being triggered on *any* signal processing, rather than the case where the signal is posted to userland. --- Peter The work of psignal() is a patchwork of special case required by the process debugging and job-control facilities... --- Kirk McKusick "The design and impelementation of the 4.4BSD Operating system" Page 105 in STABLE source, when psignal is posting a STOP signal to sleeping process and the signal action of the process is SIG_DFL, system will directly change the process state from SSLEEP to SSTOP, and when SIGCONT is posted to the stopped process, if it finds that the process is still on sleep queue, the process state will be restored to SSLEEP, and won't wakeup the process. this commit mimics the behaviour in STABLE source tree. Reviewed by: Jon Mini, Tim Robbins, Peter Wemm Approved by: julian@freebsd.org (mentor)
2002-09-03 12:56:01 +00:00
goto out;
- Relax requirements for p_numthreads, p_threads, p_swtick, and p_nice from requiring the per-process spinlock to only requiring the process lock. - Reflect these changes in the proc.h documentation and consumers throughout the kernel. This is a substantial reduction in locking cost for these fields and was made possible by recent changes to threading support.
2008-03-19 06:19:01 +00:00
}
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
} else {
/* Not in "NORMAL" state. discard the signal. */
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete(sigqueue, sig);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
goto out;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
* The process is not stopped so we need to apply the signal to all the
* running threads.
*/
runfast:
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
tdsigwakeup(td, sig, action, intrval);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
thread_unsuspend(p);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
out:
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
/* If we jump here, proc slock should not be owned. */
PROC_SLOCK_ASSERT(p, MA_NOTOWNED);
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
if (wakeup_swapper)
kick_proc0();
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
return (ret);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
* The force of a signal has been directed against a single
Fixed some style bugs in tdsigwakeup().
2004-05-21 10:02:24 +00:00
* thread. We need to see what we can do about knocking it
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* out of any sleep it may be in etc.
*/
static void
Fix a long standing race between sleep queue and thread suspension code. When a thread A is going to sleep, it calls sleepq_catch_signals() to detect any pending signals or thread suspension request, if nothing happens, it returns without holding process lock or scheduler lock, this opens a race window which allows thread B to come in and do process suspension work, however since A is still at running state, thread B can do nothing to A, thread A continues, and puts itself into actually sleeping state, but B has never seen it, and it sits there forever until B is woken up by other threads sometimes later(this can be very long delay or never happen). Fix this bug by forcing sleepq_catch_signals to return with scheduler lock held. Fix sleepq_abort() by passing it an interrupted code, previously, it worked as wakeup_one(), and the interruption can not be identified correctly by sleep queue code when the sleeping thread is resumed. Let thread_suspend_check() returns EINTR or ERESTART, so sleep queue no longer has to use SIGSTOP as a hack to build a return value. Reviewed by: jhb MFC after: 1 week
2006-02-15 23:52:01 +00:00
tdsigwakeup(struct thread *td, int sig, sig_t action, int intrval)
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
{
struct proc *p = td->td_proc;
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
int prop, wakeup_swapper;
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
- Make sigonstack() a regular function instead of an inline and add a proc lock assertion to it. - SIGPENDING() no longer needs sched_lock, so only grab sched_lock to set the TDF_NEEDSIGCHK and TDF_ASTPENDING flags in signotify(). - Add a proc lock assertion to tdsigwakeup(). - Since we always set TDF_OLDMASK while holding the proc lock, the proc lock is sufficient protection to check its state in postsig() and we only need sched_lock when clearing the actual flag.
2003-04-18 20:59:05 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
prop = sigprop(sig);
Fixed some style bugs in tdsigwakeup().
2004-05-21 10:02:24 +00:00
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
- Relax requirements for p_numthreads, p_threads, p_swtick, and p_nice from requiring the per-process spinlock to only requiring the process lock. - Reflect these changes in the proc.h documentation and consumers throughout the kernel. This is a substantial reduction in locking cost for these fields and was made possible by recent changes to threading support.
2008-03-19 06:19:01 +00:00
thread_lock(td);
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
Try clean up some of the mess that resulted from layers and layers of p4 merges from -current as things started getting different. Corroborated by: Similar patches just mailed by BDE.
2002-07-03 09:15:20 +00:00
* Bring the priority of a thread up if we want it to get
When delivering a signal with default disposition to the thread, tdsigwakeup() increases the priority of the low-priority threads, to give them a chance to be terminated timely. Also, kernel allows user to signal kernel processes. The combined effect is that signalling idle process bump a priority of the selected delivery thread, which starts eating CPU. Check for the delivery thread be an idle thread and do not raise its priority then. The signal delivery to the kernel threads must be opt-in feature. Kernel thread should explicitely declare the ability to handle signals directed to it. E.g., nfsd threads check for signal as an indication of exit request. Most threads do not handle signals at all, and queuing the signal to them causes odd side-effects. Most innocent consequence is the memory leak due to queued ksiginfo, which is never deleted from the sigqueue. Code to prevent even queuing signals to the kernel threads is trivial, but it requires careful examination of each call to kproc/kthread creation to decide should the signalling be allowed. The commit is a stop-gap measure which fixes the immediate case for now. PR: 200493 Reported and tested by: trasz Discussed with: trasz, emaste Sponsored by: The FreeBSD Foundation MFC after: 1 week
2015-05-29 16:26:08 +00:00
* killed in this lifetime. Be careful to avoid bumping the
* priority of the idle thread, since we still allow to signal
* kernel processes.
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (action == SIG_DFL && (prop & SIGPROP_KILL) != 0 &&
When delivering a signal with default disposition to the thread, tdsigwakeup() increases the priority of the low-priority threads, to give them a chance to be terminated timely. Also, kernel allows user to signal kernel processes. The combined effect is that signalling idle process bump a priority of the selected delivery thread, which starts eating CPU. Check for the delivery thread be an idle thread and do not raise its priority then. The signal delivery to the kernel threads must be opt-in feature. Kernel thread should explicitely declare the ability to handle signals directed to it. E.g., nfsd threads check for signal as an indication of exit request. Most threads do not handle signals at all, and queuing the signal to them causes odd side-effects. Most innocent consequence is the memory leak due to queued ksiginfo, which is never deleted from the sigqueue. Code to prevent even queuing signals to the kernel threads is trivial, but it requires careful examination of each call to kproc/kthread creation to decide should the signalling be allowed. The commit is a stop-gap measure which fixes the immediate case for now. PR: 200493 Reported and tested by: trasz Discussed with: trasz, emaste Sponsored by: The FreeBSD Foundation MFC after: 1 week
2015-05-29 16:26:08 +00:00
td->td_priority > PUSER && !TD_IS_IDLETHREAD(td))
- Calling sched_nice() in tdsigwakeup() is no longer required by ULE and actually causes LORs and other panics. Reported by: mlaier Approved by: re
2007-07-19 08:49:16 +00:00
sched_prio(td, PUSER);
In tdsigwakeup(), use TD_ON_SLEEPQ() rather than TD_IS_SLEEPING() to see if a thread is on a sleep queue and should have it's sleep aborted. Reported by: Thierry Herbelot thierry at herbelot dot com
2004-05-20 20:17:28 +00:00
if (TD_ON_SLEEPQ(td)) {
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
/*
* If thread is sleeping uninterruptibly
* we can't interrupt the sleep... the signal will
* be noticed when the process returns through
* trap() or syscall().
*/
Switch the sleep/wakeup and condition variable implementations to use the sleep queue interface: - Sleep queues attempt to merge some of the benefits of both sleep queues and condition variables. Having sleep qeueus in a hash table avoids having to allocate a queue head for each wait channel. Thus, struct cv has shrunk down to just a single char * pointer now. However, the hash table does not hold threads directly, but queue heads. This means that once you have located a queue in the hash bucket, you no longer have to walk the rest of the hash chain looking for threads. Instead, you have a list of all the threads sleeping on that wait channel. - Outside of the sleepq code and the sleep/cv code the kernel no longer differentiates between cv's and sleep/wakeup. For example, calls to abortsleep() and cv_abort() are replaced with a call to sleepq_abort(). Thus, the TDF_CVWAITQ flag is removed. Also, calls to unsleep() and cv_waitq_remove() have been replaced with calls to sleepq_remove(). - The sched_sleep() function no longer accepts a priority argument as sleep's no longer inherently bump the priority. Instead, this is soley a propery of msleep() which explicitly calls sched_prio() before blocking. - The TDF_ONSLEEPQ flag has been dropped as it was never used. The associated TDF_SET_ONSLEEPQ and TDF_CLR_ON_SLEEPQ macros have also been dropped and replaced with a single explicit clearing of td_wchan. TD_SET_ONSLEEPQ() would really have only made sense if it had taken the wait channel and message as arguments anyway. Now that that only happens in one place, a macro would be overkill.
2004-02-27 18:52:44 +00:00
if ((td->td_flags & TDF_SINTR) == 0)
- Relax requirements for p_numthreads, p_threads, p_swtick, and p_nice from requiring the per-process spinlock to only requiring the process lock. - Reflect these changes in the proc.h documentation and consumers throughout the kernel. This is a substantial reduction in locking cost for these fields and was made possible by recent changes to threading support.
2008-03-19 06:19:01 +00:00
goto out;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
* If SIGCONT is default (or ignored) and process is
* asleep, we are finished; the process should not
* be awakened.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if ((prop & SIGPROP_CONT) && action == SIG_DFL) {
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_unlock(td);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete(&p->p_sigqueue, sig);
Try clean up some of the mess that resulted from layers and layers of p4 merges from -current as things started getting different. Corroborated by: Similar patches just mailed by BDE.
2002-07-03 09:15:20 +00:00
/*
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
* It may be on either list in this state.
* Remove from both for now.
Try clean up some of the mess that resulted from layers and layers of p4 merges from -current as things started getting different. Corroborated by: Similar patches just mailed by BDE.
2002-07-03 09:15:20 +00:00
*/
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete(&td->td_sigqueue, sig);
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
return;
Try clean up some of the mess that resulted from layers and layers of p4 merges from -current as things started getting different. Corroborated by: Similar patches just mailed by BDE.
2002-07-03 09:15:20 +00:00
}
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
/*
* Don't awaken a sleeping thread for SIGSTOP if the
* STOP signal is deferred.
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if ((prop & SIGPROP_STOP) != 0 && (td->td_flags & (TDF_SBDRY |
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
TDF_SERESTART | TDF_SEINTR)) == TDF_SBDRY)
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
goto out;
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
/*
* Give low priority threads a better chance to run.
*/
When delivering a signal with default disposition to the thread, tdsigwakeup() increases the priority of the low-priority threads, to give them a chance to be terminated timely. Also, kernel allows user to signal kernel processes. The combined effect is that signalling idle process bump a priority of the selected delivery thread, which starts eating CPU. Check for the delivery thread be an idle thread and do not raise its priority then. The signal delivery to the kernel threads must be opt-in feature. Kernel thread should explicitely declare the ability to handle signals directed to it. E.g., nfsd threads check for signal as an indication of exit request. Most threads do not handle signals at all, and queuing the signal to them causes odd side-effects. Most innocent consequence is the memory leak due to queued ksiginfo, which is never deleted from the sigqueue. Code to prevent even queuing signals to the kernel threads is trivial, but it requires careful examination of each call to kproc/kthread creation to decide should the signalling be allowed. The commit is a stop-gap measure which fixes the immediate case for now. PR: 200493 Reported and tested by: trasz Discussed with: trasz, emaste Sponsored by: The FreeBSD Foundation MFC after: 1 week
2015-05-29 16:26:08 +00:00
if (td->td_priority > PUSER && !TD_IS_IDLETHREAD(td))
Fix a bug relavant to debugging, a masked signal unexpectedly interrupts a sleeping thread when process is being debugged. PR: GNU/77818 Tested by: Sean C. Farley <sean-freebsd at farley org>
2005-06-06 05:13:10 +00:00
sched_prio(td, PUSER);
If a thread that is swapped out is made runnable, then the setrunnable() routine wakes up proc0 so that proc0 can swap the thread back in. Historically, this has been done by waking up proc0 directly from setrunnable() itself via a wakeup(). When waking up a sleeping thread that was swapped out (the usual case when waking proc0 since only sleeping threads are eligible to be swapped out), this resulted in a bit of recursion (e.g. wakeup() -> setrunnable() -> wakeup()). With sleep queues having separate locks in 6.x and later, this caused a spin lock LOR (sleepq lock -> sched_lock/thread lock -> sleepq lock). An attempt was made to fix this in 7.0 by making the proc0 wakeup use the ithread mechanism for doing the wakeup. However, this required grabbing proc0's thread lock to perform the wakeup. If proc0 was asleep elsewhere in the kernel (e.g. waiting for disk I/O), then this degenerated into the same LOR since the thread lock would be some other sleepq lock. Fix this by deferring the wakeup of the swapper until after the sleepq lock held by the upper layer has been locked. The setrunnable() routine now returns a boolean value to indicate whether or not proc0 needs to be woken up. The end result is that consumers of the sleepq API such as *sleep/wakeup, condition variables, sx locks, and lockmgr, have to wakeup proc0 if they get a non-zero return value from sleepq_abort(), sleepq_broadcast(), or sleepq_signal(). Discussed with: jeff Glanced at by: sam Tested by: Jurgen Weber jurgen - ish com au MFC after: 2 weeks
2008-08-05 20:02:31 +00:00
wakeup_swapper = sleepq_abort(td, intrval);
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
PROC_SUNLOCK(p);
if (wakeup_swapper)
kick_proc0();
return;
}
/*
* Other states do nothing with the signal immediately,
* other than kicking ourselves if we are running.
* It will either never be noticed, or noticed very soon.
*/
Fixed some style bugs in tdsigwakeup().
2004-05-21 10:02:24 +00:00
#ifdef SMP
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
if (TD_IS_RUNNING(td) && td != curthread)
forward_signal(td);
Try clean up some of the mess that resulted from layers and layers of p4 merges from -current as things started getting different. Corroborated by: Similar patches just mailed by BDE.
2002-07-03 09:15:20 +00:00
#endif
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
- Relax requirements for p_numthreads, p_threads, p_swtick, and p_nice from requiring the per-process spinlock to only requiring the process lock. - Reflect these changes in the proc.h documentation and consumers throughout the kernel. This is a substantial reduction in locking cost for these fields and was made possible by recent changes to threading support.
2008-03-19 06:19:01 +00:00
out:
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
- Relax requirements for p_numthreads, p_threads, p_swtick, and p_nice from requiring the per-process spinlock to only requiring the process lock. - Reflect these changes in the proc.h documentation and consumers throughout the kernel. This is a substantial reduction in locking cost for these fields and was made possible by recent changes to threading support.
2008-03-19 06:19:01 +00:00
thread_unlock(td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
static int
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
sig_suspend_threads(struct thread *td, struct proc *p, int sending)
{
struct thread *td2;
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
int wakeup_swapper;
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK_ASSERT(p, MA_OWNED);
Let ptracestop() suspend threads sleeping in an SBDRY section. When a thread enters ptracestop(), for example because it had received SIGSTOP from ptrace(PT_ATTACH), it attempts to suspend other threads in the same process. In the case of a thread sleeping interruptibly in an SBDRY section, sig_suspend_threads() must wake the thread and allow it to reach the user-mode boundary. However, sig_suspend_threads() would erroneously avoid waking up such threads, resulting in an apparent hang. Reviewed by: kib Tested by: pho MFC after: 2 weeks Sponsored by: Dell EMC Isilon
2017-05-11 17:03:45 +00:00
MPASS(sending || td == curthread);
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
wakeup_swapper = 0;
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
FOREACH_THREAD_IN_PROC(p, td2) {
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_lock(td2);
- Add a new td flag TDF_NEEDSUSPCHK that is set whenever a thread needs to enter thread_suspend_check(). - Set TDF_ASTPENDING along with TDF_NEEDSUSPCHK so we can move the thread_suspend_check() to ast() rather than userret(). - Check TDF_NEEDSUSPCHK in the sleepq_catch_signals() optimization so that we don't miss a suspend request. If this is set use the expensive signal path. - Set NEEDSUSPCHK when creating a new thread in thr in case the creating thread is due to be suspended as well but has not yet. Reviewed by: davidxu (Authored original patch)
2008-03-21 08:23:25 +00:00
td2->td_flags |= TDF_ASTPENDING | TDF_NEEDSUSPCHK;
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
if ((TD_IS_SLEEPING(td2) || TD_IS_SWAPPED(td2)) &&
Add new msleep(9) flag PBDY that shall be specified together with PCATCH, to indicate that thread shall not be stopped upon receipt of SIGSTOP until it reaches the kernel->usermode boundary. Also change thread_single(SINGLE_NO_EXIT) to only stop threads at the user boundary unconditionally. Tested by: pho Reviewed by: jhb Approved by: re (kensmith)
2009-07-14 22:52:46 +00:00
(td2->td_flags & TDF_SINTR)) {
if (td2->td_flags & TDF_SBDRY) {
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
/*
* Once a thread is asleep with
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
* TDF_SBDRY and without TDF_SERESTART
* or TDF_SEINTR set, it should never
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
* become suspended due to this check.
*/
KASSERT(!TD_IS_SUSPENDED(td2),
("thread with deferred stops suspended"));
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
if (TD_SBDRY_INTR(td2)) {
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
wakeup_swapper |= sleepq_abort(td2,
TD_SBDRY_ERRNO(td2));
schedlock 1/4 Eliminate recursion from most thread_lock consumers. Return from sched_add() without the thread_lock held. This eliminates unnecessary atomics and lock word loads as well as reducing the hold time for scheduler locks. This will eventually allow for lockless remote adds. Discussed with: kib Reviewed by: jhb Tested by: pho Differential Revision: https://reviews.freebsd.org/D22626
2019-12-15 21:11:15 +00:00
continue;
}
} else if (!TD_IS_SUSPENDED(td2))
Add new msleep(9) flag PBDY that shall be specified together with PCATCH, to indicate that thread shall not be stopped upon receipt of SIGSTOP until it reaches the kernel->usermode boundary. Also change thread_single(SINGLE_NO_EXIT) to only stop threads at the user boundary unconditionally. Tested by: pho Reviewed by: jhb Approved by: re (kensmith)
2009-07-14 22:52:46 +00:00
thread_suspend_one(td2);
} else if (!TD_IS_SUSPENDED(td2)) {
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
if (sending || td != td2)
td2->td_flags |= TDF_ASTPENDING;
#ifdef SMP
if (TD_IS_RUNNING(td2) && td2 != td)
forward_signal(td2);
#endif
}
Commit 4/14 of sched_lock decomposition. - Use thread_lock() rather than sched_lock for per-thread scheduling sychronization. - Use the per-process spinlock rather than the sched_lock for per-process scheduling synchronization. - Move some common code into thread_suspend_switch() to handle the mechanics of suspending a thread. The locking here is incredibly convoluted and should be simplified. Tested by: kris, current@ Tested on: i386, amd64, ULE, 4BSD, libthr, libkse, PREEMPTION, etc. Discussed with: kris, attilio, kmacy, jhb, julian, bde (small parts each)
2007-06-04 23:52:24 +00:00
thread_unlock(td2);
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
}
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
return (wakeup_swapper);
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
}
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
/*
* Stop the process for an event deemed interesting to the debugger. If si is
* non-NULL, this is a signal exchange; the new signal requested by the
* debugger will be returned for handling. If si is NULL, this is some other
* type of interesting event. The debugger may request a signal be delivered in
* that case as well, however it will be deferred until it can be handled.
*/
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
int
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
ptracestop(struct thread *td, int sig, ksiginfo_t *si)
Move some tracing related code into its own function as it will be needed for system call related ptrace functionality I plan to commit soon.
2003-09-26 15:09:46 +00:00
{
struct proc *p = td->td_proc;
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
struct thread *td2;
ksiginfo_t ksi;
Move some tracing related code into its own function as it will be needed for system call related ptrace functionality I plan to commit soon.
2003-09-26 15:09:46 +00:00
Assert process lock in ptracestop(), since we're going to rely on it, and later unlock it.
2004-01-29 00:58:21 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Ignore stop and continue signals sent to an exiting process. Stop signals set p_xstat to the signal that triggered the stop, but p_xstat is also used to hold the exit status of an exiting process. Without this change, a stop signal that arrived after a process was marked P_WEXIT but before it was marked a zombie would overwrite the exit status with the stop signal number. Reviewed by: kib MFC after: 1 week
2012-09-13 15:51:18 +00:00
KASSERT(!(p->p_flag & P_WEXIT), ("Stopping exiting process"));
Move some tracing related code into its own function as it will be needed for system call related ptrace functionality I plan to commit soon.
2003-09-26 15:09:46 +00:00
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK,
Rename the 'mtx_object', 'rw_object', and 'sx_object' members of mutexes, rwlocks, and sx locks to 'lock_object'.
2007-03-21 21:20:51 +00:00
&p->p_mtx.lock_object, "Stopping for traced signal");
Move some tracing related code into its own function as it will be needed for system call related ptrace functionality I plan to commit soon.
2003-09-26 15:09:46 +00:00
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
td->td_xsig = sig;
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if (si == NULL || (si->ksi_flags & KSI_PTRACE) == 0) {
td->td_dbgflags |= TDB_XSIG;
CTR4(KTR_PTRACE, "ptracestop: tid %d (pid %d) flags %#x sig %d",
td->td_tid, p->p_pid, td->td_dbgflags, sig);
PROC_SLOCK(p);
while ((p->p_flag & P_TRACED) && (td->td_dbgflags & TDB_XSIG)) {
if (P_KILLED(p)) {
/*
* Ensure that, if we've been PT_KILLed, the
* exit status reflects that. Another thread
* may also be in ptracestop(), having just
* received the SIGKILL, but this thread was
* unsuspended first.
*/
td->td_dbgflags &= ~TDB_XSIG;
td->td_xsig = SIGKILL;
p->p_ptevents = 0;
break;
}
if (p->p_flag & P_SINGLE_EXIT &&
!(td->td_dbgflags & TDB_EXIT)) {
/*
* Ignore ptrace stops except for thread exit
* events when the process exits.
*/
td->td_dbgflags &= ~TDB_XSIG;
PROC_SUNLOCK(p);
return (0);
}
Add ptrace(2) reporting for LWP events. Add two new LWPINFO flags: PL_FLAG_BORN and PL_FLAG_EXITED for reporting thread creation and destruction. Newly created threads will stop to report PL_FLAG_BORN before returning to userland and exiting threads will stop to report PL_FLAG_EXIT before exiting completely. Both of these events are only enabled and reported if PT_LWP_EVENTS is enabled on a process.
2015-12-29 23:25:26 +00:00
/*
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
* Make wait(2) work. Ensure that right after the
* attach, the thread which was decided to become the
* leader of attach gets reported to the waiter.
* Otherwise, just avoid overwriting another thread's
* assignment to p_xthread. If another thread has
* already set p_xthread, the current thread will get
* a chance to report itself upon the next iteration.
Add ptrace(2) reporting for LWP events. Add two new LWPINFO flags: PL_FLAG_BORN and PL_FLAG_EXITED for reporting thread creation and destruction. Newly created threads will stop to report PL_FLAG_BORN before returning to userland and exiting threads will stop to report PL_FLAG_EXIT before exiting completely. Both of these events are only enabled and reported if PT_LWP_EVENTS is enabled on a process.
2015-12-29 23:25:26 +00:00
*/
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if ((td->td_dbgflags & TDB_FSTP) != 0 ||
((p->p_flag2 & P2_PTRACE_FSTP) == 0 &&
p->p_xthread == NULL)) {
p->p_xsig = sig;
p->p_xthread = td;
Do not go into sleep in sleepq_catch_signals() when SIGSTOP from PT_ATTACH was consumed. In particular, do not clear TDP_FSTP in ptracestop() if td_wchan is non-NULL. Leave it to sleepq_catch_signal() to clear and convert zero return code to EINTR. Otherwise, per submitter report, if the PT_ATTACH SIGSTOP was delivered right after the thread was added to the sleepqueue but not yet really sleep, and cursig() caused debugger attach, the thread sleeps instead of returning to the userspace boundary with EINTR. PR: 231445 Reported by: Efi Weiss <valmarelox@gmail.com> Reviewed by: markj Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D20381
2019-05-29 14:05:27 +00:00
/*
* If we are on sleepqueue already,
* let sleepqueue code decide if it
* needs to go sleep after attach.
*/
if (td->td_wchan == NULL)
td->td_dbgflags &= ~TDB_FSTP;
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
p->p_flag2 &= ~P2_PTRACE_FSTP;
p->p_flag |= P_STOPPED_SIG | P_STOPPED_TRACE;
sig_suspend_threads(td, p, 0);
}
if ((td->td_dbgflags & TDB_STOPATFORK) != 0) {
td->td_dbgflags &= ~TDB_STOPATFORK;
}
stopme:
thread_suspend_switch(td, p);
if (p->p_xthread == td)
p->p_xthread = NULL;
if (!(p->p_flag & P_TRACED))
break;
if (td->td_dbgflags & TDB_SUSPEND) {
if (p->p_flag & P_SINGLE_EXIT)
break;
goto stopme;
}
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
}
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
PROC_SUNLOCK(p);
}
When a debugger attaches to the process, SIGSTOP is sent to the target. Due to a way issignal() selects the next signal to deliver and report, if the simultaneous or already pending another signal exists, that signal might be reported by the next waitpid(2) call. This causes minor annoyance for debuggers, which must be prepared to take any signal as the first event, then filter SIGSTOP later. More importantly, for tools like gcore(1), which attach and then detach without processing events, SIGSTOP might leak to be delivered after PT_DETACH. This results in the process being unintentionally stopped after detach, which is fatal for automatic tools. The solution is to force SIGSTOP to be the first signal reported after the attach. Attach code is modified to set P2_PTRACE_FSTP to indicate that the attaching ritual was not yet finished, and issignal() prefers SIGSTOP in that condition. Also, the thread which handles P2_PTRACE_FSTP is made to guarantee to own p_xthread during the first waitpid(2). All that ensures that SIGSTOP is consumed first. Additionally, if P2_PTRACE_FSTP is still set on detach, which means that waitpid(2) was not called at all, SIGSTOP is removed from the queue, ensuring that the process is resumed on detach. In issignal(), when acting on STOPing signals, remove the signal from queue before suspending. Otherwise parallel attach could result in ptracestop() acting on that STOP as if it was the STOP signal from the attach. Then SIGSTOP from attach leaks again. As a minor refactoring, some bits of the common attach code is moved to new helper proc_set_traced(). Reported by: markj Reviewed by: jhb, markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D7256
2016-07-28 08:41:13 +00:00
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
if (si != NULL && sig == td->td_xsig) {
/* Parent wants us to take the original signal unchanged. */
si->ksi_flags |= KSI_HEAD;
if (sigqueue_add(&td->td_sigqueue, sig, si) != 0)
si->ksi_signo = 0;
} else if (td->td_xsig != 0) {
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
/*
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
* If parent wants us to take a new signal, then it will leave
* it in td->td_xsig; otherwise we just look for signals again.
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
*/
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
ksiginfo_init(&ksi);
ksi.ksi_signo = td->td_xsig;
ksi.ksi_flags |= KSI_PTRACE;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
td2 = sigtd(p, td->td_xsig, false);
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
tdsendsignal(p, td2, td->td_xsig, &ksi);
if (td != td2)
return (0);
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
}
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
Add code to support debugging threaded process. 1. Add tm_lwpid into kse_thr_mailbox to indicate which kernel thread current user thread is running on. Add tm_dflags into kse_thr_mailbox, the flags is written by debugger, it tells UTS and kernel what should be done when the process is being debugged, current, there two flags TMDF_SSTEP and TMDF_DONOTRUNUSER. TMDF_SSTEP is used to tell kernel to turn on single stepping, or turn off if it is not set. TMDF_DONOTRUNUSER is used to tell kernel to schedule upcall whenever possible, to UTS, it means do not run the user thread until debugger clears it, this behaviour is necessary because gdb wants to resume only one thread when the thread's pc is at a breakpoint, and thread needs to go forward, in order to avoid other threads sneak pass the breakpoints, it needs to remove breakpoint, only wants one thread to go. Also, add km_lwp to kse_mailbox, the lwp id is copied to kse_thr_mailbox at context switch time when process is not being debugged, so when process is attached, debugger can map kernel thread to user thread. 2. Add p_xthread to proc strcuture and td_xsig to thread structure. p_xthread is used by a thread when it wants to report event to debugger, every thread can set the pointer, especially, when it is used in ptracestop, it is the last thread reporting event will win the race. Every thread has a td_xsig to exchange signal with debugger, thread uses TDF_XSIG flag to indicate it is reporting signal to debugger, if the flag is not cleared, thread will keep retrying until it is cleared by debugger, p_xthread may be used by debugger to indicate CURRENT thread. The p_xstat is still in proc structure to keep wait() to work, in future, we may just use td_xsig. 3. Add TDF_DBSUSPEND flag, the flag is used by debugger to suspend a thread. When process stops, debugger can set the flag for thread, thread will check the flag in thread_suspend_check, enters a loop, unless it is cleared by debugger, process is detached or process is existing. The flag is also checked in ptracestop, so debugger can temporarily suspend a thread even if the thread wants to exchange signal. 4. Current, in ptrace, we always resume all threads, but if a thread has already a TDF_DBSUSPEND flag set by debugger, it won't run. Encouraged by: marcel, julian, deischen
2004-07-13 07:20:10 +00:00
return (td->td_xsig);
Move some tracing related code into its own function as it will be needed for system call related ptrace functionality I plan to commit soon.
2003-09-26 15:09:46 +00:00
}
Return reschedule_signals() to being static again. It was used after sigfastblock_setpend() call in in ast() when current thread fast-blocks signals. Add a flag to sigfastblock_setpend() to request reschedule, and remove the direct use of the function from subr_trap.c Tested by: pho Sponsored by: The FreeBSD Foundation
2020-03-10 20:04:38 +00:00
static void
Trapsignal() and postsig() call kern_sigprocmask() with both process lock and curproc->p_sigacts->ps_mtx. Reschedule_signals may need to have ps_mtx locked to decide and wakeup a thread, causing recursion on the mutex. Inform kern_sigprocmask() and reschedule_signals() about lock state of the ps_mtx by new flag SIGPROCMASK_PS_LOCKED to avoid recursion. Reported and tested by: keramida MFC after: 1 month
2009-10-30 10:10:39 +00:00
reschedule_signals(struct proc *p, sigset_t block, int flags)
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
{
struct sigacts *ps;
struct thread *td;
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
int sig;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
bool fastblk, pslocked;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Assert the state of the process lock and sigact mutex in kern_sigprocmask() and reschedule_signals(). Discussed with: rea Sponsored by: The FreeBSD Foundation MFC after: 1 week
2014-11-28 10:20:00 +00:00
ps = p->p_sigacts;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
pslocked = (flags & SIGPROCMASK_PS_LOCKED) != 0;
mtx_assert(&ps->ps_mtx, pslocked ? MA_OWNED : MA_NOTOWNED);
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
if (SIGISEMPTY(p->p_siglist))
return;
SIGSETAND(block, p->p_siglist);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
fastblk = (flags & SIGPROCMASK_FASTBLK) != 0;
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
while ((sig = sig_ffs(&block)) != 0) {
SIGDELSET(block, sig);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
td = sigtd(p, sig, fastblk);
/*
* If sigtd() selected us despite sigfastblock is
* blocking, do not activate AST or wake us, to avoid
* loop in AST handler.
*/
if (fastblk && td == curthread)
continue;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
signotify(td);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
if (!pslocked)
Trapsignal() and postsig() call kern_sigprocmask() with both process lock and curproc->p_sigacts->ps_mtx. Reschedule_signals may need to have ps_mtx locked to decide and wakeup a thread, causing recursion on the mutex. Inform kern_sigprocmask() and reschedule_signals() about lock state of the ps_mtx by new flag SIGPROCMASK_PS_LOCKED to avoid recursion. Reported and tested by: keramida MFC after: 1 month
2009-10-30 10:10:39 +00:00
mtx_lock(&ps->ps_mtx);
Do not wake up sleeping thread in reschedule_signals() if the signal is blocked. The spurious wakeup might result in spurious EINTR. The reschedule_signals() function is called when the calling thread has the signal mask changed. For each newly blocked signal, we try to find a thread which might have the signal not blocked. If no such thread exists, sigtd() returns random thread, which must not be waken up. I decided that re-checking, as suggested by PR submitter, is more reasonable change than to change sigtd() interface, due to other uses of sigtd(). signotify() already performs this check. Submitted by: Duane <parakleta@darkreality.org> PR: 219228 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-05-12 15:34:59 +00:00
if (p->p_flag & P_TRACED ||
(SIGISMEMBER(ps->ps_sigcatch, sig) &&
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
!SIGISMEMBER(td->td_sigmask, sig))) {
In kern_sigtimedwait(), move initialization code out of process lock, instead of using SIGISMEMBER to test every interesting signal, just unmask the signal set and let cursig() return one, get the signal after it returns, call reschedule_signal() after signals are blocked again. In kern_sigprocmask(), don't call reschedule_signal() when it is unnecessary. In reschedule_signal(), replace SIGISEMPTY() + SIGISMEMBER() with sig_ffs(), rename variable 'i' to sig.
2010-10-14 08:01:33 +00:00
tdsigwakeup(td, sig, SIG_CATCH,
(SIGISMEMBER(ps->ps_sigintr, sig) ? EINTR :
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
ERESTART));
}
if (!pslocked)
Trapsignal() and postsig() call kern_sigprocmask() with both process lock and curproc->p_sigacts->ps_mtx. Reschedule_signals may need to have ps_mtx locked to decide and wakeup a thread, causing recursion on the mutex. Inform kern_sigprocmask() and reschedule_signals() about lock state of the ps_mtx by new flag SIGPROCMASK_PS_LOCKED to avoid recursion. Reported and tested by: keramida MFC after: 1 month
2009-10-30 10:10:39 +00:00
mtx_unlock(&ps->ps_mtx);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
}
}
void
tdsigcleanup(struct thread *td)
{
struct proc *p;
sigset_t unblocked;
p = td->td_proc;
PROC_LOCK_ASSERT(p, MA_OWNED);
sigqueue_flush(&td->td_sigqueue);
if (p->p_numthreads == 1)
return;
/*
* Since we cannot handle signals, notify signal post code
* about this by filling the sigmask.
*
* Also, if needed, wake up thread(s) that do not block the
* same signals as the exiting thread, since the thread might
* have been selected for delivery and woken up.
*/
SIGFILLSET(unblocked);
SIGSETNAND(unblocked, td->td_sigmask);
SIGFILLSET(td->td_sigmask);
Trapsignal() and postsig() call kern_sigprocmask() with both process lock and curproc->p_sigacts->ps_mtx. Reschedule_signals may need to have ps_mtx locked to decide and wakeup a thread, causing recursion on the mutex. Inform kern_sigprocmask() and reschedule_signals() about lock state of the ps_mtx by new flag SIGPROCMASK_PS_LOCKED to avoid recursion. Reported and tested by: keramida MFC after: 1 month
2009-10-30 10:10:39 +00:00
reschedule_signals(p, unblocked, 0);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
}
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
static int
sigdeferstop_curr_flags(int cflags)
{
MPASS((cflags & (TDF_SEINTR | TDF_SERESTART)) == 0 ||
(cflags & TDF_SBDRY) != 0);
return (cflags & (TDF_SBDRY | TDF_SEINTR | TDF_SERESTART));
}
Further refine the handling of stop signals in the NFS client. The changes in r246417 were incomplete as they did not add explicit calls to sigdeferstop() around all the places that previously passed SBDRY to _sleep(). In addition, nfs_getcacheblk() could trigger a write RPC from getblk() resulting in sigdeferstop() recursing. Rather than manually deferring stop signals in specific places, change the VFS_*() and VOP_*() methods to defer stop signals for filesystems which request this behavior via a new VFCF_SBDRY flag. Note that this has to be a VFC flag rather than a MNTK flag so that it works properly with VFS_MOUNT() when the mount is not yet fully constructed. For now, only the NFS clients are set this new flag in VFS_SET(). A few other related changes: - Add an assertion to ensure that TDF_SBDRY doesn't leak to userland. - When a lookup request uses VOP_READLINK() to follow a symlink, mark the request as being on behalf of the thread performing the lookup (cnp_thread) rather than using a NULL thread pointer. This causes NFS to properly handle signals during this VOP on an interruptible mount. PR: kern/176179 Reported by: Russell Cattelan (sigdeferstop() recursion) Reviewed by: kib MFC after: 1 month
2013-02-21 19:02:50 +00:00
/*
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
* Defer the delivery of SIGSTOP for the current thread, according to
* the requested mode. Returns previous flags, which must be restored
* by sigallowstop().
*
* TDF_SBDRY, TDF_SEINTR, and TDF_SERESTART flags are only set and
* cleared by the current thread, which allow the lock-less read-only
* accesses below.
Further refine the handling of stop signals in the NFS client. The changes in r246417 were incomplete as they did not add explicit calls to sigdeferstop() around all the places that previously passed SBDRY to _sleep(). In addition, nfs_getcacheblk() could trigger a write RPC from getblk() resulting in sigdeferstop() recursing. Rather than manually deferring stop signals in specific places, change the VFS_*() and VOP_*() methods to defer stop signals for filesystems which request this behavior via a new VFCF_SBDRY flag. Note that this has to be a VFC flag rather than a MNTK flag so that it works properly with VFS_MOUNT() when the mount is not yet fully constructed. For now, only the NFS clients are set this new flag in VFS_SET(). A few other related changes: - Add an assertion to ensure that TDF_SBDRY doesn't leak to userland. - When a lookup request uses VOP_READLINK() to follow a symlink, mark the request as being on behalf of the thread performing the lookup (cnp_thread) rather than using a NULL thread pointer. This causes NFS to properly handle signals during this VOP on an interruptible mount. PR: kern/176179 Reported by: Russell Cattelan (sigdeferstop() recursion) Reviewed by: kib MFC after: 1 month
2013-02-21 19:02:50 +00:00
*/
int
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
sigdeferstop_impl(int mode)
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
{
Further refine the handling of stop signals in the NFS client. The changes in r246417 were incomplete as they did not add explicit calls to sigdeferstop() around all the places that previously passed SBDRY to _sleep(). In addition, nfs_getcacheblk() could trigger a write RPC from getblk() resulting in sigdeferstop() recursing. Rather than manually deferring stop signals in specific places, change the VFS_*() and VOP_*() methods to defer stop signals for filesystems which request this behavior via a new VFCF_SBDRY flag. Note that this has to be a VFC flag rather than a MNTK flag so that it works properly with VFS_MOUNT() when the mount is not yet fully constructed. For now, only the NFS clients are set this new flag in VFS_SET(). A few other related changes: - Add an assertion to ensure that TDF_SBDRY doesn't leak to userland. - When a lookup request uses VOP_READLINK() to follow a symlink, mark the request as being on behalf of the thread performing the lookup (cnp_thread) rather than using a NULL thread pointer. This causes NFS to properly handle signals during this VOP on an interruptible mount. PR: kern/176179 Reported by: Russell Cattelan (sigdeferstop() recursion) Reviewed by: kib MFC after: 1 month
2013-02-21 19:02:50 +00:00
struct thread *td;
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
int cflags, nflags;
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
Further refine the handling of stop signals in the NFS client. The changes in r246417 were incomplete as they did not add explicit calls to sigdeferstop() around all the places that previously passed SBDRY to _sleep(). In addition, nfs_getcacheblk() could trigger a write RPC from getblk() resulting in sigdeferstop() recursing. Rather than manually deferring stop signals in specific places, change the VFS_*() and VOP_*() methods to defer stop signals for filesystems which request this behavior via a new VFCF_SBDRY flag. Note that this has to be a VFC flag rather than a MNTK flag so that it works properly with VFS_MOUNT() when the mount is not yet fully constructed. For now, only the NFS clients are set this new flag in VFS_SET(). A few other related changes: - Add an assertion to ensure that TDF_SBDRY doesn't leak to userland. - When a lookup request uses VOP_READLINK() to follow a symlink, mark the request as being on behalf of the thread performing the lookup (cnp_thread) rather than using a NULL thread pointer. This causes NFS to properly handle signals during this VOP on an interruptible mount. PR: kern/176179 Reported by: Russell Cattelan (sigdeferstop() recursion) Reviewed by: kib MFC after: 1 month
2013-02-21 19:02:50 +00:00
td = curthread;
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
cflags = sigdeferstop_curr_flags(td->td_flags);
switch (mode) {
case SIGDEFERSTOP_NOP:
nflags = cflags;
break;
case SIGDEFERSTOP_OFF:
nflags = 0;
break;
case SIGDEFERSTOP_SILENT:
nflags = (cflags | TDF_SBDRY) & ~(TDF_SEINTR | TDF_SERESTART);
break;
case SIGDEFERSTOP_EINTR:
nflags = (cflags | TDF_SBDRY | TDF_SEINTR) & ~TDF_SERESTART;
break;
case SIGDEFERSTOP_ERESTART:
nflags = (cflags | TDF_SBDRY | TDF_SERESTART) & ~TDF_SEINTR;
break;
default:
panic("sigdeferstop: invalid mode %x", mode);
break;
}
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
if (cflags == nflags)
return (SIGDEFERSTOP_VAL_NCHG);
thread_lock(td);
td->td_flags = (td->td_flags & ~cflags) | nflags;
thread_unlock(td);
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
return (cflags);
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
}
/*
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
* Restores the STOP handling mode, typically permitting the delivery
* of SIGSTOP for the current thread. This does not immediately
* suspend if a stop was posted. Instead, the thread will suspend
* either via ast() or a subsequent interruptible sleep.
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
*/
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
void
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
sigallowstop_impl(int prev)
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
{
Further refine the handling of stop signals in the NFS client. The changes in r246417 were incomplete as they did not add explicit calls to sigdeferstop() around all the places that previously passed SBDRY to _sleep(). In addition, nfs_getcacheblk() could trigger a write RPC from getblk() resulting in sigdeferstop() recursing. Rather than manually deferring stop signals in specific places, change the VFS_*() and VOP_*() methods to defer stop signals for filesystems which request this behavior via a new VFCF_SBDRY flag. Note that this has to be a VFC flag rather than a MNTK flag so that it works properly with VFS_MOUNT() when the mount is not yet fully constructed. For now, only the NFS clients are set this new flag in VFS_SET(). A few other related changes: - Add an assertion to ensure that TDF_SBDRY doesn't leak to userland. - When a lookup request uses VOP_READLINK() to follow a symlink, mark the request as being on behalf of the thread performing the lookup (cnp_thread) rather than using a NULL thread pointer. This causes NFS to properly handle signals during this VOP on an interruptible mount. PR: kern/176179 Reported by: Russell Cattelan (sigdeferstop() recursion) Reviewed by: kib MFC after: 1 month
2013-02-21 19:02:50 +00:00
struct thread *td;
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
int cflags;
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
KASSERT(prev != SIGDEFERSTOP_VAL_NCHG, ("failed sigallowstop"));
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
KASSERT((prev & ~(TDF_SBDRY | TDF_SEINTR | TDF_SERESTART)) == 0,
("sigallowstop: incorrect previous mode %x", prev));
Further refine the handling of stop signals in the NFS client. The changes in r246417 were incomplete as they did not add explicit calls to sigdeferstop() around all the places that previously passed SBDRY to _sleep(). In addition, nfs_getcacheblk() could trigger a write RPC from getblk() resulting in sigdeferstop() recursing. Rather than manually deferring stop signals in specific places, change the VFS_*() and VOP_*() methods to defer stop signals for filesystems which request this behavior via a new VFCF_SBDRY flag. Note that this has to be a VFC flag rather than a MNTK flag so that it works properly with VFS_MOUNT() when the mount is not yet fully constructed. For now, only the NFS clients are set this new flag in VFS_SET(). A few other related changes: - Add an assertion to ensure that TDF_SBDRY doesn't leak to userland. - When a lookup request uses VOP_READLINK() to follow a symlink, mark the request as being on behalf of the thread performing the lookup (cnp_thread) rather than using a NULL thread pointer. This causes NFS to properly handle signals during this VOP on an interruptible mount. PR: kern/176179 Reported by: Russell Cattelan (sigdeferstop() recursion) Reviewed by: kib MFC after: 1 month
2013-02-21 19:02:50 +00:00
td = curthread;
Rewrite sigdeferstop(9) and sigallowstop(9) into more flexible framework allowing to set the suspension policy for the dynamic block. Extend the currently possible policies of stopping on interruptible sleeps and ignoring such sleeps by two more: do not suspend at interruptible sleeps, but interrupt them with either EINTR or ERESTART. Reviewed by: jilles Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-06-26 20:07:24 +00:00
cflags = sigdeferstop_curr_flags(td->td_flags);
if (cflags != prev) {
thread_lock(td);
td->td_flags = (td->td_flags & ~cflags) | prev;
thread_unlock(td);
}
Rework the handling of stop signals in the NFS client. The changes in 195702, 195703, and 195821 prevented a thread from suspending while holding locks inside of NFS by forcing the thread to fail sleeps with EINTR or ERESTART but defer the thread suspension to the user boundary. However, this had the effect that stopping a process during an NFS request could abort the request and trigger EINTR errors that were visible to userland processes (previously the thread would have suspended and completed the request once it was resumed). This change instead effectively masks stop signals while in the NFS client. It uses the existing TDF_SBDRY flag to effect this since SIGSTOP cannot be masked directly. Also, instead of setting PBDRY on individual sleeps, the NFS client now sets the TDF_SBDRY flag around each NFS request and stop signals are masked for all sleeps during that region (the previous change missed sleeps in lockmgr locks). The end result is that stop signals sent to threads performing an NFS request are completely ignored until after the NFS request has finished processing and the thread prepares to return to userland. This restores the behavior of stop signals being transparent to userland processes while still preventing threads from suspending while holding NFS locks. Reviewed by: kib MFC after: 1 month
2013-02-06 17:06:51 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* If the current process has received a signal (should be caught or cause
* termination, should interrupt current syscall), return the signal number.
* Stop signals with default action are processed immediately, then cleared;
* they aren't returned. This is checked after each entry to the system for
* a syscall or trap (though this can usually be done without calling issignal
CURSIG() is not a macro so rename it cursig(). Obtained from: KSE tree
2002-05-29 23:44:32 +00:00
* by checking the pending signal masks in cursig.) The normal call
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* sequence is
*
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
* while (sig = cursig(curthread))
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
* postsig(sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Make issignal() private to kern_sig.c since it is only called from cursig() and cursig() is now a function rather than a macro.
2003-05-05 21:16:28 +00:00
static int
Partially revert r195702. Deferring stops is now implemented via a set of calls to toggle TDF_SBDRY rather than passing PBDRY to individual sleep calls. - Remove the stop_allowed parameters from cursig() and issignal(). issignal() checks TDF_SBDRY directly. - Remove the PBDRY and SLEEPQ_STOP_ON_BDRY flags.
2013-03-18 17:23:58 +00:00
issignal(struct thread *td)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
struct proc *p;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
struct sigacts *ps;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
struct sigqueue *queue;
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
sigset_t sigpending;
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
ksiginfo_t ksi;
Retire procfs-based process debugging. Modern debuggers and process tracers use ptrace() rather than procfs for debugging. ptrace() has a supserset of functionality available via procfs and new debugging features are only added to ptrace(). While the two debugging services share some fields in struct proc, they each use dedicated fields and separate code. This results in extra complexity to support a feature that hasn't been enabled in the default install for several years. PR: 244939 (exp-run) Reviewed by: kib, mjg (earlier version) Relnotes: yes Differential Revision: https://reviews.freebsd.org/D23837
2020-04-01 19:22:09 +00:00
int prop, sig;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Part 1 of KSE-III The ability to schedule multiple threads per process (one one cpu) by making ALL system calls optionally asynchronous. to come: ia64 and power-pc patches, patches for gdb, test program (in tools) Reviewed by: Almost everyone who counts (at various times, peter, jhb, matt, alfred, mini, bernd, and a cast of thousands) NOTE: this is still Beta code, and contains lots of debugging stuff. expect slight instability in signals..
2002-06-29 17:26:22 +00:00
p = td->td_proc;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
ps = p->p_sigacts;
mtx_assert(&ps->ps_mtx, MA_OWNED);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
for (;;) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigpending = td->td_sigqueue.sq_signals;
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
SIGSETOR(sigpending, p->p_sigqueue.sq_signals);
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
SIGSETNAND(sigpending, td->td_sigmask);
Complete r302215. TDF_SBDRY | TDF_SERESTART and TDF_SBDRY | TDF_SEINTR flags values, unlike TDF_SBDRY, must be treated almost as if TDF_SBDRY is not set for STOP signal delivery. The only difference is that sig_suspend_threads() should abort the sleep instead of doing immediate suspension. Reported by: ngie Sponsored by: The FreeBSD Foundation MFC after: 12 days Approved by: re (gjb)
2016-06-28 16:41:50 +00:00
if ((p->p_flag & P_PPWAIT) != 0 || (td->td_flags &
(TDF_SBDRY | TDF_SERESTART | TDF_SEINTR)) == TDF_SBDRY)
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
SIG_STOPSIGMASK(sigpending);
if (SIGISEMPTY(sigpending)) /* no signal to send */
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (0);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
/*
* Do fast sigblock if requested by usermode. Since
* we do know that there was a signal pending at this
* point, set the FAST_SIGBLOCK_PEND as indicator for
* usermode to perform a dummy call to
* FAST_SIGBLOCK_UNBLOCK, which causes immediate
* delivery of postponed pending signal.
*/
if ((td->td_pflags & TDP_SIGFASTBLOCK) != 0) {
if (td->td_sigblock_val != 0)
SIGSETNAND(sigpending, fastblock_mask);
if (SIGISEMPTY(sigpending)) {
td->td_pflags |= TDP_SIGFASTPENDING;
return (0);
}
}
When a debugger attaches to the process, SIGSTOP is sent to the target. Due to a way issignal() selects the next signal to deliver and report, if the simultaneous or already pending another signal exists, that signal might be reported by the next waitpid(2) call. This causes minor annoyance for debuggers, which must be prepared to take any signal as the first event, then filter SIGSTOP later. More importantly, for tools like gcore(1), which attach and then detach without processing events, SIGSTOP might leak to be delivered after PT_DETACH. This results in the process being unintentionally stopped after detach, which is fatal for automatic tools. The solution is to force SIGSTOP to be the first signal reported after the attach. Attach code is modified to set P2_PTRACE_FSTP to indicate that the attaching ritual was not yet finished, and issignal() prefers SIGSTOP in that condition. Also, the thread which handles P2_PTRACE_FSTP is made to guarantee to own p_xthread during the first waitpid(2). All that ensures that SIGSTOP is consumed first. Additionally, if P2_PTRACE_FSTP is still set on detach, which means that waitpid(2) was not called at all, SIGSTOP is removed from the queue, ensuring that the process is resumed on detach. In issignal(), when acting on STOPing signals, remove the signal from queue before suspending. Otherwise parallel attach could result in ptracestop() acting on that STOP as if it was the STOP signal from the attach. Then SIGSTOP from attach leaks again. As a minor refactoring, some bits of the common attach code is moved to new helper proc_set_traced(). Reported by: markj Reviewed by: jhb, markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D7256
2016-07-28 08:41:13 +00:00
if ((p->p_flag & (P_TRACED | P_PPTRACE)) == P_TRACED &&
(p->p_flag2 & P2_PTRACE_FSTP) != 0 &&
SIGISMEMBER(sigpending, SIGSTOP)) {
/*
* If debugger just attached, always consume
* SIGSTOP from ptrace(PT_ATTACH) first, to
* execute the debugger attach ritual in
* order.
*/
sig = SIGSTOP;
td->td_dbgflags |= TDB_FSTP;
} else {
sig = sig_ffs(&sigpending);
}
Changes to allow event-based process monitoring and control.
1997-12-06 04:11:14 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* We should see pending but ignored signals
* only if P_TRACED was on when they were posted.
*/
Retire procfs-based process debugging. Modern debuggers and process tracers use ptrace() rather than procfs for debugging. ptrace() has a supserset of functionality available via procfs and new debugging features are only added to ptrace(). While the two debugging services share some fields in struct proc, they each use dedicated fields and separate code. This results in extra complexity to support a feature that hasn't been enabled in the default install for several years. PR: 244939 (exp-run) Reviewed by: kib, mjg (earlier version) Relnotes: yes Differential Revision: https://reviews.freebsd.org/D23837
2020-04-01 19:22:09 +00:00
if (SIGISMEMBER(ps->ps_sigignore, sig) &&
(p->p_flag & P_TRACED) == 0) {
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
sigqueue_delete(&td->td_sigqueue, sig);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
sigqueue_delete(&p->p_sigqueue, sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
continue;
}
When a debugger attaches to the process, SIGSTOP is sent to the target. Due to a way issignal() selects the next signal to deliver and report, if the simultaneous or already pending another signal exists, that signal might be reported by the next waitpid(2) call. This causes minor annoyance for debuggers, which must be prepared to take any signal as the first event, then filter SIGSTOP later. More importantly, for tools like gcore(1), which attach and then detach without processing events, SIGSTOP might leak to be delivered after PT_DETACH. This results in the process being unintentionally stopped after detach, which is fatal for automatic tools. The solution is to force SIGSTOP to be the first signal reported after the attach. Attach code is modified to set P2_PTRACE_FSTP to indicate that the attaching ritual was not yet finished, and issignal() prefers SIGSTOP in that condition. Also, the thread which handles P2_PTRACE_FSTP is made to guarantee to own p_xthread during the first waitpid(2). All that ensures that SIGSTOP is consumed first. Additionally, if P2_PTRACE_FSTP is still set on detach, which means that waitpid(2) was not called at all, SIGSTOP is removed from the queue, ensuring that the process is resumed on detach. In issignal(), when acting on STOPing signals, remove the signal from queue before suspending. Otherwise parallel attach could result in ptracestop() acting on that STOP as if it was the STOP signal from the attach. Then SIGSTOP from attach leaks again. As a minor refactoring, some bits of the common attach code is moved to new helper proc_set_traced(). Reported by: markj Reviewed by: jhb, markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D7256
2016-07-28 08:41:13 +00:00
if ((p->p_flag & (P_TRACED | P_PPTRACE)) == P_TRACED) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Remove trace_req(). Reviewed by: alfred, jhb, peter
2002-05-09 04:13:41 +00:00
* If traced, always stop.
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
* Remove old signal from queue before the stop.
* XXX shrug off debugger, it causes siginfo to
* be thrown away.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
queue = &td->td_sigqueue;
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
ksiginfo_init(&ksi);
if (sigqueue_get(queue, sig, &ksi) == 0) {
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
queue = &p->p_sigqueue;
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
sigqueue_get(queue, sig, &ksi);
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
}
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
td->td_si = ksi.ksi_info;
When traced process is about to receive the signal, the process is stopped and debugger may modify or drop the signal. After the changes to keep process-targeted signals on the process sigqueue, another thread may note the old signal on the queue and act before the thread removes changed or dropped signal from the process queue. Since process is traced, it usually gets stopped. Or, if the same signal is delivered while process was stopped, the thread may erronously remove it, intending to remove the original signal. Remove the signal from the queue before notifying the debugger. Restore the siginfo to the head of sigqueue when signal is allowed to be delivered to the debugee, using newly introduced KSI_HEAD ksiginfo_t flag. This preserves required order of delivery. Always restore the unchanged signal on the curthread sigqueue, not to the process queue, since the thread is about to get it anyway, because sigmask cannot be changed. Handle failure of reinserting the siginfo into the queue by falling back to sq_kill method, calling sigqueue_add with NULL ksi. If debugger changed the signal to be delivered, use sigqueue_add() with NULL ksi instead of only setting sq_signals bit. Reported by: Gardner Bell <gbell72 rogers com> Analyzed and first version of fix by: Tijl Coosemans <tijl coosemans org> PR: 142757 Reviewed by: davidxu MFC after: 2 weeks
2010-01-20 11:58:04 +00:00
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
sig = ptracestop(td, sig, &ksi);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_lock(&ps->ps_mtx);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Don't report stale signal information for non-signal events in ptrace_lwpinfo. Once a signal's siginfo was copied to 'td_si' as part of the signal exchange in issignal(), it was never cleared. This caused future thread events that are reported as SIGTRAP events without signal information to report the stale siginfo in 'td_si'. For example, if a debugger created a new process and used SIGSTOP to stop it after PT_ATTACH, future system call entry / exit events would set PL_FLAG_SI with the SIGSTOP siginfo in pl_siginfo. This broke 'catch syscall' in current versions of gdb as it assumed PL_FLAG_SI with SIGTRAP indicates a breakpoint or single step trap. Reviewed by: kib MFC after: 1 week Differential Revision: https://reviews.freebsd.org/D18487
2018-12-10 19:39:24 +00:00
td->td_si.si_signo = 0;
Defer ptracestop() signals that cannot be delivered immediately When a thread is stopped in ptracestop(), the ptrace(2) user may request a signal be delivered upon resumption of the thread. Heretofore, those signals were discarded unless ptracestop()'s caller was issignal(). Fix this by modifying ptracestop() to queue up signals requested by the ptrace user that will be delivered when possible. Take special care when the signal is SIGKILL (usually generated from a PT_KILL request); no new stop events should be triggered after a PT_KILL. Add a number of tests for the new functionality. Several tests were authored by jhb. PR: 212607 Reviewed by: kib Approved by: kib (mentor) MFC after: 2 weeks Sponsored by: Dell EMC In collaboration with: jhb Differential Revision: https://reviews.freebsd.org/D9260
2017-02-20 15:53:16 +00:00
/*
Re-evaluate thread' signal mask after ptracestop(). The stop drops process lock, which allows the signal mask to be changed and our selected signal might become blocked, i.e. should be returned to the process queue instead of delivery. Also, for the existing check of the process no longer having an attached debugger, we should not loose the signal, but requeue it. Reported and tested by: bdrewery Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:21:51 +00:00
* Keep looking if the debugger discarded or
* replaced the signal.
Fix long standing bug that prevents the PT_CONTINUE, PT_KILL and PT_DETACH ptrace(2) requests from functioning as advertised in the manual page. As described in kern/35175, the PT_DETACH request will, under certain circumstances, pass an unwanted signal on to the traced process upan detaching from it. The PT_CONTINUE request will sometimes fail if you make it pass a signal that has "properties" that differ from the properties of the signal that origionally caused the traced process to be stopped. Since PT_KILL is nothing than PT_CONTINUE with SIGKILL, it is broken too. In the PT_KILL case, this leads to an unkillable process. PR: 44011 Submitted by: Mark Kettenis <kettenis@chello.nl> Approved by: re(jhb)
2003-05-16 01:34:23 +00:00
*/
Re-evaluate thread' signal mask after ptracestop(). The stop drops process lock, which allows the signal mask to be changed and our selected signal might become blocked, i.e. should be returned to the process queue instead of delivery. Also, for the existing check of the process no longer having an attached debugger, we should not loose the signal, but requeue it. Reported and tested by: bdrewery Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:21:51 +00:00
if (sig == 0)
Fix long standing bug that prevents the PT_CONTINUE, PT_KILL and PT_DETACH ptrace(2) requests from functioning as advertised in the manual page. As described in kern/35175, the PT_DETACH request will, under certain circumstances, pass an unwanted signal on to the traced process upan detaching from it. The PT_CONTINUE request will sometimes fail if you make it pass a signal that has "properties" that differ from the properties of the signal that origionally caused the traced process to be stopped. Since PT_KILL is nothing than PT_CONTINUE with SIGKILL, it is broken too. In the PT_KILL case, this leads to an unkillable process. PR: 44011 Submitted by: Mark Kettenis <kettenis@chello.nl> Approved by: re(jhb)
2003-05-16 01:34:23 +00:00
continue;
Re-evaluate thread' signal mask after ptracestop(). The stop drops process lock, which allows the signal mask to be changed and our selected signal might become blocked, i.e. should be returned to the process queue instead of delivery. Also, for the existing check of the process no longer having an attached debugger, we should not loose the signal, but requeue it. Reported and tested by: bdrewery Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:21:51 +00:00
/*
* If the signal became masked, re-queue it.
*/
if (SIGISMEMBER(td->td_sigmask, sig)) {
ksi.ksi_flags |= KSI_HEAD;
sigqueue_add(&p->p_sigqueue, sig, &ksi);
continue;
}
/*
* If the traced bit got turned off, requeue
* the signal and go back up to the top to
* rescan signals. This ensures that p_sig*
* and p_sigact are consistent.
*/
if ((p->p_flag & P_TRACED) == 0) {
ksi.ksi_flags |= KSI_HEAD;
sigqueue_add(queue, sig, &ksi);
continue;
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fix long standing bug that prevents the PT_CONTINUE, PT_KILL and PT_DETACH ptrace(2) requests from functioning as advertised in the manual page. As described in kern/35175, the PT_DETACH request will, under certain circumstances, pass an unwanted signal on to the traced process upan detaching from it. The PT_CONTINUE request will sometimes fail if you make it pass a signal that has "properties" that differ from the properties of the signal that origionally caused the traced process to be stopped. Since PT_KILL is nothing than PT_CONTINUE with SIGKILL, it is broken too. In the PT_KILL case, this leads to an unkillable process. PR: 44011 Submitted by: Mark Kettenis <kettenis@chello.nl> Approved by: re(jhb)
2003-05-16 01:34:23 +00:00
prop = sigprop(sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Decide whether the signal should be returned.
* Return the signal's number, or fall through
* to clear it from the pending mask.
*/
Don't cast a pointer to (intptr_t) and then on to (int) when we cannot be sure that (int) is large enough. Instead cast only to (intptr_t) and cast the switch/case values to (intptr_t) as well.
2002-12-17 19:13:03 +00:00
switch ((intptr_t)p->p_sigacts->ps_sigact[_SIG_IDX(sig)]) {
case (intptr_t)SIG_DFL:
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Don't take default actions on system processes.
*/
if (p->p_pid <= 1) {
#ifdef DIAGNOSTIC
/*
* Are you sure you want to ignore SIGSEGV
* in init? XXX
*/
Cosmetics. related to getting prototypes into view.
1994-10-10 01:00:49 +00:00
printf("Process (pid %lu) got signal %d\n",
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
(u_long)p->p_pid, sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#endif
break; /* == ignore */
}
/*
don't stop in issignal() if P_SINGLE_EXIT is set Suppose a traced process is stopped in ptracestop() due to receipt of a SIGSTOP signal, and is awaiting orders from the tracing process on how to handle the signal. Before sending any such orders, the tracing process exits. This should kill the traced process. But suppose a second thread handles the SIGKILL and proceeds to exit1(), calling thread_single(). The first thread will now awaken and will have a chance to check once more if it should go to sleep due to the SIGSTOP. It must not sleep after P_SINGLE_EXIT has been set; this would prevent the SIGKILL from taking effect, leaving a stopped orphan behind after the tracing process dies. Also add new tests for this condition. Reviewed by: kib MFC after: 2 weeks Sponsored by: Dell EMC Differential Revision: https://reviews.freebsd.org/D9890
2017-03-07 13:41:01 +00:00
* If there is a pending stop signal to process with
* default action, stop here, then clear the signal.
* Traced or exiting processes should ignore stops.
* Additionally, a member of an orphaned process group
* should ignore tty stops.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if (prop & SIGPROP_STOP) {
don't stop in issignal() if P_SINGLE_EXIT is set Suppose a traced process is stopped in ptracestop() due to receipt of a SIGSTOP signal, and is awaiting orders from the tracing process on how to handle the signal. Before sending any such orders, the tracing process exits. This should kill the traced process. But suppose a second thread handles the SIGKILL and proceeds to exit1(), calling thread_single(). The first thread will now awaken and will have a chance to check once more if it should go to sleep due to the SIGSTOP. It must not sleep after P_SINGLE_EXIT has been set; this would prevent the SIGKILL from taking effect, leaving a stopped orphan behind after the tracing process dies. Also add new tests for this condition. Reviewed by: kib MFC after: 2 weeks Sponsored by: Dell EMC Differential Revision: https://reviews.freebsd.org/D9890
2017-03-07 13:41:01 +00:00
if (p->p_flag &
(P_TRACED | P_WEXIT | P_SINGLE_EXIT) ||
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
(p->p_pgrp->pg_jobc == 0 &&
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
prop & SIGPROP_TTYSTOP))
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break; /* == ignore */
Provide helper macros to detect 'non-silent SBDRY' state and to calculate appropriate return value for stops. Simplify the code by using them. Fix typo in sig_suspend_threads(). The thread which sleep must be aborted is td2. (*) In issignal(), when handling stopping signal for thread in TD_SBDRY_INTR state, do not stop, this is wrong and fires assert. This is yet another place where execution should be forced out of SBDRY-protected region. For such case, return -1 from issignal() and translate it to corresponding error code in sleepq_catch_signals(). Assert that other consumers of cursig() are not affected by the new return value. (*) Micro-optimize, mostly VFS and VOP methods, by avoiding calling the functions when SIGDEFERSTOP_NOP non-change is requested. (**) Reported and tested by: pho (*) Requested by: bde (**) Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb)
2016-07-03 18:19:48 +00:00
if (TD_SBDRY_INTR(td)) {
KASSERT((td->td_flags & TDF_SBDRY) != 0,
("lost TDF_SBDRY"));
return (-1);
}
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
WITNESS_WARN(WARN_GIANTOK | WARN_SLEEPOK,
Rename the 'mtx_object', 'rw_object', and 'sx_object' members of mutexes, rwlocks, and sx locks to 'lock_object'.
2007-03-21 21:20:51 +00:00
&p->p_mtx.lock_object, "Catching SIGSTOP");
When a debugger attaches to the process, SIGSTOP is sent to the target. Due to a way issignal() selects the next signal to deliver and report, if the simultaneous or already pending another signal exists, that signal might be reported by the next waitpid(2) call. This causes minor annoyance for debuggers, which must be prepared to take any signal as the first event, then filter SIGSTOP later. More importantly, for tools like gcore(1), which attach and then detach without processing events, SIGSTOP might leak to be delivered after PT_DETACH. This results in the process being unintentionally stopped after detach, which is fatal for automatic tools. The solution is to force SIGSTOP to be the first signal reported after the attach. Attach code is modified to set P2_PTRACE_FSTP to indicate that the attaching ritual was not yet finished, and issignal() prefers SIGSTOP in that condition. Also, the thread which handles P2_PTRACE_FSTP is made to guarantee to own p_xthread during the first waitpid(2). All that ensures that SIGSTOP is consumed first. Additionally, if P2_PTRACE_FSTP is still set on detach, which means that waitpid(2) was not called at all, SIGSTOP is removed from the queue, ensuring that the process is resumed on detach. In issignal(), when acting on STOPing signals, remove the signal from queue before suspending. Otherwise parallel attach could result in ptracestop() acting on that STOP as if it was the STOP signal from the attach. Then SIGSTOP from attach leaks again. As a minor refactoring, some bits of the common attach code is moved to new helper proc_set_traced(). Reported by: markj Reviewed by: jhb, markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D7256
2016-07-28 08:41:13 +00:00
sigqueue_delete(&td->td_sigqueue, sig);
sigqueue_delete(&p->p_sigqueue, sig);
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
p->p_flag |= P_STOPPED_SIG;
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
p->p_xsig = sig;
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
In order to speed up process suspension on MP machine, send IPI to remote CPU. While here, abstract thread suspension code into a function called sig_suspend_threads, the function is called when a process received a STOP signal.
2006-02-13 03:16:55 +00:00
sig_suspend_threads(td, p, 0);
Add facility to stop all userspace processes. The supposed use of the feature is to quisce the system before suspend. Stop is implemented by reusing the thread_single(9) with the special mode SINGLE_ALLPROC. SINGLE_ALLPROC differs from the existing single-threading modes by allowing (requiring) caller to operate on other process. Interruptible sleeps for !TDF_SBDRY threads are suspended like SIGSTOP does it, instead of aborting the sleep, like SINGLE_NO_EXIT, to avoid spurious EINTRs on resume. Provide debugging sysctl debug.stop_all_proc, which causes total stop and suspends syncer, while waiting for variable reset for resume. It is used for debugging; should be removed after the real use of the interface is added. In collaboration with: pho Discussed with: avg Sponsored by: The FreeBSD Foundation MFC after: 2 weeks
2014-12-13 16:18:29 +00:00
thread_suspend_switch(td, p);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_lock(&ps->ps_mtx);
When a debugger attaches to the process, SIGSTOP is sent to the target. Due to a way issignal() selects the next signal to deliver and report, if the simultaneous or already pending another signal exists, that signal might be reported by the next waitpid(2) call. This causes minor annoyance for debuggers, which must be prepared to take any signal as the first event, then filter SIGSTOP later. More importantly, for tools like gcore(1), which attach and then detach without processing events, SIGSTOP might leak to be delivered after PT_DETACH. This results in the process being unintentionally stopped after detach, which is fatal for automatic tools. The solution is to force SIGSTOP to be the first signal reported after the attach. Attach code is modified to set P2_PTRACE_FSTP to indicate that the attaching ritual was not yet finished, and issignal() prefers SIGSTOP in that condition. Also, the thread which handles P2_PTRACE_FSTP is made to guarantee to own p_xthread during the first waitpid(2). All that ensures that SIGSTOP is consumed first. Additionally, if P2_PTRACE_FSTP is still set on detach, which means that waitpid(2) was not called at all, SIGSTOP is removed from the queue, ensuring that the process is resumed on detach. In issignal(), when acting on STOPing signals, remove the signal from queue before suspending. Otherwise parallel attach could result in ptracestop() acting on that STOP as if it was the STOP signal from the attach. Then SIGSTOP from attach leaks again. As a minor refactoring, some bits of the common attach code is moved to new helper proc_set_traced(). Reported by: markj Reviewed by: jhb, markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D7256
2016-07-28 08:41:13 +00:00
goto next;
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
} else if (prop & SIGPROP_IGNORE) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Except for SIGCONT, shouldn't get here.
* Default action is to ignore; drop it.
*/
break; /* == ignore */
} else
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*NOTREACHED*/
Don't cast a pointer to (intptr_t) and then on to (int) when we cannot be sure that (int) is large enough. Instead cast only to (intptr_t) and cast the switch/case values to (intptr_t) as well.
2002-12-17 19:13:03 +00:00
case (intptr_t)SIG_IGN:
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Masking above should prevent us ever trying
* to take action on an ignored signal other
* than SIGCONT, unless process is traced.
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if ((prop & SIGPROP_CONT) == 0 &&
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
(p->p_flag & P_TRACED) == 0)
printf("issignal\n");
break; /* == ignore */
default:
/*
* This signal has an action, let
* postsig() process it.
*/
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
return (sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
sigqueue_delete(&td->td_sigqueue, sig); /* take the signal! */
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
sigqueue_delete(&p->p_sigqueue, sig);
When a debugger attaches to the process, SIGSTOP is sent to the target. Due to a way issignal() selects the next signal to deliver and report, if the simultaneous or already pending another signal exists, that signal might be reported by the next waitpid(2) call. This causes minor annoyance for debuggers, which must be prepared to take any signal as the first event, then filter SIGSTOP later. More importantly, for tools like gcore(1), which attach and then detach without processing events, SIGSTOP might leak to be delivered after PT_DETACH. This results in the process being unintentionally stopped after detach, which is fatal for automatic tools. The solution is to force SIGSTOP to be the first signal reported after the attach. Attach code is modified to set P2_PTRACE_FSTP to indicate that the attaching ritual was not yet finished, and issignal() prefers SIGSTOP in that condition. Also, the thread which handles P2_PTRACE_FSTP is made to guarantee to own p_xthread during the first waitpid(2). All that ensures that SIGSTOP is consumed first. Additionally, if P2_PTRACE_FSTP is still set on detach, which means that waitpid(2) was not called at all, SIGSTOP is removed from the queue, ensuring that the process is resumed on detach. In issignal(), when acting on STOPing signals, remove the signal from queue before suspending. Otherwise parallel attach could result in ptracestop() acting on that STOP as if it was the STOP signal from the attach. Then SIGSTOP from attach leaks again. As a minor refactoring, some bits of the common attach code is moved to new helper proc_set_traced(). Reported by: markj Reviewed by: jhb, markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D7256
2016-07-28 08:41:13 +00:00
next:;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/* NOTREACHED */
}
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
void
thread_stopped(struct proc *p)
{
int n;
PROC_LOCK_ASSERT(p, MA_OWNED);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK_ASSERT(p, MA_OWNED);
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
n = p->p_suspcount;
Code cleanup, simply compare with curproc.
2006-02-23 05:50:55 +00:00
if (p == curproc)
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
n++;
if ((p->p_flag & P_STOPPED_SIG) && (n == p->p_numthreads)) {
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SUNLOCK(p);
Oops, forgot to update this file. Fix a race condition between kern_wait() and thread_stopped(). Problem is in kern_wait(), parent process steps through children list, once a child process is skipped, and later even if the child is stopped, parent process still sleeps in msleep(), the race happens if parent masked SIGCHLD. Submitted by : Peter Edwards peadar.edwards at gmail dot com MFC after : 4 days
2005-04-19 08:11:28 +00:00
p->p_flag &= ~P_WAITED;
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
PROC_LOCK(p->p_pptr);
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
childproc_stopped(p, (p->p_flag & P_TRACED) ?
CLD_TRAPPED : CLD_STOPPED);
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
PROC_UNLOCK(p->p_pptr);
Revert rev 184216 and 184199, due to the way the thread_lock works, it may cause a lockup. Noticed by: peter, jhb
2008-11-05 03:01:23 +00:00
PROC_SLOCK(p);
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
}
}
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Take the action for the specified signal
* from the current set of pending signals.
*/
In r198506, kern_sigsuspend() started doing cursig/postsig loop to make sure that a signal was delivered to the thread before returning from syscall. Signal delivery puts new return frame on the user stack, and modifies trap frame to enter signal handler. As a consequence, syscall return code sets EINTR as error return for signal frame, instead of the syscall return. Also, for ia64, due to different registers layout for those two kind of frames, usermode sigsegfaulted when returned from signal handler. Use newly-introduced cpu_set_syscall_retval(9) to set syscall result, and return EJUSTRETURN from kern_sigsuspend() to prevent syscall return code from modifying this frame [1]. Another issue is that pending SIGCONT might be cancelled by SIGSTOP, causing postsig() not to deliver any catched signal [2]. Modify postsig() to return 1 if signal was posted, and 0 otherwise, and use this in the kern_sigsuspend loop. Proposed by: marcel [1] Noted by: davidxu [2] Reviewed by: marcel, davidxu MFC after: 1 month
2009-11-10 11:46:53 +00:00
int
Style. Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:11:29 +00:00
postsig(int sig)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Style. Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:11:29 +00:00
struct thread *td;
struct proc *p;
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
struct sigacts *ps;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
sig_t action;
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksiginfo_t ksi;
Fix SA_SIGINFO | SA_RESETHAND handling. The sysent' sv_sendsig() method needs pre-reset state of the ps_siginfo to correctly construct signal frame. Move sigdflt() call after the sv_sendsig() invocation in postsig(). Simultaneously extract common code from trapsignal() and postsig() into new helper postsig_done(). Submitted by: rea MFC after: 1 week
2014-11-26 14:09:04 +00:00
sigset_t returnmask;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
KASSERT(sig != 0, ("postsig"));
Split DIAGNOSTIC -> DIAGNOSTIC, INVARIANTS, and INVARIANT_SUPPORT as discussed on -hackers. Introduce 'KASSERT(assertion, ("panic message", args))' for simple check + panic. Reviewed by: msmith
1999-01-08 17:31:30 +00:00
Style. Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:11:29 +00:00
td = curthread;
p = td->td_proc;
- Change CURSIG() and postsig() to require that the proc lock is held rather than grabbing it and releasing it themselves. This allows callers of these functions to get the lock to close race conditions. - Grab Giant around ktrace in postsig. - Count the switches performed on SIGSTOP's as involuntary context switches in the resource usage stats. Reported by: tegge (signal race), bde (missing csw stats)
2001-06-22 23:02:37 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
ps = p->p_sigacts;
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_assert(&ps->ps_mtx, MA_OWNED);
1. Make ksiginfo_alloc and ksiginfo_free public. 2. Introduce flags KSI_EXT and KSI_INS. The flag KSI_EXT allows a ksiginfo to be managed by outside code, the KSI_INS indicates sigqueue_add should directly insert passed ksiginfo into queue other than copy it.
2005-10-23 04:12:26 +00:00
ksiginfo_init(&ksi);
Currently, when signal is delivered to the process and there is a thread not blocking the signal, signal is placed on the thread sigqueue. If the selected thread is in kernel executing thr_exit() or sigprocmask() syscalls, then signal might be not delivered to usermode for arbitrary amount of time, and for exiting thread it is lost. Put process-directed signals to the process queue unconditionally, selecting the thread to deliver the signal only by the thread returning to usermode, since only then the thread can handle delivery of signal reliably. For exiting thread or thread that has blocked some signals, check whether the newly blocked signal is queued for the process, and try to find a thread to wakeup for delivery, in reschedule_signal(). For exiting thread, assume that all signals are blocked. Change cursig() and postsig() to look both into the thread and process signal queues. When there is a signal that thread returning to usermode could consume, TDF_NEEDSIGCHK flag is not neccessary set now. Do unlocked read of p_siglist and p_pendingcnt to check for queued signals. Note that thread that has a signal unblocked might get spurious wakeup and EINTR from the interruptible system call now, due to the possibility of being selected by reschedule_signals(), while other thread returned to usermode earlier and removed the signal from process queue. This should not cause compliance issues, since the thread has not blocked a signal and thus should be ready to receive it anyway. Reported by: Justin Teller <justin.teller gmail com> Reviewed by: davidxu, jilles MFC after: 1 month
2009-10-11 16:49:30 +00:00
if (sigqueue_get(&td->td_sigqueue, sig, &ksi) == 0 &&
sigqueue_get(&p->p_sigqueue, sig, &ksi) == 0)
In r198506, kern_sigsuspend() started doing cursig/postsig loop to make sure that a signal was delivered to the thread before returning from syscall. Signal delivery puts new return frame on the user stack, and modifies trap frame to enter signal handler. As a consequence, syscall return code sets EINTR as error return for signal frame, instead of the syscall return. Also, for ia64, due to different registers layout for those two kind of frames, usermode sigsegfaulted when returned from signal handler. Use newly-introduced cpu_set_syscall_retval(9) to set syscall result, and return EJUSTRETURN from kern_sigsuspend() to prevent syscall return code from modifying this frame [1]. Another issue is that pending SIGCONT might be cancelled by SIGSTOP, causing postsig() not to deliver any catched signal [2]. Modify postsig() to return 1 if signal was posted, and 0 otherwise, and use this in the kern_sigsuspend loop. Proposed by: marcel [1] Noted by: davidxu [2] Reviewed by: marcel, davidxu MFC after: 1 month
2009-11-10 11:46:53 +00:00
return (0);
1. Change prototype of trapsignal and sendsig to use ksiginfo_t *, most changes in MD code are trivial, before this change, trapsignal and sendsig use discrete parameters, now they uses member fields of ksiginfo_t structure. For sendsig, this change allows us to pass POSIX realtime signal value to user code. 2. Remove cpu_thread_siginfo, it is no longer needed because we now always generate ksiginfo_t data and feed it to libpthread. 3. Add p_sigqueue to proc structure to hold shared signals which were blocked by all threads in the proc. 4. Add td_sigqueue to thread structure to hold all signals delivered to thread. 5. i386 and amd64 now return POSIX standard si_code, other arches will be fixed. 6. In this sigqueue implementation, pending signal set is kept as before, an extra siginfo list holds additional siginfo_t data for signals. kernel code uses psignal() still behavior as before, it won't be failed even under memory pressure, only exception is when deleting a signal, we should call sigqueue_delete to remove signal from sigqueue but not SIGDELSET. Current there is no kernel code will deliver a signal with additional data, so kernel should be as stable as before, a ksiginfo can carry more information, for example, allow signal to be delivered but throw away siginfo data if memory is not enough. SIGKILL and SIGSTOP have fast path in sigqueue_add, because they can not be caught or masked. The sigqueue() syscall allows user code to queue a signal to target process, if resource is unavailable, EAGAIN will be returned as specification said. Just before thread exits, signal queue memory will be freed by sigqueue_flush. Current, all signals are allowed to be queued, not only realtime signals. Earlier patch reviewed by: jhb, deischen Tested on: i386, amd64
2005-10-14 12:43:47 +00:00
ksi.ksi_signo = sig;
Let itimer store itimerspec instead of itimerval, so I don't have to convert to or from timeval frequently. Introduce function itimer_accept() to ack a timer signal in signal acceptance code, this allows us to return more fresh overrun counter than at signal generating time. while POSIX says: "the value returned by timer_getoverrun() shall apply to the most recent expiration signal delivery or acceptance for the timer,.." I prefer returning it at acceptance time. Introduce SIGEV_THREAD_ID notification mode, it is used by thread libary to request kernel to deliver signal to a specified thread, and in turn, the thread library may use the mechanism to implement SIGEV_THREAD which is required by POSIX. Timer signal is managed by timer code, so it can not fail even if signal queue is full filled by sigqueue syscall.
2005-10-30 02:56:08 +00:00
if (ksi.ksi_code == SI_TIMER)
itimer_accept(p, ksi.ksi_timerid, &ksi);
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
action = ps->ps_sigact[_SIG_IDX(sig)];
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#ifdef KTRACE
- trapsignal() no longer needs to acquire Giant for ktrpsig(). - Catch up to new ktrace API.
2002-06-07 05:43:02 +00:00
if (KTRPOINT(td, KTR_PSIG))
- Add a td_pflags field to struct thread for private flags accessed only by curthread. Unlike td_flags, this field does not need any locking. - Replace the td_inktr and td_inktrace variables with equivalent private thread flags. - Move TDF_OLDMASK over to the private flags field so it no longer requires sched_lock.
2003-06-09 17:38:32 +00:00
ktrpsig(sig, action, td->td_pflags & TDP_OLDMASK ?
ktrace: Log the code for all signals (PSIG events). The code provides information on how the signal was generated. Formerly, the code was only logged for traps, much like only signal handlers for traps received a meaningful si_code before FreeBSD 7.0. In rare cases, no information is available and 0 is still logged. MFC after: 1 week
2011-04-17 14:38:11 +00:00
&td->td_oldsigmask : &td->td_sigmask, ksi.ksi_code);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#endif
Changes to allow event-based process monitoring and control.
1997-12-06 04:11:14 +00:00
Make KSE a kernel option, turned on by default in all GENERIC kernel configs except sun4v (which doesn't process signals properly with KSE). Reviewed by: davidxu@
2006-10-26 21:42:22 +00:00
if (action == SIG_DFL) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Default action, where the default is to kill
* the process. (Other cases were ignored above.)
*/
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_unlock(&ps->ps_mtx);
Add support for capturing 'struct ptrace_lwpinfo' for signals resulting in a process dumping core in the corefile. Also extend procstat to view select members of 'struct ptrace_lwpinfo' from the contents of the note. Sponsored by: Dell EMC Isilon
2017-03-30 18:21:36 +00:00
proc_td_siginfo_capture(td, &ksi.ksi_info);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
sigexit(td, sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* NOTREACHED */
} else {
/*
* If we get here, the signal must be caught.
*/
Improve assertion that an ignored or blocked signal is not delivered. Split two conditions into separate asserts. Print additional details, like the signal number and action value. Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-10-16 20:15:19 +00:00
KASSERT(action != SIG_IGN, ("postsig action %p", action));
KASSERT(!SIGISMEMBER(td->td_sigmask, sig),
("postsig action: blocked sig %d", sig));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Set the new mask value and also defer further
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
* occurrences of this signal.
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
* Special case: user has done a sigsuspend. Here the
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* current mask is not of interest, but rather the
Add a per-signal flag to mark handlers registered with osigaction, so we can provide the correct context to each signal handler. Fix broken sigsuspend(): don't use p_oldsigmask as a flag, use SAS_OLDMASK as we did before the linuxthreads support merge (submitted by bde). Move ps_sigstk from to p_sigacts to the main proc structure since signal stack should not be shared among threads. Move SAS_OLDMASK and SAS_ALTSTACK flags from sigacts::ps_flags to proc::p_flag. Move PS_NOCLDSTOP and PS_NOCLDWAIT flags from proc::p_flag to procsig::ps_flag. Reviewed by: marcel, jdp, bde
1999-10-11 20:33:17 +00:00
* mask from before the sigsuspend is what we want
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
* restored after the signal processing is completed.
*/
- Add a td_pflags field to struct thread for private flags accessed only by curthread. Unlike td_flags, this field does not need any locking. - Replace the td_inktr and td_inktrace variables with equivalent private thread flags. - Move TDF_OLDMASK over to the private flags field so it no longer requires sched_lock.
2003-06-09 17:38:32 +00:00
if (td->td_pflags & TDP_OLDMASK) {
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
returnmask = td->td_oldsigmask;
- Add a td_pflags field to struct thread for private flags accessed only by curthread. Unlike td_flags, this field does not need any locking. - Replace the td_inktr and td_inktrace variables with equivalent private thread flags. - Move TDF_OLDMASK over to the private flags field so it no longer requires sched_lock.
2003-06-09 17:38:32 +00:00
td->td_pflags &= ~TDP_OLDMASK;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
} else
- Move p->p_sigmask to td->td_sigmask. Signal masks will be per thread with a follow on commit to kern_sig.c - signotify() now operates on a thread since unmasked pending signals are stored in the thread. - PS_NEEDSIGCHK moves to TDF_NEEDSIGCHK.
2003-03-31 22:49:17 +00:00
returnmask = td->td_sigmask;
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
Remove even more unneeded variable assignments. kern_time.c: - Unused variable `p'. kern_thr.c: - Variable `error' is always caught immediately, so no reason to initialize it. There is no way that error != 0 at the end of create_thread(). kern_sig.c: - Unused variable `code'. kern_synch.c: - `rval' is always assigned in all different cases. kern_rwlock.c: - `v' is always overwritten with RW_UNLOCKED further on. kern_malloc.c: - `size' is always initialized with the proper value before being used. kern_exit.c: - `error' is always caught and returned immediately. abort2() never returns a non-zero value. kern_exec.c: - `len' is always assigned inside the if-statement right below it. tty_info.c: - `td' is always overwritten by FOREACH_THREAD_IN_PROC(). Found by: LLVM's scan-build
2009-02-26 15:51:54 +00:00
if (p->p_sig == sig) {
Reviewed by: Luoqi Chen, Jordan Hubbard Submitted by: "Richard Seaman, Jr." <lists@tar.com> Obtained from: linux :-) Code to allow Linux Threads to run under FreeBSD. By default not enabled This code is dependent on the conditional COMPAT_LINUX_THREADS (suggested by Garret) This is not yet a 'real' option but will be within some number of hours.
1998-12-19 02:55:34 +00:00
p->p_sig = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Make KSE a kernel option, turned on by default in all GENERIC kernel configs except sun4v (which doesn't process signals properly with KSE). Reviewed by: davidxu@
2006-10-26 21:42:22 +00:00
(*p->p_sysent->sv_sendsig)(action, &ksi, &returnmask);
Fix SA_SIGINFO | SA_RESETHAND handling. The sysent' sv_sendsig() method needs pre-reset state of the ps_siginfo to correctly construct signal frame. Move sigdflt() call after the sv_sendsig() invocation in postsig(). Simultaneously extract common code from trapsignal() and postsig() into new helper postsig_done(). Submitted by: rea MFC after: 1 week
2014-11-26 14:09:04 +00:00
postsig_done(sig, td, ps);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
In r198506, kern_sigsuspend() started doing cursig/postsig loop to make sure that a signal was delivered to the thread before returning from syscall. Signal delivery puts new return frame on the user stack, and modifies trap frame to enter signal handler. As a consequence, syscall return code sets EINTR as error return for signal frame, instead of the syscall return. Also, for ia64, due to different registers layout for those two kind of frames, usermode sigsegfaulted when returned from signal handler. Use newly-introduced cpu_set_syscall_retval(9) to set syscall result, and return EJUSTRETURN from kern_sigsuspend() to prevent syscall return code from modifying this frame [1]. Another issue is that pending SIGCONT might be cancelled by SIGSTOP, causing postsig() not to deliver any catched signal [2]. Modify postsig() to return 1 if signal was posted, and 0 otherwise, and use this in the kern_sigsuspend loop. Proposed by: marcel [1] Noted by: davidxu [2] Reviewed by: marcel, davidxu MFC after: 1 month
2009-11-10 11:46:53 +00:00
return (1);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Refactor sleepq_catch_signals(). - Extract suspension check into sig_ast_checksusp() helper. - Extract signal check and calculation of the interruption errno into sig_ast_needsigchk() helper. The helpers are moved to kern_sig.c which is the proper place for signal-related code. Improve control flow in sleepq_catch_signals(), to handle ret == 0 (can sleep) and ret != 0 (interrupted) only once, by separating checking code into sleepq_check_ast_sq_locked(), which return value is interpreted at single location. Reviewed by: markj Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D26628
2020-10-04 16:30:05 +00:00
int
sig_ast_checksusp(struct thread *td)
{
struct proc *p;
int ret;
p = td->td_proc;
PROC_LOCK_ASSERT(p, MA_OWNED);
if ((td->td_flags & TDF_NEEDSUSPCHK) == 0)
return (0);
ret = thread_suspend_check(1);
MPASS(ret == 0 || ret == EINTR || ret == ERESTART);
return (ret);
}
int
sig_ast_needsigchk(struct thread *td)
{
struct proc *p;
struct sigacts *ps;
int ret, sig;
p = td->td_proc;
PROC_LOCK_ASSERT(p, MA_OWNED);
if ((td->td_flags & TDF_NEEDSIGCHK) == 0)
return (0);
ps = p->p_sigacts;
mtx_lock(&ps->ps_mtx);
sig = cursig(td);
if (sig == -1) {
mtx_unlock(&ps->ps_mtx);
KASSERT((td->td_flags & TDF_SBDRY) != 0, ("lost TDF_SBDRY"));
KASSERT(TD_SBDRY_INTR(td),
("lost TDF_SERESTART of TDF_SEINTR"));
KASSERT((td->td_flags & (TDF_SEINTR | TDF_SERESTART)) !=
(TDF_SEINTR | TDF_SERESTART),
("both TDF_SEINTR and TDF_SERESTART"));
ret = TD_SBDRY_ERRNO(td);
} else if (sig != 0) {
ret = SIGISMEMBER(ps->ps_sigintr, sig) ? EINTR : ERESTART;
mtx_unlock(&ps->ps_mtx);
} else {
mtx_unlock(&ps->ps_mtx);
ret = 0;
}
/*
* Do not go into sleep if this thread was the ptrace(2)
* attach leader. cursig() consumed SIGSTOP from PT_ATTACH,
* but we usually act on the signal by interrupting sleep, and
* should do that here as well.
*/
if ((td->td_dbgflags & TDB_FSTP) != 0) {
if (ret == 0)
ret = EINTR;
td->td_dbgflags &= ~TDB_FSTP;
}
return (ret);
}
Add sig_intr(9). It gives the answer would the thread sleep according to current state of signals and suspensions. Of course the answer is racy and allows for false-negatives (no sleep when signal is delivered after process lock is dropped). Also the answer might change due to signal rescheduling among threads in multi-threaded process. Still it is the best approximation I can provide, to answering the question was the thread interrupted. Reviewed by: markj Tested by: pho, rmacklem Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D26628
2020-10-04 16:33:42 +00:00
int
sig_intr(void)
{
struct thread *td;
struct proc *p;
int ret;
td = curthread;
sig_intr(9): return early if AST is not scheduled. Check td_flags for relevant AST requests lock-less. This opens the race slightly wider where sig_intr() returns false negative, but might be it is worth it. Requested by: mjg Sponsored by: The FreeBSD Foundation MFC after: 1 week
2020-10-08 22:34:34 +00:00
if ((td->td_flags & (TDF_NEEDSIGCHK | TDF_NEEDSUSPCHK)) == 0)
return (0);
Add sig_intr(9). It gives the answer would the thread sleep according to current state of signals and suspensions. Of course the answer is racy and allows for false-negatives (no sleep when signal is delivered after process lock is dropped). Also the answer might change due to signal rescheduling among threads in multi-threaded process. Still it is the best approximation I can provide, to answering the question was the thread interrupted. Reviewed by: markj Tested by: pho, rmacklem Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D26628
2020-10-04 16:33:42 +00:00
p = td->td_proc;
PROC_LOCK(p);
ret = sig_ast_checksusp(td);
if (ret == 0)
ret = sig_ast_needsigchk(td);
PROC_UNLOCK(p);
return (ret);
}
Swap in WKILLED processes. Swapped-out process that is WKILLED must be swapped in as soon as possible. The reason is that such process can be killed by OOM and its pages can be only freed if the process exits. To exit, the kernel stack of the process must be mapped. When allocating pages for the stack of the WKILLED process on swap in, use VM_ALLOC_SYSTEM requests to increase the chance of the allocation to succeed. Add counter of the swapped out processes to avoid unneeded iteration over the allprocs list when there is no work to do, reducing the allproc_lock ownership. Reviewed by: alc, markj (previous version) Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D16489
2018-08-04 20:45:43 +00:00
void
proc_wkilled(struct proc *p)
{
PROC_LOCK_ASSERT(p, MA_OWNED);
if ((p->p_flag & P_WKILLED) == 0) {
p->p_flag |= P_WKILLED;
/*
* Notify swapper that there is a process to swap in.
* The notification is racy, at worst it would take 10
* seconds for the swapper process to notice.
*/
if ((p->p_flag & (P_INMEM | P_SWAPPINGIN)) == 0)
wakeup(&proc0);
}
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Kill the current process for stated reason.
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
void
sys/kern: quiet -Wwrite-strings Quiet a variety of Wwrite-strings warnings in sys/kern at low-impact sites. This patch avoids addressing certain others which would need to plumb const through structure definitions. Reviewed by: kib, markj Differential Revision: https://reviews.freebsd.org/D23798
2020-02-23 03:32:16 +00:00
killproc(struct proc *p, const char *why)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
- Remove unneeded include of sys/ipl.h. - Require the proc lock be held for killproc() to allow for the vmdaemon to kill a process when memory is exhausted while holding the lock of the process to kill.
2001-05-15 23:13:58 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
More style fixes.
2012-11-27 10:15:58 +00:00
CTR3(KTR_PROC, "killproc: proc %p (pid %d, %s)", p, p->p_pid,
p->p_comm);
Make kernel print jail ID when logging a process exit Kernel now includes jail ID when logging a process exit. jid is 0 for unjailed processes. Submitted by: Marie Helene Kvello-Aune <freebsd@mhka.no> Relnotes: yes Sponsored by: Modirum MDPay Differential Revision: https://reviews.freebsd.org/D18618
2018-12-29 14:48:51 +00:00
log(LOG_ERR, "pid %d (%s), jid %d, uid %d, was killed: %s\n",
Simplify jail ID printing on process exit As suggested by kib@, we don't need to check p_ucred, because that's only NULL during process creation, and cr_prison is never NULL.
2018-12-29 21:36:02 +00:00
p->p_pid, p->p_comm, p->p_ucred->cr_prison->pr_id,
Remove unneeded NULL check for td_ucred td_ucred is always set, so we don't need the ternary expression to check for it.
2019-01-04 21:12:17 +00:00
p->p_ucred->cr_uid, why);
Swap in WKILLED processes. Swapped-out process that is WKILLED must be swapped in as soon as possible. The reason is that such process can be killed by OOM and its pages can be only freed if the process exits. To exit, the kernel stack of the process must be mapped. When allocating pages for the stack of the WKILLED process on swap in, use VM_ALLOC_SYSTEM requests to increase the chance of the allocation to succeed. Add counter of the swapped out processes to avoid unneeded iteration over the allprocs list when there is no work to do, reducing the allproc_lock ownership. Reviewed by: alc, markj (previous version) Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Differential revision: https://reviews.freebsd.org/D16489
2018-08-04 20:45:43 +00:00
proc_wkilled(p);
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
kern_psignal(p, SIGKILL);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* Force the current process to exit with the specified signal, dumping core
* if appropriate. We bypass the normal tests for masked and caught signals,
* allowing unrecoverable failures to terminate the process without changing
* signal state. Mark the accounting record with the signal termination.
* If dumping core, save the signal number for the debugger. Calls exit and
* does not return.
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
void
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
sigexit(struct thread *td, int sig)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct proc *p = td->td_proc;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
p->p_acflag |= AXSIG;
Suspend all other threads in the process while generating a core dump. The main reason for doing this is that the ELF dump handler expects the thread list to be fixed while the dump header is generated, so an upcall that occurs at the wrong time can lead to buffer overruns and other Bad Things. Another solution would be to grab sched_lock in the ELF dump handler, but we might as well single-thread, since the process is about to die. Furthermore, I think this should ensure that the register sets in the core file are sequentially consistent.
2005-04-10 02:31:24 +00:00
/*
* We must be single-threading to generate a core dump. This
* ensures that the registers in the core file are up-to-date.
* Also, the ELF dump handler assumes that the thread list doesn't
* change out from under it.
*
* XXX If another thread attempts to single-thread before us
* (e.g. via fork()), we won't get a dump at all.
*/
Merge from CheriBSD: Rename sigprop-table constants to SIGPROP_ from SA_ to reduce the impression of a namespace collision. Submitted by: rwatson Reviewed by: jhb, kib (slightly different versions) Obtained from: CheriBSD (814ec5771cb1cb53deba317c561de62a91ae7684) Sponsored by: DARPA, AFRL Differential Revision: https://reviews.freebsd.org/D7616
2016-09-02 18:22:56 +00:00
if ((sigprop(sig) & SIGPROP_CORE) &&
thread_single(p, SINGLE_NO_EXIT) == 0) {
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
p->p_sig = sig;
Log SA_CORE signals Obtained from: FreeBSD 1.x
1994-09-30 00:38:34 +00:00
/*
* Log signals which would cause core dumps
Remove trailing whitespace.
1995-05-30 08:16:23 +00:00
* (Log as LOG_INFO to appease those who don't want
Log SA_CORE signals Obtained from: FreeBSD 1.x
1994-09-30 00:38:34 +00:00
* these messages.)
* XXX : Todo, as well as euid, write out ruid too
- Push down Giant in exit() and wait(). - Push Giant down a bit in coredump() and call coredump() with the proc lock already held rather than unlocking it only to turn around and relock it. Requested by: peter
2004-03-05 22:39:53 +00:00
* Note that coredump() drops proc lock.
Log SA_CORE signals Obtained from: FreeBSD 1.x
1994-09-30 00:38:34 +00:00
*/
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
if (coredump(td) == 0)
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
sig |= WCOREFLAG;
Make the logging of abnormally exiting processes optional by a sysctl. PR: kern/1711 Submitted by: Nick Sayer <nsayer@kfu.com>
1998-07-28 22:34:12 +00:00
if (kern_logsigexit)
log(LOG_INFO,
Make kernel print jail ID when logging a process exit Kernel now includes jail ID when logging a process exit. jid is 0 for unjailed processes. Submitted by: Marie Helene Kvello-Aune <freebsd@mhka.no> Relnotes: yes Sponsored by: Modirum MDPay Differential Revision: https://reviews.freebsd.org/D18618
2018-12-29 14:48:51 +00:00
"pid %d (%s), jid %d, uid %d: exited on "
Simplify jail ID printing on process exit As suggested by kib@, we don't need to check p_ucred, because that's only NULL during process creation, and cr_prison is never NULL.
2018-12-29 21:36:02 +00:00
"signal %d%s\n", p->p_pid, p->p_comm,
p->p_ucred->cr_prison->pr_id,
Remove unneeded NULL check for td_ucred td_ucred is always set, so we don't need the ternary expression to check for it.
2019-01-04 21:12:17 +00:00
td->td_ucred->cr_uid,
sigset_t change (part 2 of 5) ----------------------------- The core of the signalling code has been rewritten to operate on the new sigset_t. No methodological changes have been made. Most references to a sigset_t object are through macros (see signalvar.h) to create a level of abstraction and to provide a basis for further improvements. The NSIG constant has not been changed to reflect the maximum number of signals possible. The reason is that it breaks programs (especially shells) which assume that all signals have a non-null name in sys_signame. See src/bin/sh/trap.c for an example. Instead _SIG_MAXSIG has been introduced to hold the maximum signal possible with the new sigset_t. struct sigprop has been moved from signalvar.h to kern_sig.c because a) it is only used there, and b) access must be done though function sigprop(). The latter because the table doesn't holds properties for all signals, but only for the first NSIG signals. signal.h has been reorganized to make reading easier and to add the new and/or modified structures. The "old" structures are moved to signalvar.h to prevent namespace polution. Especially the coda filesystem suffers from the change, because it contained lines like (p->p_sigmask == SIGIO), which is easy to do for integral types, but not for compound types. NOTE: kdump (and port linux_kdump) must be recompiled. Thanks to Garrett Wollman and Daniel Eischen for pressing the importance of changing sigreturn as well.
1999-09-29 15:03:48 +00:00
sig &~ WCOREFLAG,
sig & WCOREFLAG ? " (core dumped)" : "");
- Push down Giant in exit() and wait(). - Push Giant down a bit in coredump() and call coredump() with the proc lock already held rather than unlocking it only to turn around and relock it. Requested by: peter
2004-03-05 22:39:53 +00:00
} else
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
exit1(td, 0, sig);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* NOTREACHED */
}
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
/*
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
* Send queued SIGCHLD to parent when child process's state
* is changed.
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
*/
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
static void
sigparent(struct proc *p, int reason, int status)
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
{
PROC_LOCK_ASSERT(p, MA_OWNED);
PROC_LOCK_ASSERT(p->p_pptr, MA_OWNED);
if (p->p_ksi != NULL) {
p->p_ksi->ksi_signo = SIGCHLD;
p->p_ksi->ksi_code = reason;
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
p->p_ksi->ksi_status = status;
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
p->p_ksi->ksi_pid = p->p_pid;
p->p_ksi->ksi_uid = p->p_ucred->cr_ruid;
if (KSI_ONQ(p->p_ksi))
return;
}
Tweak the in-kernel API for sending signals to threads: - Rename tdsignal() to tdsendsignal() and make it private to kern_sig.c. - Add tdsignal() and tdksignal() routines that mirror psignal() and pksignal() except that they accept a thread as an argument instead of a process. They send a signal to a specific thread rather than to an individual process. Reviewed by: kib
2010-06-29 20:41:52 +00:00
pksignal(p->p_pptr, SIGCHLD, p->p_ksi);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
}
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
static void
Fix siginfo_t.si_status for wait6/waitid/SIGCHLD. Per POSIX, si_status should contain the value passed to exit() for si_code==CLD_EXITED and the signal number for other si_code. This was incorrect for CLD_EXITED and CLD_DUMPED. This is still not fully POSIX-compliant (Austin group issue #594 says that the full value passed to exit() shall be returned via si_status, not just the low 8 bits) but is sufficient for a si_status-related test in libnih (upstart, Debian/kFreeBSD). PR: kern/184002 Reported by: Dmitrijs Ledkovs Tested by: Dmitrijs Ledkovs
2013-11-17 22:31:23 +00:00
childproc_jobstate(struct proc *p, int reason, int sig)
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
{
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
struct sigacts *ps;
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
PROC_LOCK_ASSERT(p->p_pptr, MA_OWNED);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
/*
* Wake up parent sleeping in kern_wait(), also send
* SIGCHLD to parent, but SIGCHLD does not guarantee
* that parent will awake, because parent may masked
* the signal.
*/
p->p_pptr->p_flag |= P_STATCHILD;
wakeup(p->p_pptr);
ps = p->p_pptr->p_sigacts;
mtx_lock(&ps->ps_mtx);
if ((ps->ps_flag & PS_NOCLDSTOP) == 0) {
mtx_unlock(&ps->ps_mtx);
Fix siginfo_t.si_status for wait6/waitid/SIGCHLD. Per POSIX, si_status should contain the value passed to exit() for si_code==CLD_EXITED and the signal number for other si_code. This was incorrect for CLD_EXITED and CLD_DUMPED. This is still not fully POSIX-compliant (Austin group issue #594 says that the full value passed to exit() shall be returned via si_status, not just the low 8 bits) but is sufficient for a si_status-related test in libnih (upstart, Debian/kFreeBSD). PR: kern/184002 Reported by: Dmitrijs Ledkovs Tested by: Dmitrijs Ledkovs
2013-11-17 22:31:23 +00:00
sigparent(p, reason, sig);
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
} else
mtx_unlock(&ps->ps_mtx);
}
void
childproc_stopped(struct proc *p, int reason)
{
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
childproc_jobstate(p, reason, p->p_xsig);
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
}
void
childproc_continued(struct proc *p)
{
childproc_jobstate(p, CLD_CONTINUED, SIGCONT);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
}
void
childproc_exited(struct proc *p)
{
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
int reason, status;
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
The si_status field of the siginfo_t, provided by the waitid(2) and SIGCHLD signal, should keep full 32 bits of the status passed to the _exit(2). Split the combined p_xstat of the struct proc into the separate exit status p_xexit for normal process exit, and signalled termination information p_xsig. Kernel-visible macro KW_EXITCODE() reconstructs old p_xstat from p_xexit and p_xsig. p_xexit contains complete status and copied out into si_status. Requested by: Joerg Schilling Reviewed by: jilles (previous version), pho Tested by: pho Sponsored by: The FreeBSD Foundation
2015-07-18 09:02:50 +00:00
if (WCOREDUMP(p->p_xsig)) {
reason = CLD_DUMPED;
status = WTERMSIG(p->p_xsig);
} else if (WIFSIGNALED(p->p_xsig)) {
reason = CLD_KILLED;
status = WTERMSIG(p->p_xsig);
} else {
reason = CLD_EXITED;
status = p->p_xexit;
}
Create childproc_jobstate function to report job control state, this also fixes a bug in childproc_continued which ignored PS_NOCLDSTOP.
2006-02-04 14:10:57 +00:00
/*
* XXX avoid calling wakeup(p->p_pptr), the work is
* done in exit1().
*/
sigparent(p, reason, status);
Add support for queueing SIGCHLD same as other UNIX systems did. For each child process whose status has been changed, a SIGCHLD instance is queued, if the signal is stilling pending, and process changed status several times, signal information is updated to reflect latest process status. If wait() returns because the status of a child process is available, pending SIGCHLD signal associated with the child process is discarded. Any other pending SIGCHLD signals remain pending. The signal information is allocated at the same time when proc structure is allocated, if process signal queue is fully filled or there is a memory shortage, it can still send the signal to process. There is a booting time tunable kern.sigqueue.queue_sigchild which can control the behavior, setting it to zero disables the SIGCHLD queueing feature, the tunable will be removed if the function is proved that it is stable enough. Tested on: i386 (SMP and UP)
2005-11-08 09:09:26 +00:00
}
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
#define MAX_NUM_CORE_FILES 100000
Add the NUM_CORE_FILES kernel config option which specifies the limit for the number of core files allowed by a particular process when using the %I core file name pattern. Sanity check at compile time to ensure the value is within the valid range of 0-10. Reviewed by: jtl, sjg Approved by: sjg (mentor) Sponsored by: Juniper Networks, Inc. Differential Revision: https://reviews.freebsd.org/D6812
2016-07-27 03:21:02 +00:00
#ifndef NUM_CORE_FILES
#define NUM_CORE_FILES 5
#endif
CTASSERT(NUM_CORE_FILES >= 0 && NUM_CORE_FILES <= MAX_NUM_CORE_FILES);
static int num_cores = NUM_CORE_FILES;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
static int
sysctl_debug_num_cores_check (SYSCTL_HANDLER_ARGS)
{
int error;
int new_val;
Do not report a stack garbage as the old value for debug.ncores sysctl. Reported by: brucec
2010-06-21 09:51:25 +00:00
new_val = num_cores;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
error = sysctl_handle_int(oidp, &new_val, 0, req);
if (error != 0 || req->newptr == NULL)
return (error);
Add the NUM_CORE_FILES kernel config option which specifies the limit for the number of core files allowed by a particular process when using the %I core file name pattern. Sanity check at compile time to ensure the value is within the valid range of 0-10. Reviewed by: jtl, sjg Approved by: sjg (mentor) Sponsored by: Juniper Networks, Inc. Differential Revision: https://reviews.freebsd.org/D6812
2016-07-27 03:21:02 +00:00
if (new_val > MAX_NUM_CORE_FILES)
new_val = MAX_NUM_CORE_FILES;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
if (new_val < 0)
new_val = 0;
num_cores = new_val;
return (0);
}
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
SYSCTL_PROC(_debug, OID_AUTO, ncores,
CTLTYPE_INT | CTLFLAG_RW | CTLFLAG_NEEDGIANT, 0, sizeof(int),
sysctl_debug_num_cores_check, "I",
"Maximum number of generated process corefiles while using index format");
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
Add support for zstd-compressed user and kernel core dumps. This works similarly to the existing gzip compression support, but zstd is typically faster and gives better compression ratios. Support for this functionality must be configured by adding ZSTDIO to one's kernel configuration file. dumpon(8)'s new -Z option is used to configure zstd compression for kernel dumps. savecore(8) now recognizes and saves zstd-compressed kernel dumps with a .zst extension. Submitted by: cem (original version) Relnotes: yes Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D13101, https://reviews.freebsd.org/D13633
2018-02-13 19:28:02 +00:00
#define GZIP_SUFFIX ".gz"
#define ZSTD_SUFFIX ".zst"
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
Generalize the gzio API. We currently use a set of subroutines in kern_gzio.c to perform compression of user and kernel core dumps. In the interest of adding support for other compression algorithms (zstd) in this role without complicating the API consumers, add a simple compressor API which can be used to select an algorithm. Also change the (non-default) GZIO kernel option to not enable compressed user cores by default. It's not clear that such a default would be desirable with support for multiple algorithms implemented, and it's inconsistent in that it isn't applied to kernel dumps. Reviewed by: cem Differential Revision: https://reviews.freebsd.org/D13632
2018-01-08 21:27:41 +00:00
int compress_user_cores = 0;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
Generalize the gzio API. We currently use a set of subroutines in kern_gzio.c to perform compression of user and kernel core dumps. In the interest of adding support for other compression algorithms (zstd) in this role without complicating the API consumers, add a simple compressor API which can be used to select an algorithm. Also change the (non-default) GZIO kernel option to not enable compressed user cores by default. It's not clear that such a default would be desirable with support for multiple algorithms implemented, and it's inconsistent in that it isn't applied to kernel dumps. Reviewed by: cem Differential Revision: https://reviews.freebsd.org/D13632
2018-01-08 21:27:41 +00:00
static int
sysctl_compress_user_cores(SYSCTL_HANDLER_ARGS)
{
int error, val;
val = compress_user_cores;
error = sysctl_handle_int(oidp, &val, 0, req);
if (error != 0 || req->newptr == NULL)
return (error);
if (val != 0 && !compressor_avail(val))
return (EINVAL);
compress_user_cores = val;
return (error);
}
Mark more nodes as CTLFLAG_MPSAFE or CTLFLAG_NEEDGIANT (17 of many) r357614 added CTLFLAG_NEEDGIANT to make it easier to find nodes that are still not MPSAFE (or already are but aren’t properly marked). Use it in preparation for a general review of all nodes. This is non-functional change that adds annotations to SYSCTL_NODE and SYSCTL_PROC nodes using one of the soon-to-be-required flags. Mark all obvious cases as MPSAFE. All entries that haven't been marked as MPSAFE before are by default marked as NEEDGIANT Approved by: kib (mentor, blanket) Commented by: kib, gallatin, melifaro Differential Revision: https://reviews.freebsd.org/D23718
2020-02-26 14:26:36 +00:00
SYSCTL_PROC(_kern, OID_AUTO, compress_user_cores,
CTLTYPE_INT | CTLFLAG_RWTUN | CTLFLAG_NEEDGIANT, 0, sizeof(int),
sysctl_compress_user_cores, "I",
Add support for zstd-compressed user and kernel core dumps. This works similarly to the existing gzip compression support, but zstd is typically faster and gives better compression ratios. Support for this functionality must be configured by adding ZSTDIO to one's kernel configuration file. dumpon(8)'s new -Z option is used to configure zstd compression for kernel dumps. savecore(8) now recognizes and saves zstd-compressed kernel dumps with a .zst extension. Submitted by: cem (original version) Relnotes: yes Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D13101, https://reviews.freebsd.org/D13633
2018-02-13 19:28:02 +00:00
"Enable compression of user corefiles ("
__XSTRING(COMPRESS_GZIP) " = gzip, "
__XSTRING(COMPRESS_ZSTD) " = zstd)");
Generalize the gzio API. We currently use a set of subroutines in kern_gzio.c to perform compression of user and kernel core dumps. In the interest of adding support for other compression algorithms (zstd) in this role without complicating the API consumers, add a simple compressor API which can be used to select an algorithm. Also change the (non-default) GZIO kernel option to not enable compressed user cores by default. It's not clear that such a default would be desirable with support for multiple algorithms implemented, and it's inconsistent in that it isn't applied to kernel dumps. Reviewed by: cem Differential Revision: https://reviews.freebsd.org/D13632
2018-01-08 21:27:41 +00:00
int compress_user_cores_level = 6;
SYSCTL_INT(_kern, OID_AUTO, compress_user_cores_level, CTLFLAG_RWTUN,
&compress_user_cores_level, 0,
"Corefile compression level");
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
coredump: protect corefilename access with a lock Previously format string traversal could happen while the string itself was being modified. Use allproc_lock as coredumping is a rare operation and as such we don't have to create a dedicated lock. Submitted by: Tiwei Bie <btw mail.ustc.edu.cn> Reviewed by: kib X-Additional: JuniorJobs project
2015-03-21 04:39:33 +00:00
/*
* Protect the access to corefilename[] by allproc_lock.
*/
#define corefilename_lock allproc_lock
Fix an off by one error. MAXPATHLEN already has +1.
2004-11-15 20:51:32 +00:00
static char corefilename[MAXPATHLEN] = {"%N.core"};
Define tunable instead of using CTLFLAG_RWTUN flag with kern.corefile. The allproc_lock lock used in the sysctl_kern_corefile function is initialized in the procinit function which is called after setting sysctl values at boot. That means if we set kern.corefile at boot we will be trying to use lock with is uninitialized and machine will crash. If we define kern.corefile as tunable instead of using CTFLAG_RWTUN we will not call the sysctl_kern_corefile function and we will not use an uninitialized lock. When machine will boot then we will start using function depending on the lock. Reviewed by: pjd
2016-06-09 20:23:30 +00:00
TUNABLE_STR("kern.corefile", corefilename, sizeof(corefilename));
coredump: protect corefilename access with a lock Previously format string traversal could happen while the string itself was being modified. Use allproc_lock as coredumping is a rare operation and as such we don't have to create a dedicated lock. Submitted by: Tiwei Bie <btw mail.ustc.edu.cn> Reviewed by: kib X-Additional: JuniorJobs project
2015-03-21 04:39:33 +00:00
static int
sysctl_kern_corefile(SYSCTL_HANDLER_ARGS)
{
int error;
sx_xlock(&corefilename_lock);
error = sysctl_handle_string(oidp, corefilename, sizeof(corefilename),
req);
sx_xunlock(&corefilename_lock);
return (error);
}
Define tunable instead of using CTLFLAG_RWTUN flag with kern.corefile. The allproc_lock lock used in the sysctl_kern_corefile function is initialized in the procinit function which is called after setting sysctl values at boot. That means if we set kern.corefile at boot we will be trying to use lock with is uninitialized and machine will crash. If we define kern.corefile as tunable instead of using CTFLAG_RWTUN we will not call the sysctl_kern_corefile function and we will not use an uninitialized lock. When machine will boot then we will start using function depending on the lock. Reviewed by: pjd
2016-06-09 20:23:30 +00:00
SYSCTL_PROC(_kern, OID_AUTO, corefile, CTLTYPE_STRING | CTLFLAG_RW |
coredump: protect corefilename access with a lock Previously format string traversal could happen while the string itself was being modified. Use allproc_lock as coredumping is a rare operation and as such we don't have to create a dedicated lock. Submitted by: Tiwei Bie <btw mail.ustc.edu.cn> Reviewed by: kib X-Additional: JuniorJobs project
2015-03-21 04:39:33 +00:00
CTLFLAG_MPSAFE, 0, 0, sysctl_kern_corefile, "A",
"Process corefile name format string");
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
static void
vnode_close_locked(struct thread *td, struct vnode *vp)
{
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(vp);
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
vn_close(vp, FWRITE, td->td_ucred, td);
}
/*
* If the core format has a %I in it, then we need to check
* for existing corefiles before defining a name.
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
* To do this we iterate over 0..ncores to find a
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
* non-existing core file name to use. If all core files are
* already used we choose the oldest one.
*/
static int
corefile_open_last(struct thread *td, char *name, int indexpos,
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
int indexlen, int ncores, struct vnode **vpp)
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
{
struct vnode *oldvp, *nextvp, *vp;
struct vattr vattr;
struct nameidata nd;
int error, i, flags, oflags, cmode;
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
char ch;
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
struct timespec lasttime;
nextvp = oldvp = NULL;
cmode = S_IRUSR | S_IWUSR;
oflags = VN_OPEN_NOAUDIT | VN_OPEN_NAMECACHE |
(capmode_coredump ? VN_OPEN_NOCAPCHECK : 0);
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
for (i = 0; i < ncores; i++) {
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
flags = O_CREAT | FWRITE | O_NOFOLLOW;
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
ch = name[indexpos + indexlen];
(void)snprintf(name + indexpos, indexlen + 1, "%.*u", indexlen,
i);
name[indexpos + indexlen] = ch;
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_SYSSPACE, name, td);
error = vn_open_cred(&nd, &flags, cmode, oflags, td->td_ucred,
NULL);
if (error != 0)
break;
vp = nd.ni_vp;
NDFREE(&nd, NDF_ONLY_PNBUF);
if ((flags & O_CREAT) == O_CREAT) {
nextvp = vp;
break;
}
error = VOP_GETATTR(vp, &vattr, td->td_ucred);
if (error != 0) {
vnode_close_locked(td, vp);
break;
}
if (oldvp == NULL ||
lasttime.tv_sec > vattr.va_mtime.tv_sec ||
(lasttime.tv_sec == vattr.va_mtime.tv_sec &&
lasttime.tv_nsec >= vattr.va_mtime.tv_nsec)) {
if (oldvp != NULL)
corefile_open_last: don't keep a locked vnode while locking other ones Consider this scenario: - kern.corefile=/var/coredumps/%N.%U.%I.core - multiple processes with the same name crash at the same time It's possible that one process selects existing file N as oldvp while it keeps looking for an unused file number. Another process scans through files and stumbles upon N. That process would be blocked on the vnode lock while holding the directory vnode exclusively locked. The first process would, thus, get blocked on the directory's vnode lock. More generally, holding a file's vnode lock (oldvp) while trying to lock its directory (for the next lookup) is a violation of the vnode locking order. I have observed this deadlock in the wild. So, the change to keep oldvp "opened" but unlocked and to lock it again only if it's to be returned as the result. As kib noted, an alternative would be to keep the directory locked and to use VOP_LOOKUP directly for scanning through existing core files. Reviewed by: kib MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D25027
2020-05-29 07:44:02 +00:00
vn_close(oldvp, FWRITE, td->td_ucred, td);
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
oldvp = vp;
corefile_open_last: don't keep a locked vnode while locking other ones Consider this scenario: - kern.corefile=/var/coredumps/%N.%U.%I.core - multiple processes with the same name crash at the same time It's possible that one process selects existing file N as oldvp while it keeps looking for an unused file number. Another process scans through files and stumbles upon N. That process would be blocked on the vnode lock while holding the directory vnode exclusively locked. The first process would, thus, get blocked on the directory's vnode lock. More generally, holding a file's vnode lock (oldvp) while trying to lock its directory (for the next lookup) is a violation of the vnode locking order. I have observed this deadlock in the wild. So, the change to keep oldvp "opened" but unlocked and to lock it again only if it's to be returned as the result. As kib noted, an alternative would be to keep the directory locked and to use VOP_LOOKUP directly for scanning through existing core files. Reviewed by: kib MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D25027
2020-05-29 07:44:02 +00:00
VOP_UNLOCK(oldvp);
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
lasttime = vattr.va_mtime;
} else {
vnode_close_locked(td, vp);
}
}
if (oldvp != NULL) {
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
if (nextvp == NULL) {
if ((td->td_proc->p_flag & P_SUGID) != 0) {
error = EFAULT;
corefile_open_last: don't keep a locked vnode while locking other ones Consider this scenario: - kern.corefile=/var/coredumps/%N.%U.%I.core - multiple processes with the same name crash at the same time It's possible that one process selects existing file N as oldvp while it keeps looking for an unused file number. Another process scans through files and stumbles upon N. That process would be blocked on the vnode lock while holding the directory vnode exclusively locked. The first process would, thus, get blocked on the directory's vnode lock. More generally, holding a file's vnode lock (oldvp) while trying to lock its directory (for the next lookup) is a violation of the vnode locking order. I have observed this deadlock in the wild. So, the change to keep oldvp "opened" but unlocked and to lock it again only if it's to be returned as the result. As kib noted, an alternative would be to keep the directory locked and to use VOP_LOOKUP directly for scanning through existing core files. Reviewed by: kib MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D25027
2020-05-29 07:44:02 +00:00
vn_close(oldvp, FWRITE, td->td_ucred, td);
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
} else {
nextvp = oldvp;
corefile_open_last: don't keep a locked vnode while locking other ones Consider this scenario: - kern.corefile=/var/coredumps/%N.%U.%I.core - multiple processes with the same name crash at the same time It's possible that one process selects existing file N as oldvp while it keeps looking for an unused file number. Another process scans through files and stumbles upon N. That process would be blocked on the vnode lock while holding the directory vnode exclusively locked. The first process would, thus, get blocked on the directory's vnode lock. More generally, holding a file's vnode lock (oldvp) while trying to lock its directory (for the next lookup) is a violation of the vnode locking order. I have observed this deadlock in the wild. So, the change to keep oldvp "opened" but unlocked and to lock it again only if it's to be returned as the result. As kib noted, an alternative would be to keep the directory locked and to use VOP_LOOKUP directly for scanning through existing core files. Reviewed by: kib MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D25027
2020-05-29 07:44:02 +00:00
error = vn_lock(nextvp, LK_EXCLUSIVE);
if (error != 0) {
vn_close(nextvp, FWRITE, td->td_ucred,
td);
nextvp = NULL;
}
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
}
} else {
corefile_open_last: don't keep a locked vnode while locking other ones Consider this scenario: - kern.corefile=/var/coredumps/%N.%U.%I.core - multiple processes with the same name crash at the same time It's possible that one process selects existing file N as oldvp while it keeps looking for an unused file number. Another process scans through files and stumbles upon N. That process would be blocked on the vnode lock while holding the directory vnode exclusively locked. The first process would, thus, get blocked on the directory's vnode lock. More generally, holding a file's vnode lock (oldvp) while trying to lock its directory (for the next lookup) is a violation of the vnode locking order. I have observed this deadlock in the wild. So, the change to keep oldvp "opened" but unlocked and to lock it again only if it's to be returned as the result. As kib noted, an alternative would be to keep the directory locked and to use VOP_LOOKUP directly for scanning through existing core files. Reviewed by: kib MFC after: 2 weeks Differential Revision: https://reviews.freebsd.org/D25027
2020-05-29 07:44:02 +00:00
vn_close(oldvp, FWRITE, td->td_ucred, td);
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
}
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
}
if (error != 0) {
if (nextvp != NULL)
vnode_close_locked(td, oldvp);
} else {
*vpp = nextvp;
}
return (error);
}
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
/*
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
* corefile_open(comm, uid, pid, td, compress, vpp, namep)
* Expand the name described in corefilename, using name, uid, and pid
* and open/create core file.
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
* corefilename is a printf-like string, with three format specifiers:
* %N name of process ("name")
* %P process id (pid)
* %U user id (uid)
* For example, "%N.core" is the default; they can be disabled completely
* by using "/dev/null", or all core files can be stored in "/cores/%U/%N-%P".
* This is controlled by the sysctl variable kern.corefile (see above).
*/
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
static int
corefile_open(const char *comm, uid_t uid, pid_t pid, struct thread *td,
Adds signal number format to kern.corefile Add format capability to core file names to include signal that generated the core. This can help various validation workflows where all cores should not be considered equally (SIGQUIT is often intentional and not an error unlike SIGSEGV or SIGBUS) Submitted by: David Leimbach (leimy2k@gmail.com) Reviewed by: markj MFC after: 1 week Relnotes: sysctl kern.corefile can now include the signal number Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D20970
2019-07-16 15:51:09 +00:00
int compress, int signum, struct vnode **vpp, char **namep)
expand_name fixes: .) don't use MAXPATHLEN + 1, fix logic to compensate. .) style(9) function parameters. .) fix line wrapping. .) remove duplicated error and string handling code. .) don't NUL terminate already NUL terminated string. .) all string length variables changed from int to size_t. .) constify variables. .) catch when corename would be truncated. .) cast pid_t and uid_t args for format string. .) add parens around return arguments. Help and suggestions from: bde
2002-05-08 09:06:47 +00:00
{
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
struct sbuf sb;
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
struct nameidata nd;
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
const char *format;
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
char *hostname, *name;
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
int cmode, error, flags, i, indexpos, indexlen, oflags, ncores;
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
Avoid allocating MAXHOSTNAMELEN bytes on the stack in expand_name(), use the heap instead. Obtained from: Juniper Networks Reviewed by: jhb
2010-04-30 03:15:00 +00:00
hostname = NULL;
expand_name fixes: .) don't use MAXPATHLEN + 1, fix logic to compensate. .) style(9) function parameters. .) fix line wrapping. .) remove duplicated error and string handling code. .) don't NUL terminate already NUL terminated string. .) all string length variables changed from int to size_t. .) constify variables. .) catch when corename would be truncated. .) cast pid_t and uid_t args for format string. .) add parens around return arguments. Help and suggestions from: bde
2002-05-08 09:06:47 +00:00
format = corefilename;
The expand_name() function isn't called with the process lock held anymore, so we can safely use malloc(M_WAITOK) now. Pointed out by: kib
2012-12-19 12:00:09 +00:00
name = malloc(MAXPATHLEN, M_TEMP, M_WAITOK | M_ZERO);
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
indexlen = 0;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
indexpos = -1;
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
ncores = num_cores;
Better variables naming in expand_name() to be more consistent with coredump(). Obtained from: WHEEL Systems
2012-12-17 10:48:10 +00:00
(void)sbuf_new(&sb, name, MAXPATHLEN, SBUF_FIXEDLEN);
coredump: protect corefilename access with a lock Previously format string traversal could happen while the string itself was being modified. Use allproc_lock as coredumping is a rare operation and as such we don't have to create a dedicated lock. Submitted by: Tiwei Bie <btw mail.ustc.edu.cn> Reviewed by: kib X-Additional: JuniorJobs project
2015-03-21 04:39:33 +00:00
sx_slock(&corefilename_lock);
Style cleanups.
2012-12-19 12:10:14 +00:00
for (i = 0; format[i] != '\0'; i++) {
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
switch (format[i]) {
case '%': /* Format character */
i++;
switch (format[i]) {
case '%':
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
sbuf_putc(&sb, '%');
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
break;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
case 'H': /* hostname */
Avoid allocating MAXHOSTNAMELEN bytes on the stack in expand_name(), use the heap instead. Obtained from: Juniper Networks Reviewed by: jhb
2010-04-30 03:15:00 +00:00
if (hostname == NULL) {
hostname = malloc(MAXHOSTNAMELEN,
The expand_name() function isn't called with the process lock held anymore, so we can safely use malloc(M_WAITOK) now. Pointed out by: kib
2012-12-19 12:00:09 +00:00
M_TEMP, M_WAITOK);
Style cleanups.
2012-12-19 12:10:14 +00:00
}
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
getcredhostname(td->td_ucred, hostname,
Avoid allocating MAXHOSTNAMELEN bytes on the stack in expand_name(), use the heap instead. Obtained from: Juniper Networks Reviewed by: jhb
2010-04-30 03:15:00 +00:00
MAXHOSTNAMELEN);
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
sbuf_printf(&sb, "%s", hostname);
break;
Style cleanups.
2012-12-19 12:10:14 +00:00
case 'I': /* autoincrementing index */
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
if (indexpos != -1) {
sbuf_printf(&sb, "%%I");
break;
}
indexpos = sbuf_len(&sb);
sbuf_printf(&sb, "%u", ncores - 1);
indexlen = sbuf_len(&sb) - indexpos;
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
break;
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
case 'N': /* process name */
Better variables naming in expand_name() to be more consistent with coredump(). Obtained from: WHEEL Systems
2012-12-17 10:48:10 +00:00
sbuf_printf(&sb, "%s", comm);
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
break;
case 'P': /* process id */
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
sbuf_printf(&sb, "%u", pid);
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
break;
Adds signal number format to kern.corefile Add format capability to core file names to include signal that generated the core. This can help various validation workflows where all cores should not be considered equally (SIGQUIT is often intentional and not an error unlike SIGSEGV or SIGBUS) Submitted by: David Leimbach (leimy2k@gmail.com) Reviewed by: markj MFC after: 1 week Relnotes: sysctl kern.corefile can now include the signal number Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D20970
2019-07-16 15:51:09 +00:00
case 'S': /* signal number */
sbuf_printf(&sb, "%i", signum);
break;
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
case 'U': /* user id */
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
sbuf_printf(&sb, "%u", uid);
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
break;
default:
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
log(LOG_ERR,
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
"Unknown format character %c in "
"corename `%s'\n", format[i], format);
Style cleanups.
2012-12-19 12:10:14 +00:00
break;
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
}
break;
default:
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
sbuf_putc(&sb, format[i]);
Add break to the default case.
2013-02-17 11:47:58 +00:00
break;
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
}
}
coredump: protect corefilename access with a lock Previously format string traversal could happen while the string itself was being modified. Use allproc_lock as coredumping is a rare operation and as such we don't have to create a dedicated lock. Submitted by: Tiwei Bie <btw mail.ustc.edu.cn> Reviewed by: kib X-Additional: JuniorJobs project
2015-03-21 04:39:33 +00:00
sx_sunlock(&corefilename_lock);
Avoid allocating MAXHOSTNAMELEN bytes on the stack in expand_name(), use the heap instead. Obtained from: Juniper Networks Reviewed by: jhb
2010-04-30 03:15:00 +00:00
free(hostname, M_TEMP);
Generalize the gzio API. We currently use a set of subroutines in kern_gzio.c to perform compression of user and kernel core dumps. In the interest of adding support for other compression algorithms (zstd) in this role without complicating the API consumers, add a simple compressor API which can be used to select an algorithm. Also change the (non-default) GZIO kernel option to not enable compressed user cores by default. It's not clear that such a default would be desirable with support for multiple algorithms implemented, and it's inconsistent in that it isn't applied to kernel dumps. Reviewed by: cem Differential Revision: https://reviews.freebsd.org/D13632
2018-01-08 21:27:41 +00:00
if (compress == COMPRESS_GZIP)
Add support for zstd-compressed user and kernel core dumps. This works similarly to the existing gzip compression support, but zstd is typically faster and gives better compression ratios. Support for this functionality must be configured by adding ZSTDIO to one's kernel configuration file. dumpon(8)'s new -Z option is used to configure zstd compression for kernel dumps. savecore(8) now recognizes and saves zstd-compressed kernel dumps with a .zst extension. Submitted by: cem (original version) Relnotes: yes Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D13101, https://reviews.freebsd.org/D13633
2018-02-13 19:28:02 +00:00
sbuf_printf(&sb, GZIP_SUFFIX);
else if (compress == COMPRESS_ZSTD)
sbuf_printf(&sb, ZSTD_SUFFIX);
Replace sbuf_overflowed() with sbuf_error(), which returns any error code associated with overflow or with the drain function. While this function is not expected to be used often, it produces more information in the form of an errno that sbuf_overflowed() did.
2010-09-10 16:42:16 +00:00
if (sbuf_error(&sb) != 0) {
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
log(LOG_ERR, "pid %ld (%s), uid (%lu): corename is too "
Better variables naming in expand_name() to be more consistent with coredump(). Obtained from: WHEEL Systems
2012-12-17 10:48:10 +00:00
"long\n", (long)pid, comm, (u_long)uid);
Avoid allocating MAXHOSTNAMELEN bytes on the stack in expand_name(), use the heap instead. Obtained from: Juniper Networks Reviewed by: jhb
2010-04-30 03:15:00 +00:00
sbuf_delete(&sb);
Better variables naming in expand_name() to be more consistent with coredump(). Obtained from: WHEEL Systems
2012-12-17 10:48:10 +00:00
free(name, M_TEMP);
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
return (ENOMEM);
Use sbuf routines to construct core dump filenames rather than custom string buffer handling, making the code both easier to read and more robust against string-handling bugs. MFC after: 1 week
2008-03-08 16:31:29 +00:00
}
sbuf_finish(&sb);
sbuf_delete(&sb);
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
if (indexpos != -1) {
Extend amount of possible coredumps from 10 to 100000 when using index format. The amount of digits in the name of corefile is assigned dynamically. Reviewed by: kib Differential Revision: https://reviews.freebsd.org/D16118
2018-07-15 17:10:12 +00:00
error = corefile_open_last(td, name, indexpos, indexlen, ncores,
vpp);
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
if (error != 0) {
log(LOG_ERR,
"pid %d (%s), uid (%u): Path `%s' failed "
"on initial open test, error = %d\n",
pid, comm, uid, name, error);
}
} else {
cmode = S_IRUSR | S_IWUSR;
oflags = VN_OPEN_NOAUDIT | VN_OPEN_NAMECACHE |
(capmode_coredump ? VN_OPEN_NOCAPCHECK : 0);
flags = O_CREAT | FWRITE | O_NOFOLLOW;
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
if ((td->td_proc->p_flag & P_SUGID) != 0)
flags |= O_EXCL;
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
NDINIT(&nd, LOOKUP, NOFOLLOW, UIO_SYSSPACE, name, td);
error = vn_open_cred(&nd, &flags, cmode, oflags, td->td_ucred,
NULL);
if (error == 0) {
*vpp = nd.ni_vp;
NDFREE(&nd, NDF_ONLY_PNBUF);
Merge projects/enhanced_coredumps (r204346) into HEAD: Enhanced process coredump routines. This brings in the following features: 1) Limit number of cores per process via the %I coredump formatter. Example: if corefilename is set to %N.%I.core AND num_cores = 3, then if a process "rpd" cores, then the corefile will be named "rpd.0.core", however if it cores again, then the kernel will generate "rpd.1.core" until we hit the limit of "num_cores". this is useful to get several corefiles, but also prevent filling the machine with corefiles. 2) Encode machine hostname in core dump name via %H. 3) Compress coredumps, useful for embedded platforms with limited space. A sysctl kern.compress_user_cores is made available if turned on. To enable compressed coredumps, the following config options need to be set: options COMPRESS_USER_CORES device zlib # brings in the zlib requirements. device gzio # brings in the kernel vnode gzip output module. 4) Eventhandlers are fired to indicate coredumps in progress. 5) The imgact sv_coredump routine has grown a flag to pass in more state, currently this is used only for passing a flag down to compress the coredump or not. Note that the gzio facility can be used for generic output of gzip'd streams via vnodes. Obtained from: Juniper Networks Reviewed by: kan
2010-03-02 06:58:58 +00:00
}
}
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
core(5): overwrite the oldest core dump The '%I' format in the kern.corefile sysctl limits the number of core files that a process can generate to the number stored in the debug.ncores sysctl. The '%I' format is replaced by the single digit index. Previously, if all indexes were taken the kernel would overwrite only a core file with the highest index in a filename. Currently the system will create a new core file if there is a free index or if all slots are taken it will overwrite the oldest one. Reviewed by: kib(code), bcr (updating) Differential Revision: https://reviews.freebsd.org/D15991 Differential Revision: https://reviews.freebsd.org/D16084
2018-07-01 17:28:46 +00:00
if (error != 0) {
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
#ifdef AUDIT
audit_proc_coredump(td, name, error);
#endif
free(name, M_TEMP);
return (error);
}
*namep = name;
return (0);
Add support for run-time configuration of core file names. In a nutshell, you can specify the corefile name by using: sysctl -w kern.corefile="format" where format is a pathname (relative or absolute -- default is "%N.core"), with "%N" (process name), "%P" (process ID), and "%U" (user ID) formats. Reviewed by: Mike Smith, with strong requests by Julian :)
1998-07-08 06:38:39 +00:00
}
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
/*
* Dump a process' core. The main routine does some
* policy checking, and creates the name of the coredump;
* then it passes on a vnode and a size limit to the process-specific
* coredump routine if there is one; if there _is not_ one, it returns
* ENOSYS; otherwise it returns the error from the process-specific routine.
*/
static int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
coredump(struct thread *td)
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct proc *p = td->td_proc;
Minor style tweaks. Obtained from: WHEEL Systems
2012-12-17 10:51:22 +00:00
struct ucred *cred = td->td_ucred;
struct vnode *vp;
This brings in a Yahoo coredump patch from Paul, with additional mods by me (addition of vn_rdwr_inchunks). The problem Yahoo is solving is that if you have large process images core dumping, or you have a large number of forked processes all core dumping at the same time, the original coredump code would leave the vnode locked throughout. This can cause the directory vnode to get locked up, which can cause the parent directory vnode to get locked up, and so on all the way to the root node, locking the entire machine up for extremely long periods of time. This patch solves the problem in two ways. First it uses an advisory non-blocking lock to abort multiple processes trying to core to the same file. Second (my contribution) it chunks up the writes and uses bwillwrite() to avoid holding the vnode locked while blocking in the buffer cache. Submitted by: ps Reviewed by: dillon MFC after: 2 weeks
2001-09-08 20:02:33 +00:00
struct flock lf;
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
struct vattr vattr;
cache: replace kern___getcwd with vn_getcwd The previous routine was resulting in extra data copies most notably in linux_getcwd.
2020-02-01 20:38:38 +00:00
size_t fullpathsize;
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
int error, error1, locked;
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
char *name; /* name of corefile */
Fixes for i/o during coredumping: - Do not dump into system files. - Do not acquire write reference to the mount point where img.core is written, in the coredump(). The vn_rdwr() calls from ELF imgact request the write ref from vn_rdwr(). Recursive acqusition of the write ref deadlocks with the unmount. - Instead, take the range lock for the whole core file. This prevents parallel dumping from two processes executing the same image, converting the useless interleaved dump into sequential dumping, with second core overwriting the first. Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks
2014-10-04 18:35:00 +00:00
void *rl_cookie;
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
off_t limit;
Notify devd(8) when a process crashed. This change implements a notification (via devctl) to userland when the kernel produces coredumps after a process has crashed. devd can then run a specific command to produce a human readable crash report. The command is most usually a helper that runs gdb/lldb commands on the file/coredump pair. It's possible to use this functionality for implementing automatic generation of crash reports. devd(8) will be notified of the full path of the binary that crashed and the full path of the coredump file.
2015-02-09 23:13:50 +00:00
char *fullpath, *freepath = NULL;
Fix devctl generation for core files. We have a problem with vn_fullpath_global when the file exists. Work around it by printing the full path if the core file name starts with /, or current working directory followed by the filename if not. Sponsored by: Netflix Differential Review: https://reviews.freebsd.org/D16026
2018-06-27 04:11:09 +00:00
struct sbuf *sb;
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
- Push down Giant in exit() and wait(). - Push Giant down a bit in coredump() and call coredump() with the proc lock already held rather than unlocking it only to turn around and relock it. Requested by: peter
2004-03-05 22:39:53 +00:00
PROC_LOCK_ASSERT(p, MA_OWNED);
Suspend all other threads in the process while generating a core dump. The main reason for doing this is that the ELF dump handler expects the thread list to be fixed while the dump header is generated, so an upcall that occurs at the wrong time can lead to buffer overruns and other Bad Things. Another solution would be to grab sched_lock in the ELF dump handler, but we might as well single-thread, since the process is about to die. Furthermore, I think this should ensure that the register sets in the core file are sequentially consistent.
2005-04-10 02:31:24 +00:00
MPASS((p->p_flag & P_HADTHREADS) == 0 || p->p_singlethread == td);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
Add procctl(2) PROC_TRACE_CTL command to enable or disable debugger attachment to the process. Note that the command is not intended to be a security measure, rather it is an obfuscation feature, implemented for parity with other operating systems. Discussed with: jilles, rwatson Man page fixes by: rwatson Sponsored by: The FreeBSD Foundation MFC after: 1 week
2015-01-18 15:13:11 +00:00
if (!do_coredump || (!sugid_coredump && (p->p_flag & P_SUGID) != 0) ||
(p->p_flag2 & P2_NOTRACE) != 0) {
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
return (EFAULT);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
}
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
/*
Bail out of the process early if the coredumpfile limit is 0. PR: kern/14540 Reviewed by: Nate Williams
1999-10-30 18:55:11 +00:00
* Note that the bulk of limit checking is done after
* the corefile is created. The exception is if the limit
* for corefiles is 0, in which case we don't bother
* creating the corefile at all. This layout means that
* a corefile is truncated instead of not being created,
* if it is larger than the limit.
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
*/
Implement lockless resource limits. Use the same scheme implemented to manage credentials. Code needing to look at process's credentials (as opposed to thred's) is provided with *_proc variants of relevant functions. Places which possibly had to take the proc lock anyway still use the proc pointer to access limits.
2015-06-10 10:48:12 +00:00
limit = (off_t)lim_cur(td, RLIMIT_CORE);
Fix support for RACCT_CORE by merging forgotten file.
2011-05-26 18:54:07 +00:00
if (limit == 0 || racct_get_available(p, RACCT_CORE) == 0) {
PROC_UNLOCK(p);
Locking for the per-process resource limits structure. - struct plimit includes a mutex to protect a reference count. The plimit structure is treated similarly to struct ucred in that is is always copy on write, so having a reference to a structure is sufficient to read from it without needing a further lock. - The proc lock protects the p_limit pointer and must be held while reading limits from a process to keep the limit structure from changing out from under you while reading from it. - Various global limits that are ints are not protected by a lock since int writes are atomic on all the archs we support and thus a lock wouldn't buy us anything. - All accesses to individual resource limits from a process are abstracted behind a simple lim_rlimit(), lim_max(), and lim_cur() API that return either an rlimit, or the current or max individual limit of the specified resource from a process. - dosetrlimit() was renamed to kern_setrlimit() to match existing style of other similar syscall helper functions. - The alpha OSF/1 compat layer no longer calls getrlimit() and setrlimit() (it didn't used the stackgap when it should have) but uses lim_rlimit() and kern_setrlimit() instead. - The svr4 compat no longer uses the stackgap for resource limits calls, but uses lim_rlimit() and kern_setrlimit() instead. - The ibcs2 compat no longer uses the stackgap for resource limits. It also no longer uses the stackgap for accessing sysctl's for the ibcs2_sysconf() syscall but uses kernel_sysctl() instead. As a result, ibcs2_sysconf() no longer needs Giant. - The p_rlimit macro no longer exists. Submitted by: mtm (mostly, I only did a few cleanups and catchups) Tested on: i386 Compiled on: alpha, amd64
2004-02-04 21:52:57 +00:00
return (EFBIG);
Implement AUE_CORE, which adds process core dump support into the kernel. This change introduces audit_proc_coredump() which is called by coredump(9) to create an audit record for the coredump event. When a process dumps a core, it could be security relevant. It could be an indicator that a stack within the process has been overflowed with an incorrectly constructed malicious payload or a number of other events. The record that is generated looks like this: header,111,10,process dumped core,0,Thu Oct 25 19:36:29 2007, + 179 msec argument,0,0xb,signal path,/usr/home/csjp/test.core subject,csjp,csjp,staff,csjp,staff,1101,1095,50457,10.37.129.2 return,success,1 trailer,111 - We allocate a completely new record to make sure we arent clobbering the audit data associated with the syscall that produced the core (assuming the core is being generated in response to SIGABRT and not an invalid memory access). - Shuffle around expand_name() so we can use the coredump name at the very beginning of the coredump call. Make sure we free the storage referenced by "name" if we need to bail out early. - Audit both successful and failed coredump creation efforts Obtained from: TrustedBSD Project Reviewed by: rwatson MFC after: 1 month
2007-10-26 01:23:07 +00:00
}
Fix support for RACCT_CORE by merging forgotten file.
2011-05-26 18:54:07 +00:00
PROC_UNLOCK(p);
Bail out of the process early if the coredumpfile limit is 0. PR: kern/14540 Reviewed by: Nate Williams
1999-10-30 18:55:11 +00:00
Reimplement support for userland core dump compression using a new interface in kern_gzio.c. The old gzio interface was somewhat inflexible and has not worked properly since r272535: currently, the gzio functions are called with a range lock held on the output vnode, but kern_gzio.c does not pass the IO_RANGELOCKED flag to vn_rdwr() calls, resulting in deadlock when vn_rdwr() attempts to reacquire the range lock. Moreover, the new gzio interface can be used to implement kernel core compression. This change also modifies the kernel configuration options needed to enable userland core dump compression support: gzio is now an option rather than a device, and the COMPRESS_USER_CORES option is removed. Core dump compression is enabled using the kern.compress_user_cores sysctl/tunable. Differential Revision: https://reviews.freebsd.org/D1832 Reviewed by: rpaulo Discussed with: kib
2015-03-09 03:50:53 +00:00
error = corefile_open(p->p_comm, cred->cr_uid, p->p_pid, td,
Adds signal number format to kern.corefile Add format capability to core file names to include signal that generated the core. This can help various validation workflows where all cores should not be considered equally (SIGQUIT is often intentional and not an error unlike SIGSEGV or SIGBUS) Submitted by: David Leimbach (leimy2k@gmail.com) Reviewed by: markj MFC after: 1 week Relnotes: sysctl kern.corefile can now include the signal number Sponsored by: Dell EMC Isilon Differential Revision: https://reviews.freebsd.org/D20970
2019-07-16 15:51:09 +00:00
compress_user_cores, p->p_sig, &vp, &name);
Replace expand_name() function with corefile_open() function, which not only returns name, but also vnode of corefile to use. This simplifies the code and closes few races, especially in %I handling. Reviewed by: kib Obtained from: WHEEL Systems
2012-12-19 23:59:48 +00:00
if (error != 0)
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
return (error);
This brings in a Yahoo coredump patch from Paul, with additional mods by me (addition of vn_rdwr_inchunks). The problem Yahoo is solving is that if you have large process images core dumping, or you have a large number of forked processes all core dumping at the same time, the original coredump code would leave the vnode locked throughout. This can cause the directory vnode to get locked up, which can cause the parent directory vnode to get locked up, and so on all the way to the root node, locking the entire machine up for extremely long periods of time. This patch solves the problem in two ways. First it uses an advisory non-blocking lock to abort multiple processes trying to core to the same file. Second (my contribution) it chunks up the writes and uses bwillwrite() to avoid holding the vnode locked while blocking in the buffer cache. Submitted by: ps Reviewed by: dillon MFC after: 2 weeks
2001-09-08 20:02:33 +00:00
Fixes for i/o during coredumping: - Do not dump into system files. - Do not acquire write reference to the mount point where img.core is written, in the coredump(). The vn_rdwr() calls from ELF imgact request the write ref from vn_rdwr(). Recursive acqusition of the write ref deadlocks with the unmount. - Instead, take the range lock for the whole core file. This prevents parallel dumping from two processes executing the same image, converting the useless interleaved dump into sequential dumping, with second core overwriting the first. Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks
2014-10-04 18:35:00 +00:00
/*
* Don't dump to non-regular files or files with links.
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
* Do not dump into system files. Effective user must own the corefile.
Fixes for i/o during coredumping: - Do not dump into system files. - Do not acquire write reference to the mount point where img.core is written, in the coredump(). The vn_rdwr() calls from ELF imgact request the write ref from vn_rdwr(). Recursive acqusition of the write ref deadlocks with the unmount. - Instead, take the range lock for the whole core file. This prevents parallel dumping from two processes executing the same image, converting the useless interleaved dump into sequential dumping, with second core overwriting the first. Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks
2014-10-04 18:35:00 +00:00
*/
Minor style tweaks. Obtained from: WHEEL Systems
2012-12-17 10:51:22 +00:00
if (vp->v_type != VREG || VOP_GETATTR(vp, &vattr, cred) != 0 ||
coredump: avoid writing to core files not owned by the real user. Reported by: blake frantz <trew@hick.org> PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 18:35:11 +00:00
vattr.va_nlink != 1 || (vp->v_vflag & VV_SYSTEM) != 0 ||
Switch to check for effective user id in r349320, and disable dumping into existing files for sugid processes. Despite using real user id pronounces the intent, it actually breaks suid coredumps, while not making any difference for non-sugid processes. The reason for the breakage is that non-existent core file is created with the effective uid (unless weird hacks like SUIDDIR are configured). Then, if user enabled kern.sugid_coredump, core dumping should not overwrite core files owned by effective uid, but we cannot pretend to use real uid for dumping. PR: 68905 admbugs: 358 Sponsored by: The FreeBSD Foundation MFC after: 1 week
2019-06-23 21:15:31 +00:00
vattr.va_uid != cred->cr_uid) {
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(vp);
Rearrange the code so that it checks whether the file is something valid to write a core dump to before doing the preparations to actually write to the file. Call VOP_GETATTR() before dropping the initial vnode lock.
2002-07-10 06:31:35 +00:00
error = EFAULT;
Keep a reference on the coredump vnode for vn_fullpath() call. Do it by moving vn_close() after the point where notification is sent. Reported by: sbruno Tested by: pho, sbruno Sponsored by: The FreeBSD Foundation
2015-02-24 13:07:31 +00:00
goto out;
Rearrange the code so that it checks whether the file is something valid to write a core dump to before doing the preparations to actually write to the file. Call VOP_GETATTR() before dropping the initial vnode lock.
2002-07-10 06:31:35 +00:00
}
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(vp);
Fixes for i/o during coredumping: - Do not dump into system files. - Do not acquire write reference to the mount point where img.core is written, in the coredump(). The vn_rdwr() calls from ELF imgact request the write ref from vn_rdwr(). Recursive acqusition of the write ref deadlocks with the unmount. - Instead, take the range lock for the whole core file. This prevents parallel dumping from two processes executing the same image, converting the useless interleaved dump into sequential dumping, with second core overwriting the first. Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks
2014-10-04 18:35:00 +00:00
/* Postpone other writers, including core dumps of other processes. */
rl_cookie = vn_rangelock_wlock(vp, 0, OFF_MAX);
This brings in a Yahoo coredump patch from Paul, with additional mods by me (addition of vn_rdwr_inchunks). The problem Yahoo is solving is that if you have large process images core dumping, or you have a large number of forked processes all core dumping at the same time, the original coredump code would leave the vnode locked throughout. This can cause the directory vnode to get locked up, which can cause the parent directory vnode to get locked up, and so on all the way to the root node, locking the entire machine up for extremely long periods of time. This patch solves the problem in two ways. First it uses an advisory non-blocking lock to abort multiple processes trying to core to the same file. Second (my contribution) it chunks up the writes and uses bwillwrite() to avoid holding the vnode locked while blocking in the buffer cache. Submitted by: ps Reviewed by: dillon MFC after: 2 weeks
2001-09-08 20:02:33 +00:00
lf.l_whence = SEEK_SET;
lf.l_start = 0;
lf.l_len = 0;
lf.l_type = F_WRLCK;
When generate a core dump, use advisory locking in an advisory way: if we do acquire an advisory lock, great! We'll release it later. However, if we fail to acquire a lock, we perform the coredump anyway. This problem became particularly visible with NFS after the introduction of rpc.lockd: if the lock manager isn't running, then locking calls will fail, aborting the core dump (resulting in a zero-byte dump file). Reported by: Yogeshwar Shenoy <ynshenoy@alumni.cs.ucsb.edu>
2003-10-25 16:14:09 +00:00
locked = (VOP_ADVLOCK(vp, (caddr_t)p, F_SETLK, &lf, F_FLOCK) == 0);
This brings in a Yahoo coredump patch from Paul, with additional mods by me (addition of vn_rdwr_inchunks). The problem Yahoo is solving is that if you have large process images core dumping, or you have a large number of forked processes all core dumping at the same time, the original coredump code would leave the vnode locked throughout. This can cause the directory vnode to get locked up, which can cause the parent directory vnode to get locked up, and so on all the way to the root node, locking the entire machine up for extremely long periods of time. This patch solves the problem in two ways. First it uses an advisory non-blocking lock to abort multiple processes trying to core to the same file. Second (my contribution) it chunks up the writes and uses bwillwrite() to avoid holding the vnode locked while blocking in the buffer cache. Submitted by: ps Reviewed by: dillon MFC after: 2 weeks
2001-09-08 20:02:33 +00:00
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
VATTR_NULL(&vattr);
vattr.va_size = 0;
add option to automaticly mark core dumps with the nodump flag PR: 57065 Submitted by: Walter C. Pelissero
2004-08-09 05:46:46 +00:00
if (set_core_nodump_flag)
vattr.va_flags = UF_NODUMP;
vn_lock() is currently only used with the 'curthread' passed as argument. Remove this argument and pass curthread directly to underlying VOP_LOCK1() VFS method. This modify makes the code cleaner and in particular remove an annoying dependence helping next lockmgr() cleanup. KPI results, obviously, changed. Manpage and FreeBSD_version will be updated through further commits. As a side note, would be valuable to say that next commits will address a similar cleanup about VFS methods, in particular vop_lock1 and vop_unlock. Tested by: Diego Sardina <siarodx at gmail dot com>, Andrea Di Pasquale <whyx dot it at gmail dot com>
2008-01-10 01:10:58 +00:00
vn_lock(vp, LK_EXCLUSIVE | LK_RETRY);
Decontextualize the couplet VOP_GETATTR / VOP_SETATTR as the passed thread was always curthread and totally unuseful. Tested by: Giovanni Trematerra <giovanni dot trematerra at gmail dot com>
2008-08-28 15:23:18 +00:00
VOP_SETATTR(vp, &vattr, cred);
vfs: drop the mostly unused flags argument from VOP_UNLOCK Filesystems which want to use it in limited capacity can employ the VOP_UNLOCK_FLAGS macro. Reviewed by: kib (previous version) Differential Revision: https://reviews.freebsd.org/D21427
2020-01-03 22:29:58 +00:00
VOP_UNLOCK(vp);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
p->p_acflag |= ACORE;
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
if (p->p_sysent->sv_coredump != NULL) {
Generalize the gzio API. We currently use a set of subroutines in kern_gzio.c to perform compression of user and kernel core dumps. In the interest of adding support for other compression algorithms (zstd) in this role without complicating the API consumers, add a simple compressor API which can be used to select an algorithm. Also change the (non-default) GZIO kernel option to not enable compressed user cores by default. It's not clear that such a default would be desirable with support for multiple algorithms implemented, and it's inconsistent in that it isn't applied to kernel dumps. Reviewed by: cem Differential Revision: https://reviews.freebsd.org/D13632
2018-01-08 21:27:41 +00:00
error = p->p_sysent->sv_coredump(td, vp, limit, 0);
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
} else {
error = ENOSYS;
}
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
When generate a core dump, use advisory locking in an advisory way: if we do acquire an advisory lock, great! We'll release it later. However, if we fail to acquire a lock, we perform the coredump anyway. This problem became particularly visible with NFS after the introduction of rpc.lockd: if the lock manager isn't running, then locking calls will fail, aborting the core dump (resulting in a zero-byte dump file). Reported by: Yogeshwar Shenoy <ynshenoy@alumni.cs.ucsb.edu>
2003-10-25 16:14:09 +00:00
if (locked) {
lf.l_type = F_UNLCK;
VOP_ADVLOCK(vp, (caddr_t)p, F_UNLCK, &lf, F_FLOCK);
}
Fixes for i/o during coredumping: - Do not dump into system files. - Do not acquire write reference to the mount point where img.core is written, in the coredump(). The vn_rdwr() calls from ELF imgact request the write ref from vn_rdwr(). Recursive acqusition of the write ref deadlocks with the unmount. - Instead, take the range lock for the whole core file. This prevents parallel dumping from two processes executing the same image, converting the useless interleaved dump into sequential dumping, with second core overwriting the first. Tested by: pho Sponsored by: The FreeBSD Foundation MFC after: 2 weeks
2014-10-04 18:35:00 +00:00
vn_rangelock_unlock(vp, rl_cookie);
Keep a reference on the coredump vnode for vn_fullpath() call. Do it by moving vn_close() after the point where notification is sent. Reported by: sbruno Tested by: pho, sbruno Sponsored by: The FreeBSD Foundation
2015-02-24 13:07:31 +00:00
Notify devd(8) when a process crashed. This change implements a notification (via devctl) to userland when the kernel produces coredumps after a process has crashed. devd can then run a specific command to produce a human readable crash report. The command is most usually a helper that runs gdb/lldb commands on the file/coredump pair. It's possible to use this functionality for implementing automatic generation of crash reports. devd(8) will be notified of the full path of the binary that crashed and the full path of the coredump file.
2015-02-09 23:13:50 +00:00
/*
* Notify the userland helper that a process triggered a core dump.
* This allows the helper to run an automated debugging session.
*/
Keep a reference on the coredump vnode for vn_fullpath() call. Do it by moving vn_close() after the point where notification is sent. Reported by: sbruno Tested by: pho, sbruno Sponsored by: The FreeBSD Foundation
2015-02-24 13:07:31 +00:00
if (error != 0 || coredump_devctl == 0)
Notify devd(8) when a process crashed. This change implements a notification (via devctl) to userland when the kernel produces coredumps after a process has crashed. devd can then run a specific command to produce a human readable crash report. The command is most usually a helper that runs gdb/lldb commands on the file/coredump pair. It's possible to use this functionality for implementing automatic generation of crash reports. devd(8) will be notified of the full path of the binary that crashed and the full path of the coredump file.
2015-02-09 23:13:50 +00:00
goto out;
Fix devctl generation for core files. We have a problem with vn_fullpath_global when the file exists. Work around it by printing the full path if the core file name starts with /, or current working directory followed by the filename if not. Sponsored by: Netflix Differential Review: https://reviews.freebsd.org/D16026
2018-06-27 04:11:09 +00:00
sb = sbuf_new_auto();
cache: drop the always curthread argument from reverse lookup routines Note VOP_VPTOCNP keeps getting it as temporary compatibility for zfs. Tested by: pho
2020-08-24 08:57:02 +00:00
if (vn_fullpath_global(p->p_textvp, &fullpath, &freepath) != 0)
Fix devctl generation for core files. We have a problem with vn_fullpath_global when the file exists. Work around it by printing the full path if the core file name starts with /, or current working directory followed by the filename if not. Sponsored by: Netflix Differential Review: https://reviews.freebsd.org/D16026
2018-06-27 04:11:09 +00:00
goto out2;
sbuf_printf(sb, "comm=\"");
devctl_safe_quote_sb(sb, fullpath);
Sanitise the coredump file names sent to devd. While there, add a sysctl to turn this feature off as requested by kib@.
2015-02-10 04:34:39 +00:00
free(freepath, M_TEMP);
Fix devctl generation for core files. We have a problem with vn_fullpath_global when the file exists. Work around it by printing the full path if the core file name starts with /, or current working directory followed by the filename if not. Sponsored by: Netflix Differential Review: https://reviews.freebsd.org/D16026
2018-06-27 04:11:09 +00:00
sbuf_printf(sb, "\" core=\"");
/*
* We can't lookup core file vp directly. When we're replacing a core, and
* other random times, we flush the name cache, so it will fail. Instead,
* if the path of the core is relative, add the current dir in front if it.
*/
if (name[0] != '/') {
cache: replace kern___getcwd with vn_getcwd The previous routine was resulting in extra data copies most notably in linux_getcwd.
2020-02-01 20:38:38 +00:00
fullpathsize = MAXPATHLEN;
freepath = malloc(fullpathsize, M_TEMP, M_WAITOK);
cache: drop the always curthread argument from reverse lookup routines Note VOP_VPTOCNP keeps getting it as temporary compatibility for zfs. Tested by: pho
2020-08-24 08:57:02 +00:00
if (vn_getcwd(freepath, &fullpath, &fullpathsize) != 0) {
cache: replace kern___getcwd with vn_getcwd The previous routine was resulting in extra data copies most notably in linux_getcwd.
2020-02-01 20:38:38 +00:00
free(freepath, M_TEMP);
Fix devctl generation for core files. We have a problem with vn_fullpath_global when the file exists. Work around it by printing the full path if the core file name starts with /, or current working directory followed by the filename if not. Sponsored by: Netflix Differential Review: https://reviews.freebsd.org/D16026
2018-06-27 04:11:09 +00:00
goto out2;
}
devctl_safe_quote_sb(sb, fullpath);
cache: replace kern___getcwd with vn_getcwd The previous routine was resulting in extra data copies most notably in linux_getcwd.
2020-02-01 20:38:38 +00:00
free(freepath, M_TEMP);
Fix devctl generation for core files. We have a problem with vn_fullpath_global when the file exists. Work around it by printing the full path if the core file name starts with /, or current working directory followed by the filename if not. Sponsored by: Netflix Differential Review: https://reviews.freebsd.org/D16026
2018-06-27 04:11:09 +00:00
sbuf_putc(sb, '/');
}
devctl_safe_quote_sb(sb, name);
sbuf_printf(sb, "\"");
if (sbuf_finish(sb) == 0)
devctl_notify("kernel", "signal", "coredump", sbuf_data(sb));
out2:
sbuf_delete(sb);
Notify devd(8) when a process crashed. This change implements a notification (via devctl) to userland when the kernel produces coredumps after a process has crashed. devd can then run a specific command to produce a human readable crash report. The command is most usually a helper that runs gdb/lldb commands on the file/coredump pair. It's possible to use this functionality for implementing automatic generation of crash reports. devd(8) will be notified of the full path of the binary that crashed and the full path of the coredump file.
2015-02-09 23:13:50 +00:00
out:
Keep a reference on the coredump vnode for vn_fullpath() call. Do it by moving vn_close() after the point where notification is sent. Reported by: sbruno Tested by: pho, sbruno Sponsored by: The FreeBSD Foundation
2015-02-24 13:07:31 +00:00
error1 = vn_close(vp, FWRITE, cred, td);
if (error == 0)
error = error1;
Implement AUE_CORE, which adds process core dump support into the kernel. This change introduces audit_proc_coredump() which is called by coredump(9) to create an audit record for the coredump event. When a process dumps a core, it could be security relevant. It could be an indicator that a stack within the process has been overflowed with an incorrectly constructed malicious payload or a number of other events. The record that is generated looks like this: header,111,10,process dumped core,0,Thu Oct 25 19:36:29 2007, + 179 msec argument,0,0xb,signal path,/usr/home/csjp/test.core subject,csjp,csjp,staff,csjp,staff,1101,1095,50457,10.37.129.2 return,success,1 trailer,111 - We allocate a completely new record to make sure we arent clobbering the audit data associated with the syscall that produced the core (assuming the core is being generated in response to SIGABRT and not an invalid memory access). - Shuffle around expand_name() so we can use the coredump name at the very beginning of the coredump call. Make sure we free the storage referenced by "name" if we need to bail out early. - Audit both successful and failed coredump creation efforts Obtained from: TrustedBSD Project Reviewed by: rwatson MFC after: 1 month
2007-10-26 01:23:07 +00:00
#ifdef AUDIT
audit_proc_coredump(td, name, error);
#endif
free(name, M_TEMP);
General cleanup of core-dumping code. Submitted by: Sean Fagan,
1999-09-01 00:29:56 +00:00
return (error);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
Remove 'MPSAFE' annotations from the comments above most system calls: all system calls now enter without Giant held, and then in some cases, acquire Giant explicitly. Remove a number of other MPSAFE annotations in the credential code and tweak one or two other adjacent comments.
2007-03-04 22:36:48 +00:00
* Nonexistent system call-- signal process (may want to handle it). Flag
* error in case process won't see signal immediately (blocked or ignored).
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct nosys_args {
int dummy;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
nosys(struct thread *td, struct nosys_args *args)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Print unimplemented syscall number to the ctty on SIGSYS, if enabled by the knob kern.lognosys. Discussed with: imp Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 3 weeks X-Differential revision: https://reviews.freebsd.org/D11080
2017-06-12 21:11:11 +00:00
struct proc *p;
p = td->td_proc;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(p);
Deliver SIGSYS to the guilty thread, not to the process. MFC after: 1 week
2012-08-18 18:17:10 +00:00
tdsignal(td, SIGSYS);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
Make it possible to request nosys logging to console. New kern.lognosys values are 1 - log to ctty 2 - log to console 3 - log to both. Inspired by: eugen Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-07-27 20:45:41 +00:00
if (kern_lognosys == 1 || kern_lognosys == 3) {
Print unimplemented syscall number to the ctty on SIGSYS, if enabled by the knob kern.lognosys. Discussed with: imp Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 3 weeks X-Differential revision: https://reviews.freebsd.org/D11080
2017-06-12 21:11:11 +00:00
uprintf("pid %d comm %s: nosys %d\n", p->p_pid, p->p_comm,
td->td_sa.code);
Make it possible to request nosys logging to console. New kern.lognosys values are 1 - log to ctty 2 - log to console 3 - log to both. Inspired by: eugen Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-07-27 20:45:41 +00:00
}
Always report ENOSYS in init While rare, encountering an unimplemented system call early in init is catastrophic and difficult to debug. Even after a SIGSYS handler is registered, such configurations are problematic. As such, always report such events for pid 1 (following kern.lognosys if non-zero). Reviewed by: kevans, imp Obtained from: CheriBSD (plus suggestions from kevans) MFC after: 1 week Sponsored by: DARPA Differential Revision: https://reviews.freebsd.org/D26288
2020-09-02 23:17:33 +00:00
if (kern_lognosys == 2 || kern_lognosys == 3 ||
(p->p_pid == 1 && (kern_lognosys & 3) == 0)) {
Make it possible to request nosys logging to console. New kern.lognosys values are 1 - log to ctty 2 - log to console 3 - log to both. Inspired by: eugen Sponsored by: The FreeBSD Foundation MFC after: 1 week
2017-07-27 20:45:41 +00:00
printf("pid %d comm %s: nosys %d\n", p->p_pid, p->p_comm,
td->td_sa.code);
}
Return the correct error code (ENOSYS, not EINVAL) from nosys(). Getting killed by SIGSYS for unimlemented syscalls is bad enough. Obtained from: Lite2 branch The Lite2 branch has some other interesting unmerged (?) bits in this file. They are well hidden among cosmetic regressions.
2002-05-05 04:50:47 +00:00
return (ENOSYS);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
/*
Remove 'MPSAFE' annotations from the comments above most system calls: all system calls now enter without Giant held, and then in some cases, acquire Giant explicitly. Remove a number of other MPSAFE annotations in the credential code and tweak one or two other adjacent comments.
2007-03-04 22:36:48 +00:00
* Send a SIGIO or SIGURG signal to a process or process group using stored
* credentials rather than those of the current process.
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
*/
void
kern_sig.c: ANSIfy and remove archaic register keyword Sponsored by: The FreeBSD Foundation
2017-03-02 22:17:53 +00:00
pgsigio(struct sigio **sigiop, int sig, int checkctty)
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
{
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_t ksi;
Redo the sigio locking. Turn the sigio sx into a mutex. Sigio lock is really only needed to protect interrupts from dereferencing the sigio pointer in an object when the sigio itself is being destroyed. In order to do this in the most unintrusive manner change pgsigio's sigio * argument into a **, that way we can lock internally to the function.
2002-05-01 20:44:46 +00:00
struct sigio *sigio;
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
Among signal generation syscalls, only sigqueue(2) is allowed by POSIX to fail due to lack of resources to queue siginfo. Add KSI_SIGQ flag that allows sigqueue_add() to fail while trying to allocate memory for new siginfo. When the flag is not set, behaviour is the same as for KSI_TRAP: if memory cannot be allocated, set bit in sq_kill. KSI_TRAP is kept to preserve KBI. Add SI_KERNEL si_code, to be used in siginfo.si_code when signal is generated by kernel. Deliver siginfo when signal is generated by kill(2) family of syscalls (SI_USER with properly filled si_uid and si_pid), or by kernel (SI_KERNEL, mostly job control or SIGIO). Since KSI_SIGQ flag is not set for the ksi, low memory condition cause old behaviour. Keep psignal(9) KBI intact, but modify it to generate SI_KERNEL si_code. Pgsignal(9) and gsignal(9) now take ksi explicitely. Add pksignal(9) that behaves like psignal but takes ksi, and ddb kill command implemented as pksignal(..., ksi = NULL) to not do allocation while in debugger. While there, remove some register specifiers and use ANSI C prototypes. Reviewed by: davidxu MFC after: 1 month
2009-11-17 11:39:15 +00:00
ksiginfo_init(&ksi);
ksi.ksi_signo = sig;
ksi.ksi_code = SI_KERNEL;
Redo the sigio locking. Turn the sigio sx into a mutex. Sigio lock is really only needed to protect interrupts from dereferencing the sigio pointer in an object when the sigio itself is being destroyed. In order to do this in the most unintrusive manner change pgsigio's sigio * argument into a **, that way we can lock internally to the function.
2002-05-01 20:44:46 +00:00
SIGIO_LOCK();
sigio = *sigiop;
if (sigio == NULL) {
SIGIO_UNLOCK();
return;
}
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
if (sigio->sio_pgid > 0) {
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_LOCK(sigio->sio_proc);
o Revert kern_sig.c#1.143, as cr_cansignal() doesn't currently permit a number of desirable cases in which SIGIO/SIGURG are delivered. We'll keep tweaking. Reported by: Alexander Kabaev <ak03@gte.com>
2002-01-10 01:25:35 +00:00
if (CANSIGIO(sigio->sio_ucred, sigio->sio_proc->p_ucred))
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
kern_psignal(sigio->sio_proc, sig);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(sigio->sio_proc);
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
} else if (sigio->sio_pgid < 0) {
struct proc *p;
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
PGRP_LOCK(sigio->sio_pgrp);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
LIST_FOREACH(p, &sigio->sio_pgrp->pg_members, p_pglist) {
PROC_LOCK(p);
Fix several places to ignore processes that are not yet fully constructed. MFC after: 1 week
2011-04-06 17:47:22 +00:00
if (p->p_state == PRS_NORMAL &&
CANSIGIO(sigio->sio_ucred, p->p_ucred) &&
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
(checkctty == 0 || (p->p_flag & P_CONTROLT)))
In order to maximize the re-usability of kernel code in user space this patch modifies makesyscalls.sh to prefix all of the non-compatibility calls (e.g. not linux_, freebsd32_) with sys_ and updates the kernel entry points and all places in the code that use them. It also fixes an additional name space collision between the kernel function psignal and the libc function of the same name by renaming the kernel psignal kern_psignal(). By introducing this change now we will ease future MFCs that change syscalls. Reviewed by: rwatson Approved by: re (bz)
2011-09-16 13:58:51 +00:00
kern_psignal(p, sig);
- Proc locking. Most of signal handling is now MP safe and doesn't require Giant. The only exception is the CANSIGNAL() macro. Unlocking the proc lock around sendsig() in trapsignal() is also questionable. Note that the functions sigexit(), psignal(), and issignal() must be called with the proc lock of the process in question held. postsig() and trapsignal() should not be called with the proc lock held, but they also do not require Giant anymore either. - Remove spl's that are now no longer needed as they are fully replaced.
2001-03-07 02:59:54 +00:00
PROC_UNLOCK(p);
}
Lock struct pgrp, session and sigio. New locks are: - pgrpsess_lock which locks the whole pgrps and sessions, - pg_mtx which protects the pgrp members, and - s_mtx which protects the session members. Please refer to sys/proc.h for the coverage of these locks. Changes on the pgrp/session interface: - pgfind() needs the pgrpsess_lock held. - The caller of enterpgrp() is responsible to allocate a new pgrp and session. - Call enterthispgrp() in order to enter an existing pgrp. - pgsignal() requires a pgrp lock held. Reviewed by: jhb, alfred Tested on: cvsup.jp.FreeBSD.org (which is a quad-CPU machine running -current)
2002-02-23 11:12:57 +00:00
PGRP_UNLOCK(sigio->sio_pgrp);
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
}
Redo the sigio locking. Turn the sigio sx into a mutex. Sigio lock is really only needed to protect interrupts from dereferencing the sigio pointer in an object when the sigio itself is being destroyed. In order to do this in the most unintrusive manner change pgsigio's sigio * argument into a **, that way we can lock internally to the function.
2002-05-01 20:44:46 +00:00
SIGIO_UNLOCK();
Installed the second patch attached to kern/7899 with some changes suggested by bde, a few other tweaks to get the patch to apply cleanly again and some improvements to the comments. This change closes some fairly minor security holes associated with F_SETOWN, fixes a few bugs, and removes some limitations that F_SETOWN had on tty devices. For more details, see the description on the PR. Because this patch increases the size of the proc and pgrp structures, it is necessary to re-install the includes and recompile libkvm, the vinum lkm, fstat, gcore, gdb, ipfilter, ps, top, and w. PR: kern/7899 Reviewed by: bde, elvind
1998-11-11 10:04:13 +00:00
}
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
static int
filt_sigattach(struct knote *kn)
{
struct proc *p = curproc;
kn->kn_ptr.p_proc = p;
kn->kn_flags |= EV_CLEAR; /* automatically set */
When filt_proc() removes event from the knlist due to the process exiting (NOTE_EXIT->knlist_remove_inevent()), two things happen: - knote kn_knlist pointer is reset - INFLUX knote is removed from the process knlist. And, there are two consequences: - KN_LIST_UNLOCK() on such knote is nop - there is nothing which would block exit1() from processing past the knlist_destroy() (and knlist_destroy() resets knlist lock pointers). Both consequences result either in leaked process lock, or dereferencing NULL function pointers for locking. Handle this by stopping embedding the process knlist into struct proc. Instead, the knlist is allocated together with struct proc, but marked as autodestroy on the zombie reap, by knlist_detach() function. The knlist is freed when last kevent is removed from the list, in particular, at the zombie reap time if the list is empty. As result, the knlist_remove_inevent() is no longer needed and removed. Other changes: In filt_procattach(), clear NOTE_EXEC and NOTE_FORK desired events from kn_sfflags for knote registered by kernel to only get NOTE_CHILD notifications. The flags leak resulted in excessive NOTE_EXEC/NOTE_FORK reports. Fix immediate note activation in filt_procattach(). Condition should be either the immediate CHILD_NOTE activation, or immediate NOTE_EXIT report for the exiting process. In knote_fork(), do not perform racy check for KN_INFLUX before kq lock is taken. Besides being racy, it did not accounted for notes just added by scan (KN_SCAN). Some minor and incomplete style fixes. Analyzed and tested by: Eric Badger <eric@badgerio.us> Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb) Differential revision: https://reviews.freebsd.org/D6859
2016-06-27 21:52:17 +00:00
knlist_add(p->p_klist, kn, 0);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
return (0);
}
static void
filt_sigdetach(struct knote *kn)
{
struct proc *p = kn->kn_ptr.p_proc;
When filt_proc() removes event from the knlist due to the process exiting (NOTE_EXIT->knlist_remove_inevent()), two things happen: - knote kn_knlist pointer is reset - INFLUX knote is removed from the process knlist. And, there are two consequences: - KN_LIST_UNLOCK() on such knote is nop - there is nothing which would block exit1() from processing past the knlist_destroy() (and knlist_destroy() resets knlist lock pointers). Both consequences result either in leaked process lock, or dereferencing NULL function pointers for locking. Handle this by stopping embedding the process knlist into struct proc. Instead, the knlist is allocated together with struct proc, but marked as autodestroy on the zombie reap, by knlist_detach() function. The knlist is freed when last kevent is removed from the list, in particular, at the zombie reap time if the list is empty. As result, the knlist_remove_inevent() is no longer needed and removed. Other changes: In filt_procattach(), clear NOTE_EXEC and NOTE_FORK desired events from kn_sfflags for knote registered by kernel to only get NOTE_CHILD notifications. The flags leak resulted in excessive NOTE_EXEC/NOTE_FORK reports. Fix immediate note activation in filt_procattach(). Condition should be either the immediate CHILD_NOTE activation, or immediate NOTE_EXIT report for the exiting process. In knote_fork(), do not perform racy check for KN_INFLUX before kq lock is taken. Besides being racy, it did not accounted for notes just added by scan (KN_SCAN). Some minor and incomplete style fixes. Analyzed and tested by: Eric Badger <eric@badgerio.us> Reviewed by: jhb Sponsored by: The FreeBSD Foundation MFC after: 2 weeks Approved by: re (gjb) Differential revision: https://reviews.freebsd.org/D6859
2016-06-27 21:52:17 +00:00
knlist_remove(p->p_klist, kn, 0);
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
}
/*
Style fixes (mostly whitespaces).
2012-11-27 10:11:54 +00:00
* signal knotes are shared with proc knotes, so we apply a mask to
Introduce kqueue() and kevent(), a kernel event notification facility.
2000-04-16 18:53:38 +00:00
* the hint in order to differentiate them from process hints. This
* could be avoided by using a signal-specific knote list, but probably
* isn't worth the trouble.
*/
static int
filt_signal(struct knote *kn, long hint)
{
if (hint & NOTE_SIGNAL) {
hint &= ~NOTE_SIGNAL;
if (kn->kn_id == hint)
kn->kn_data++;
}
return (kn->kn_data != 0);
}
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
struct sigacts *
sigacts_alloc(void)
{
struct sigacts *ps;
ps = malloc(sizeof(struct sigacts), M_SUBPROC, M_WAITOK | M_ZERO);
Use refcount_init in sigacts_alloc. This change is a no-op, but fixes up an inconsistency introduced with r268634. MFC after: 3 days
2014-08-24 09:24:37 +00:00
refcount_init(&ps->ps_refcnt, 1);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
mtx_init(&ps->ps_mtx, "sigacts", NULL, MTX_DEF);
return (ps);
}
void
sigacts_free(struct sigacts *ps)
{
Manage struct sigacts refcnt with atomics instead of a mutex. MFC after: 1 week
2014-07-14 21:12:59 +00:00
if (refcount_release(&ps->ps_refcnt) == 0)
return;
mtx_destroy(&ps->ps_mtx);
free(ps, M_SUBPROC);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
}
struct sigacts *
sigacts_hold(struct sigacts *ps)
{
Manage struct sigacts refcnt with atomics instead of a mutex. MFC after: 1 week
2014-07-14 21:12:59 +00:00
refcount_acquire(&ps->ps_refcnt);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
return (ps);
}
void
sigacts_copy(struct sigacts *dest, struct sigacts *src)
{
KASSERT(dest->ps_refcnt == 1, ("sigacts_copy to shared dest"));
mtx_lock(&src->ps_mtx);
bcopy(src, dest, offsetof(struct sigacts, ps_refcnt));
mtx_unlock(&src->ps_mtx);
}
int
sigacts_shared(struct sigacts *ps)
{
Perform a lockless check in sigacts_shared. It is used only during execve (i.e. singlethreaded), so there is no fear of returning 'not shared' which soon becomes 'shared'. While here reorganize the code a little to avoid proc lock/unlock in shared case. MFC after: 1 week
2014-07-01 06:29:15 +00:00
return (ps->ps_refcnt > 1);
- Merge struct procsig with struct sigacts. - Move struct sigacts out of the u-area and malloc() it using the M_SUBPROC malloc bucket. - Add a small sigacts_*() API for managing sigacts structures: sigacts_alloc(), sigacts_free(), sigacts_copy(), sigacts_share(), and sigacts_shared(). - Remove the p_sigignore, p_sigacts, and p_sigcatch macros. - Add a mutex to struct sigacts that protects all the members of the struct. - Add sigacts locking. - Remove Giant from nosys(), kill(), killpg(), and kern_sigaction() now that sigacts is locked. - Several in-kernel functions such as psignal(), tdsignal(), trapsignal(), and thread_stopped() are now MP safe. Reviewed by: arch@ Approved by: re (rwatson)
2003-05-13 20:36:02 +00:00
}
rfork(2): add RFSPAWN flag When RFSPAWN is passed, rfork exhibits vfork(2) semantics but also resets signal handlers in the child during creation to avoid a point of corruption of parent state from the child. This flag will be used by posix_spawn(3) to handle potential signal issues. Reviewed by: jilles, kib Differential Revision: https://reviews.freebsd.org/D19058
2019-09-25 19:20:41 +00:00
void
sig_drop_caught(struct proc *p)
{
int sig;
struct sigacts *ps;
ps = p->p_sigacts;
PROC_LOCK_ASSERT(p, MA_OWNED);
mtx_assert(&ps->ps_mtx, MA_OWNED);
while (SIGNOTEMPTY(ps->ps_sigcatch)) {
sig = sig_ffs(&ps->ps_sigcatch);
sigdflt(ps, sig);
if ((sigprop(sig) & SIGPROP_IGNORE) != 0)
sigqueue_delete_proc(p, sig);
}
}
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
static void
sigfastblock_failed(struct thread *td, bool sendsig, bool write)
{
ksiginfo_t ksi;
/*
* Prevent further fetches and SIGSEGVs, allowing thread to
* issue syscalls despite corruption.
*/
sigfastblock_clear(td);
if (!sendsig)
return;
ksiginfo_init_trap(&ksi);
ksi.ksi_signo = SIGSEGV;
ksi.ksi_code = write ? SEGV_ACCERR : SEGV_MAPERR;
ksi.ksi_addr = td->td_sigblock_ptr;
trapsignal(td, &ksi);
}
static bool
sigfastblock_fetch_sig(struct thread *td, bool sendsig, uint32_t *valp)
{
uint32_t res;
if ((td->td_pflags & TDP_SIGFASTBLOCK) == 0)
return (true);
if (fueword32((void *)td->td_sigblock_ptr, &res) == -1) {
sigfastblock_failed(td, sendsig, false);
return (false);
}
*valp = res;
td->td_sigblock_val = res & ~SIGFASTBLOCK_FLAGS;
return (true);
}
sigfastblock: fix delivery of the pending signals in single-threaded processes. If single-threaded process receives a signal during critical section established by sigfastblock(2) word, unblock did not caused signal delivery because sigfastblock(SIGFASTBLOCK_UNBLOCK) failed to request ast handling of the pending signals. Set TDF_ASTPENDING | TDF_NEEDSIGCHK on unblock or when kernel forces end of sigfastblock critical section, to cause syscall exit to recheck and deliver any signal pending. Reported by: corydoras@ridiculousfish.com PR: 246385 Sponsored by: The FreeBSD Foundation
2020-05-11 22:38:32 +00:00
static void
sigfastblock_resched(struct thread *td, bool resched)
{
struct proc *p;
if (resched) {
p = td->td_proc;
PROC_LOCK(p);
reschedule_signals(p, td->td_sigmask, 0);
PROC_UNLOCK(p);
}
thread_lock(td);
td->td_flags |= TDF_ASTPENDING | TDF_NEEDSIGCHK;
thread_unlock(td);
}
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
int
sys_sigfastblock(struct thread *td, struct sigfastblock_args *uap)
{
struct proc *p;
int error, res;
uint32_t oldval;
error = 0;
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
p = td->td_proc;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
switch (uap->cmd) {
case SIGFASTBLOCK_SETPTR:
if ((td->td_pflags & TDP_SIGFASTBLOCK) != 0) {
error = EBUSY;
break;
}
if (((uintptr_t)(uap->ptr) & (sizeof(uint32_t) - 1)) != 0) {
error = EINVAL;
break;
}
td->td_pflags |= TDP_SIGFASTBLOCK;
td->td_sigblock_ptr = uap->ptr;
break;
case SIGFASTBLOCK_UNBLOCK:
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
if ((td->td_pflags & TDP_SIGFASTBLOCK) == 0) {
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
error = EINVAL;
break;
}
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
for (;;) {
res = casueword32(td->td_sigblock_ptr,
SIGFASTBLOCK_PEND, &oldval, 0);
if (res == -1) {
error = EFAULT;
sigfastblock_failed(td, false, true);
break;
}
if (res == 0)
break;
MPASS(res == 1);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
if (oldval != SIGFASTBLOCK_PEND) {
error = EBUSY;
break;
}
error = thread_check_susp(td, false);
if (error != 0)
break;
}
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
if (error != 0)
break;
/*
* td_sigblock_val is cleared there, but not on a
* syscall exit. The end effect is that a single
* interruptible sleep, while user sigblock word is
* set, might return EINTR or ERESTART to usermode
* without delivering signal. All further sleeps,
* until userspace clears the word and does
* sigfastblock(UNBLOCK), observe current word and no
* longer get interrupted. It is slight
* non-conformance, with alternative to have read the
* sigblock word on each syscall entry.
*/
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
td->td_sigblock_val = 0;
/*
* Rely on normal ast mechanism to deliver pending
* signals to current thread. But notify others about
* fake unblock.
*/
sigfastblock: fix delivery of the pending signals in single-threaded processes. If single-threaded process receives a signal during critical section established by sigfastblock(2) word, unblock did not caused signal delivery because sigfastblock(SIGFASTBLOCK_UNBLOCK) failed to request ast handling of the pending signals. Set TDF_ASTPENDING | TDF_NEEDSIGCHK on unblock or when kernel forces end of sigfastblock critical section, to cause syscall exit to recheck and deliver any signal pending. Reported by: corydoras@ridiculousfish.com PR: 246385 Sponsored by: The FreeBSD Foundation
2020-05-11 22:38:32 +00:00
sigfastblock_resched(td, error == 0 && p->p_numthreads != 1);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
break;
case SIGFASTBLOCK_UNSETPTR:
if ((td->td_pflags & TDP_SIGFASTBLOCK) == 0) {
error = EINVAL;
break;
}
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
if (!sigfastblock_fetch_sig(td, false, &oldval)) {
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
error = EFAULT;
break;
}
if (oldval != 0 && oldval != SIGFASTBLOCK_PEND) {
error = EBUSY;
break;
}
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
sigfastblock_clear(td);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
break;
default:
error = EINVAL;
break;
}
return (error);
}
void
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
sigfastblock_clear(struct thread *td)
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
{
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
bool resched;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
if ((td->td_pflags & TDP_SIGFASTBLOCK) == 0)
return;
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
td->td_sigblock_val = 0;
Fix signal delivery might be on sigfastblock clearing. When clearing sigfastblock, either by sigfastblock(UNSETPTR) call or implicitly on execve(2), kernel must check for pending signals and reschedule them if needed. E.g. on execve, all other threads are terminated, and current thread fast block pointer is cleaned. If any signal was left pending, it can now be delivered to the current thread, and we should prepare for ast() on return to userspace to notice the signals. Reported and tested by: pho Sponsored by: The FreeBSD Foundation
2020-03-10 20:25:03 +00:00
resched = (td->td_pflags & TDP_SIGFASTPENDING) != 0 ||
SIGPENDING(td);
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
td->td_pflags &= ~(TDP_SIGFASTBLOCK | TDP_SIGFASTPENDING);
sigfastblock: fix delivery of the pending signals in single-threaded processes. If single-threaded process receives a signal during critical section established by sigfastblock(2) word, unblock did not caused signal delivery because sigfastblock(SIGFASTBLOCK_UNBLOCK) failed to request ast handling of the pending signals. Set TDF_ASTPENDING | TDF_NEEDSIGCHK on unblock or when kernel forces end of sigfastblock critical section, to cause syscall exit to recheck and deliver any signal pending. Reported by: corydoras@ridiculousfish.com PR: 246385 Sponsored by: The FreeBSD Foundation
2020-05-11 22:38:32 +00:00
sigfastblock_resched(td, resched);
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
}
void
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
sigfastblock_fetch(struct thread *td)
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
{
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
uint32_t val;
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
(void)sigfastblock_fetch_sig(td, true, &val);
}
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
Return reschedule_signals() to being static again. It was used after sigfastblock_setpend() call in in ast() when current thread fast-blocks signals. Add a flag to sigfastblock_setpend() to request reschedule, and remove the direct use of the function from subr_trap.c Tested by: pho Sponsored by: The FreeBSD Foundation
2020-03-10 20:04:38 +00:00
static void
sigfastblock_setpend1(struct thread *td)
Do not read sigfastblock word on syscall entry. On machines with SMAP, fueword executes two serializing instructions which can be seen in microbenchmarks. As a measure to restore microbenchmark numbers, only read the word on the attempt to deliver signal in ast(). If the word is set, signal is not delivered and word is kept, preventing interruption of interruptible sleeps by signals until userspace calls sigfastblock(UNBLOCK) which clears the word. This way, the spurious EINTR that userspace can see while in critical section is on first interruptible sleep, if a signal is pending, and on signal posting. It is believed that it is not important for rtld and lbithr critical sections. It might be visible for the application code e.g. for the callback of dl_iterate_phdr(3), but again the belief is that the non-compliance is acceptable. Most important is that the retry of the sleeping syscall does not interrupt unless additional signal is posted. For now I added the knob kern.sigfastblock_fetch_always to enable the word read on syscall entry to be able to diagnose possible issues due to spurious EINTR. While there, do some code restructuting to have all sigfastblock() handling located in kern_sig.c. Reviewed by: jeff Discussed with: mjg Tested by: pho Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D23622
2020-02-20 15:34:02 +00:00
{
int res;
uint32_t oldval;
if ((td->td_pflags & TDP_SIGFASTBLOCK) == 0)
return;
res = fueword32((void *)td->td_sigblock_ptr, &oldval);
if (res == -1) {
sigfastblock_failed(td, true, false);
return;
}
for (;;) {
res = casueword32(td->td_sigblock_ptr, oldval, &oldval,
oldval | SIGFASTBLOCK_PEND);
if (res == -1) {
sigfastblock_failed(td, true, true);
return;
}
if (res == 0) {
td->td_sigblock_val = oldval & ~SIGFASTBLOCK_FLAGS;
td->td_pflags &= ~TDP_SIGFASTPENDING;
break;
}
MPASS(res == 1);
if (thread_check_susp(td, false) != 0)
break;
}
Add a way to manage thread signal mask using shared word, instead of syscall. A new syscall sigfastblock(2) is added which registers a uint32_t variable as containing the count of blocks for signal delivery. Its content is read by kernel on each syscall entry and on AST processing, non-zero count of blocks is interpreted same as the signal mask blocking all signals. The biggest downside of the feature that I see is that memory corruption that affects the registered fast sigblock location, would cause quite strange application misbehavior. For instance, the process would be immune to ^C (but killable by SIGKILL). With consumers (rtld and libthr added), benchmarks do not show a slow-down of the syscalls in micro-measurements, and macro benchmarks like buildworld do not demonstrate a difference. Part of the reason is that buildworld time is dominated by compiler, and clang already links to libthr. On the other hand, small utilities typically used by shell scripts have the total number of syscalls cut by half. The syscall is not exported from the stable libc version namespace on purpose. It is intended to be used only by our C runtime implementation internals. Tested by: pho Disscussed with: cem, emaste, jilles Sponsored by: The FreeBSD Foundation Differential revision: https://reviews.freebsd.org/D12773
2020-02-09 11:53:12 +00:00
}
Return reschedule_signals() to being static again. It was used after sigfastblock_setpend() call in in ast() when current thread fast-blocks signals. Add a flag to sigfastblock_setpend() to request reschedule, and remove the direct use of the function from subr_trap.c Tested by: pho Sponsored by: The FreeBSD Foundation
2020-03-10 20:04:38 +00:00
void
sigfastblock_setpend(struct thread *td, bool resched)
{
struct proc *p;
sigfastblock_setpend1(td);
if (resched) {
p = td->td_proc;
PROC_LOCK(p);
reschedule_signals(p, fastblock_mask, SIGPROCMASK_FASTBLK);
PROC_UNLOCK(p);
}
}
Reference in New Issue Copy Permalink