d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
73ba600faa
freebsd-nq/sys/kern/kern_kse.c

1446 lines
35 KiB
C
Raw Normal View History

/* -> /*- for copyright notices, minor format tweaks as necessary
2005-01-06 23:35:40 +00:00
/*-
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
* Copyright (C) 2001 Julian Elischer <julian@freebsd.org>.
* All rights reserved.
*
* 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(s), this list of conditions and the following disclaimer as
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
* the first lines of this file unmodified other than the possible
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
* addition of one or more copyright notices.
* 2. Redistributions in binary form must reproduce the above copyright
* notice(s), this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``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 COPYRIGHT HOLDER(S) 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.
*/
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
o Split out kernel part of execve(2) syscall into two parts: one that copies arguments into the kernel space and one that operates completely in the kernel space; o use kernel-only version of execve(2) to kill another stackgap in linuxlator/i386. Obtained from: DragonFlyBSD (partially) MFC after: 2 weeks
2005-01-29 23:12:00 +00:00
#include <sys/imgact.h>
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/proc.h>
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 07:33:40 +00:00
#include <sys/ptrace.h>
include smp.h. it is required by some code that was commented out until david's last commit.
2002-11-17 23:26:42 +00:00
#include <sys/smp.h>
Add an execve command for kse_thr_interrupt to allow libpthread to restore signal mask correctly, this is required by POSIX. Reviewed by: deischen
2004-10-07 13:50:10 +00:00
#include <sys/syscallsubr.h>
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
#include <sys/sysproto.h>
- Implement a mechanism for allowing schedulers to place scheduler dependant data in the scheduler independant structures (proc, ksegrp, kse, thread). - Implement unused stubs for this mechanism in sched_4bsd. Approved by: re Reviewed by: luigi, trb Tested on: x86, alpha
2002-11-21 01:22:38 +00:00
#include <sys/sched.h>
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
#include <sys/signalvar.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>
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
#include <sys/syslog.h>
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
#include <sys/kse.h>
#include <sys/ktr.h>
#include <vm/uma.h>
ia64 does not have the same degree of stealth include file nesting, so it needs an explicit #include <machine/frame.h> to get 'struct trapframe'. The fact that it needs this at this level is rather bogus but it will not compile without it.
2002-07-17 23:43:55 +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
#ifdef KSE
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
static uma_zone_t upcall_zone;
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
Allocate KSEs and KSEGRPs separatly and remove them from the proc structure. next step is to allow > 1 to be allocated per process. This would give multi-processor threads. (when the rest of the infrastructure is in place) While doing this I noticed libkvm and sys/kern/kern_proc.c:fill_kinfo_proc are diverging more than they should.. corrective action needed soon.
2002-09-15 23:52:25 +00:00
/* DEBUG ONLY */
Move the KSE ABI specific code here and separate it from code that is generic to any threading system. This commit does not link this file to the build yet, nor does it remove these functions from their current location in kern_thread.c. (that commit coming up after further review)
2004-06-07 07:25:03 +00:00
extern int virtual_cpu;
extern int thread_debug;
Shuffle some code around.
2004-06-11 17:48:20 +00:00
Move the KSE ABI specific code here and separate it from code that is generic to any threading system. This commit does not link this file to the build yet, nor does it remove these functions from their current location in kern_thread.c. (that commit coming up after further review)
2004-06-07 07:25:03 +00:00
extern int max_threads_per_proc;
extern int max_groups_per_proc;
extern int max_threads_hits;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
extern struct mtx kse_lock;
Allocate KSEs and KSEGRPs separatly and remove them from the proc structure. next step is to allow > 1 to be allocated per process. This would give multi-processor threads. (when the rest of the infrastructure is in place) While doing this I noticed libkvm and sys/kern/kern_proc.c:fill_kinfo_proc are diverging more than they should.. corrective action needed soon.
2002-09-15 23:52:25 +00:00
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
TAILQ_HEAD(, kse_upcall) zombie_upcalls =
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
TAILQ_HEAD_INITIALIZER(zombie_upcalls);
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
static int thread_update_usr_ticks(struct thread *td);
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
static int thread_alloc_spare(struct thread *td);
- Garbage collect unused concurrency functions. - Remove unused kse fields from struct proc. - Group remaining fields and #ifdef KSE them. - Move some kern_kse.c only prototypes out of proc and into kern_kse. Discussed with: Julian
2007-06-12 19:49:39 +00:00
static struct thread *thread_schedule_upcall(struct thread *td, struct kse_upcall *ku);
static struct kse_upcall *upcall_alloc(void);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
struct mtx kse_lock;
MTX_SYSINIT(kse_lock, &kse_lock, "kse lock", MTX_SPIN);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
struct kse_upcall *
upcall_alloc(void)
{
struct kse_upcall *ku;
Drop bzero and shove the responsibility of zeroing the kse upcall object on to the zone allocator. It should be noted that uma_zalloc(9) uses bzero to zero out the object so there probably wont be any real performance benefit. If UMA grows the ability to supply zeroed zones more efficiently in the future, we will not have to modify all the existing consumers. Discussed with: rwatson,julian MFC after: 1 week
2005-02-24 00:05:50 +00:00
ku = uma_zalloc(upcall_zone, M_WAITOK | M_ZERO);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
return (ku);
}
Actually, upcalls cannot be freed while destroying the thread because we should call uma_zfree() with various spinlock helds. Rearranging the code would not help here because we cannot break atomicity respect prcess spinlock, so the only one choice we have is to defer the operation. In order to do this use a global queue synchronized through the kse_lock spinlock which is freed at any thread_alloc() / thread_wait() through a call to thread_reap(). Note that this approach is not ideal as we should want a per-process list of zombie upcalls, but it follows initial guidelines of KSE authors. Tested by: jkim, pav Approved by: jeff, julian Approved by: re
2007-07-27 09:21:18 +00:00
void
upcall_reap(void)
{
TAILQ_HEAD(, kse_upcall) zupcalls;
struct kse_upcall *ku_item, *ku_tmp;
TAILQ_INIT(&zupcalls);
mtx_lock_spin(&kse_lock);
if (!TAILQ_EMPTY(&zombie_upcalls)) {
TAILQ_CONCAT(&zupcalls, &zombie_upcalls, ku_link);
TAILQ_INIT(&zombie_upcalls);
}
mtx_unlock_spin(&kse_lock);
TAILQ_FOREACH_SAFE(ku_item, &zupcalls, ku_link, ku_tmp)
uma_zfree(upcall_zone, ku_item);
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
void
upcall_remove(struct thread *td)
{
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK_ASSERT(td->td_proc, MA_OWNED);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Various small style fixes.
2004-09-22 15:24:33 +00:00
if (td->td_upcall != NULL) {
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
/*
* If we are not a bound thread then decrement the count of
* possible upcall sources
*/
if (td->td_pflags & TDP_SA)
td->td_proc->p_numupcalls--;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
mtx_lock_spin(&kse_lock);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td->td_upcall->ku_owner = NULL;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
TAILQ_REMOVE(&td->td_upcall->ku_proc->p_upcalls, td->td_upcall,
ku_link);
TAILQ_INSERT_HEAD(&zombie_upcalls, td->td_upcall, ku_link);
mtx_unlock_spin(&kse_lock);
Various small style fixes.
2004-09-22 15:24:33 +00:00
td->td_upcall = NULL;
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +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
#endif
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
#ifndef _SYS_SYSPROTO_H_
struct kse_switchin_args {
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
struct kse_thr_mailbox *tmbx;
int flags;
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
};
#endif
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
#ifdef KSE
void
kse_unlink(struct thread *td)
{
mtx_lock_spin(&kse_lock);
thread_unlink(td);
mtx_unlock_spin(&kse_lock);
- Garbage collect unused concurrency functions. - Remove unused kse fields from struct proc. - Group remaining fields and #ifdef KSE them. - Move some kern_kse.c only prototypes out of proc and into kern_kse. Discussed with: Julian
2007-06-12 19:49:39 +00:00
upcall_remove(td);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
}
#endif
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
int
kse_switchin(struct thread *td, struct kse_switchin_args *uap)
{
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
#ifdef KSE
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
struct kse_thr_mailbox tmbx;
struct kse_upcall *ku;
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
int error;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
if ((ku = td->td_upcall) == NULL || TD_CAN_UNBIND(td)) {
thread_unlock(td);
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
return (EINVAL);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
}
thread_unlock(td);
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
error = (uap->tmbx == NULL) ? EINVAL : 0;
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
if (!error)
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
error = copyin(uap->tmbx, &tmbx, sizeof(tmbx));
if (!error && (uap->flags & KSE_SWITCHIN_SETTMBX))
error = (suword(&ku->ku_mailbox->km_curthread,
(long)uap->tmbx) != 0 ? EINVAL : 0);
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
if (!error)
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
error = set_mcontext(td, &tmbx.tm_context.uc_mcontext);
if (!error) {
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 07:33:40 +00:00
suword32(&uap->tmbx->tm_lwp, td->td_tid);
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
if (uap->flags & KSE_SWITCHIN_SETTMBX) {
td->td_mailbox = uap->tmbx;
Move TDF_CAN_UNBIND to thread private flags td_pflags, this eliminates need of sched_lock in some places. Also in thread_userret, remove spare thread allocation code, it is already done in thread_user_enter. Reviewed by: julian
2004-08-28 04:08:05 +00:00
td->td_pflags |= TDP_CAN_UNBIND;
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
}
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
PROC_LOCK(td->td_proc);
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 07:33:40 +00:00
if (td->td_proc->p_flag & P_TRACED) {
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
_PHOLD(td->td_proc);
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 07:33:40 +00:00
if (tmbx.tm_dflags & TMDF_SSTEP)
ptrace_single_step(td);
else
ptrace_clear_single_step(td);
s/TMDF_DONOTRUNUSER/TMDF_SUSPEND/g Dicussed with: deischen
2004-08-03 02:23:06 +00:00
if (tmbx.tm_dflags & TMDF_SUSPEND) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
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 07:33:40 +00:00
/* fuword can block, check again */
if (td->td_upcall)
ku->ku_flags |= KUF_DOUPCALL;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
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 07:33:40 +00:00
}
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
_PRELE(td->td_proc);
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 07:33:40 +00:00
}
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
PROC_UNLOCK(td->td_proc);
Change kse_switchin to accept kse_thr_mailbox pointer, the syscall will be used heavily in debugging KSE threads. This breaks libpthread on IA64, but because libpthread was not in 5.2.1 release, I would like to change it so we needn't to introduce another syscall.
2004-07-12 07:39:20 +00:00
}
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
return ((error == 0) ? EJUSTRETURN : error);
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
#else /* !KSE */
return (EOPNOTSUPP);
#endif
Add kse_switchin(2). This syscall can be used by KSE implementations to have the kernel switch to a new thread, instead of doing it in userland. It is in fact needed on ia64 where syscall restarts do not return to userland first. It's completely handled inside the kernel. As such, any context created by the kernel as part of an upcall and caused by some syscall needs to be restored by the kernel.
2003-12-07 19:34:29 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
struct kse_thr_interrupt_args {
struct kse_thr_mailbox * tmbx;
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
int cmd;
long data;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
};
*/
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
int
kse_thr_interrupt(struct thread *td, struct kse_thr_interrupt_args *uap)
{
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
#ifdef KSE
Add an execve command for kse_thr_interrupt to allow libpthread to restore signal mask correctly, this is required by POSIX. Reviewed by: deischen
2004-10-07 13:50:10 +00:00
struct kse_execve_args args;
o Split out kernel part of execve(2) syscall into two parts: one that copies arguments into the kernel space and one that operates completely in the kernel space; o use kernel-only version of execve(2) to kill another stackgap in linuxlator/i386. Obtained from: DragonFlyBSD (partially) MFC after: 2 weeks
2005-01-29 23:12:00 +00:00
struct image_args iargs;
Add an actual implementation of kse_thr_interrupt()
2002-10-30 02:28:41 +00:00
struct proc *p;
struct thread *td2;
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
struct kse_upcall *ku;
struct kse_thr_mailbox *tmbx;
uint32_t flags;
Add an execve command for kse_thr_interrupt to allow libpthread to restore signal mask correctly, this is required by POSIX. Reviewed by: deischen
2004-10-07 13:50:10 +00:00
int error;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
KSE-enabled processes only.
2002-10-31 08:00:51 +00:00
p = td->td_proc;
Let SA process work under ULE scheduler, originally it would panic kernel. Reviewed by: jeff
2003-08-26 11:33:15 +00:00
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_LOCK(p);
if (!(p->p_flag & P_SA)) {
PROC_UNLOCK(p);
Check NULL thread mailbox pointer.
2002-10-30 05:09:29 +00:00
return (EINVAL);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
}
PROC_UNLOCK(p);
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
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
switch (uap->cmd) {
case KSE_INTR_SENDSIG:
if (uap->data < 0 || uap->data > _SIG_MAXSIG)
return (EINVAL);
case KSE_INTR_INTERRUPT:
case KSE_INTR_RESTART:
PROC_LOCK(p);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
FOREACH_THREAD_IN_PROC(p, td2) {
if (td2->td_mailbox == uap->tmbx)
break;
}
if (td2 == NULL) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
PROC_UNLOCK(p);
return (ESRCH);
}
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(td2);
PROC_SUNLOCK(p);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
if (uap->cmd == KSE_INTR_SENDSIG) {
if (uap->data > 0) {
td2->td_flags &= ~TDF_INTERRUPT;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td2);
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
tdsignal(p, td2, (int)uap->data, NULL);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
} else {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td2);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
}
} else {
td2->td_flags |= TDF_INTERRUPT | TDF_ASTPENDING;
if (TD_CAN_UNBIND(td2))
td2->td_upcall->ku_flags |= KUF_DOUPCALL;
if (uap->cmd == KSE_INTR_INTERRUPT)
td2->td_intrval = EINTR;
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
else
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
td2->td_intrval = ERESTART;
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_ON_SLEEPQ(td2) && (td2->td_flags & TDF_SINTR))
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
sleepq_abort(td2, td2->td_intrval);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td2);
Add an actual implementation of kse_thr_interrupt()
2002-10-30 02:28:41 +00:00
}
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
PROC_UNLOCK(p);
break;
case KSE_INTR_SIGEXIT:
if (uap->data < 1 || uap->data > _SIG_MAXSIG)
return (EINVAL);
PROC_LOCK(p);
sigexit(td, (int)uap->data);
break;
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
case KSE_INTR_DBSUSPEND:
/* this sub-function is only for bound thread */
if (td->td_pflags & TDP_SA)
return (EINVAL);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
ku = td->td_upcall;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
tmbx = (void *)fuword((void *)&ku->ku_mailbox->km_curthread);
if (tmbx == NULL || tmbx == (void *)-1)
return (EINVAL);
flags = 0;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_LOCK(p);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
while ((p->p_flag & P_TRACED) && !(p->p_flag & P_SINGLE_EXIT)) {
flags = fuword32(&tmbx->tm_dflags);
if (!(flags & TMDF_SUSPEND))
break;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
thread_stopped(p);
PROC_UNLOCK(p);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(td);
thread_suspend_one(td);
PROC_SUNLOCK(p);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
mi_switch(SW_VOL, NULL);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_LOCK(p);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
}
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_UNLOCK(p);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
return (0);
Add an execve command for kse_thr_interrupt to allow libpthread to restore signal mask correctly, this is required by POSIX. Reviewed by: deischen
2004-10-07 13:50:10 +00:00
case KSE_INTR_EXECVE:
error = copyin((void *)uap->data, &args, sizeof(args));
if (error)
return (error);
o Split out kernel part of execve(2) syscall into two parts: one that copies arguments into the kernel space and one that operates completely in the kernel space; o use kernel-only version of execve(2) to kill another stackgap in linuxlator/i386. Obtained from: DragonFlyBSD (partially) MFC after: 2 weeks
2005-01-29 23:12:00 +00:00
error = exec_copyin_args(&iargs, args.path, UIO_USERSPACE,
args.argv, args.envp);
if (error == 0)
error = kern_execve(td, &iargs, NULL);
Add an execve command for kse_thr_interrupt to allow libpthread to restore signal mask correctly, this is required by POSIX. Reviewed by: deischen
2004-10-07 13:50:10 +00:00
if (error == 0) {
PROC_LOCK(p);
SIGSETOR(td->td_siglist, args.sigpend);
PROC_UNLOCK(p);
kern_sigprocmask(td, SIG_SETMASK, &args.sigmask, NULL,
0);
}
return (error);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
default:
return (EINVAL);
Add an actual implementation of kse_thr_interrupt()
2002-10-30 02:28:41 +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
return (0);
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
#else /* !KSE */
return (EOPNOTSUPP);
#endif
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
struct kse_exit_args {
register_t dummy;
};
*/
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
int
kse_exit(struct thread *td, struct kse_exit_args *uap)
{
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
#ifdef KSE
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
struct proc *p;
Remove un-needed code. Don't copyin() data we are about to overwrite. Add a flag to tell userland that KSE is officially "DONE" with the mailbox and has gone away. Obtained from: davidxu@
2003-06-04 00:12:57 +00:00
struct kse_upcall *ku, *ku2;
int error, count;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
p = td->td_proc;
Shuffle some code around.
2004-06-11 17:48:20 +00:00
/*
* Ensure that this is only called from the UTS
*/
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
if ((ku = td->td_upcall) == NULL || TD_CAN_UNBIND(td)) {
thread_unlock(td);
Style fixes.
2002-10-30 03:01:28 +00:00
return (EINVAL);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
}
thread_unlock(td);
Shuffle some code around.
2004-06-11 17:48:20 +00:00
/*
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* Calculate the existing non-exiting upcalls in this process.
Shuffle some code around.
2004-06-11 17:48:20 +00:00
* If we are the last upcall but there are still other threads,
* then do not exit. We need the other threads to be able to
* complete whatever they are doing.
* XXX This relies on the userland knowing what to do if we return.
* It may be a better choice to convert ourselves into a kse_release
* ( or similar) and wait in the kernel to be needed.
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
* XXX Where are those other threads? I suppose they are waiting in
* the kernel. We should wait for them all at the user boundary after
* turning into an exit.
Shuffle some code around.
2004-06-11 17:48:20 +00:00
*/
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
count = 0;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
PROC_LOCK(p);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
FOREACH_UPCALL_IN_PROC(p, ku2) {
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
if ((ku2->ku_flags & KUF_EXITING) == 0)
Remove un-needed code. Don't copyin() data we are about to overwrite. Add a flag to tell userland that KSE is officially "DONE" with the mailbox and has gone away. Obtained from: davidxu@
2003-06-04 00:12:57 +00:00
count++;
}
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
if (count == 1 && (p->p_numthreads > 1)) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
PROC_UNLOCK(p);
return (EDEADLK);
}
Remove un-needed code. Don't copyin() data we are about to overwrite. Add a flag to tell userland that KSE is officially "DONE" with the mailbox and has gone away. Obtained from: davidxu@
2003-06-04 00:12:57 +00:00
ku->ku_flags |= KUF_EXITING;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Remove un-needed code. Don't copyin() data we are about to overwrite. Add a flag to tell userland that KSE is officially "DONE" with the mailbox and has gone away. Obtained from: davidxu@
2003-06-04 00:12:57 +00:00
PROC_UNLOCK(p);
Shuffle some code around.
2004-06-11 17:48:20 +00:00
/*
* Mark the UTS mailbox as having been finished with.
* If that fails then just go for a segfault.
* XXX need to check it that can be deliverred without a mailbox.
*/
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 07:33:40 +00:00
error = suword32(&ku->ku_mailbox->km_flags, ku->ku_mflags|KMF_DONE);
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
if (!(td->td_pflags & TDP_SA))
if (suword32(&td->td_mailbox->tm_lwp, 0))
error = EFAULT;
Remove un-needed code. Don't copyin() data we are about to overwrite. Add a flag to tell userland that KSE is officially "DONE" with the mailbox and has gone away. Obtained from: davidxu@
2003-06-04 00:12:57 +00:00
PROC_LOCK(p);
if (error)
psignal(p, SIGSEGV);
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_flush(&td->td_sigqueue);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
upcall_remove(td);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
if (p->p_numthreads != 1) {
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
thread_stopped(p);
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
thread_exit();
/* NOTREACHED */
}
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
/*
* This is the last thread. Just return to the user.
* Effectively we have left threading mode..
* The only real thing left to do is ensure that the
* scheduler sets out concurrency back to 1 as that may be a
* resource leak otherwise.
* This is an A[PB]I issue.. what SHOULD we do?
* One possibility is to return to the user. It may not cope well.
* The other possibility would be to let the process exit.
*/
Break out to a separate function, the code to revert a multithreaded process back to officially being a non-threaded program. MFC after: 4 days
2004-10-05 20:39:26 +00:00
thread_unthread(td);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
PROC_UNLOCK(p);
Fix a potential point of confusion. Art Ironport we've seen this end up with an infinite loop in and out of the kernel during process shutdown.
2006-12-12 08:01:55 +00:00
#if 0
Style fixes.
2002-10-30 03:01:28 +00:00
return (0);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
#else
Fix a potential point of confusion. Art Ironport we've seen this end up with an infinite loop in and out of the kernel during process shutdown.
2006-12-12 08:01:55 +00:00
printf("kse_exit: called on last thread. Calling exit1()");
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
exit1(td, 0);
#endif
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
#else /* !KSE */
return (EOPNOTSUPP);
#endif
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
Unbreak the KSE code. Keep track of zobie threads using the Per-CPU storage during the context switch. Rearrange thread cleanups to avoid problems with Giant. Clean threads when freed or when recycled. Approved by: re (jhb)
2002-12-10 02:33:45 +00:00
/*
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
* Either becomes an upcall or waits for an awakening event and
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* then becomes an upcall. Only error cases return.
Unbreak the KSE code. Keep track of zobie threads using the Per-CPU storage during the context switch. Rearrange thread cleanups to avoid problems with Giant. Clean threads when freed or when recycled. Approved by: re (jhb)
2002-12-10 02:33:45 +00:00
*/
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
struct kse_release_args {
Add a timeout parameter to kse_release.
2003-02-20 08:18:15 +00:00
struct timespec *timeout;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
};
*/
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
int
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
kse_release(struct thread *td, struct kse_release_args *uap)
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +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
#ifdef KSE
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
struct proc *p;
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
struct kse_upcall *ku;
struct timespec timeout;
Add a timeout parameter to kse_release.
2003-02-20 08:18:15 +00:00
struct timeval tv;
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
sigset_t sigset;
Add a timeout parameter to kse_release.
2003-02-20 08:18:15 +00:00
int error;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
p = td->td_proc;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
Fix a potential point of confusion. Art Ironport we've seen this end up with an infinite loop in and out of the kernel during process shutdown.
2006-12-12 08:01:55 +00:00
if ((ku = td->td_upcall) == NULL || TD_CAN_UNBIND(td)) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
Fix a potential point of confusion. Art Ironport we've seen this end up with an infinite loop in and out of the kernel during process shutdown.
2006-12-12 08:01:55 +00:00
printf("kse_release: called outside of threading. exiting");
exit1(td, 0);
}
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
Add a timeout parameter to kse_release.
2003-02-20 08:18:15 +00:00
if (uap->timeout != NULL) {
if ((error = copyin(uap->timeout, &timeout, sizeof(timeout))))
return (error);
TIMESPEC_TO_TIMEVAL(&tv, &timeout);
}
Move TDF_SA from td_flags to td_pflags (and rename it accordingly) so that it is no longer necessary to hold sched_lock while manipulating it. Reviewed by: davidxu
2004-06-02 07:52:36 +00:00
if (td->td_pflags & TDP_SA)
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
td->td_pflags |= TDP_UPCALLING;
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
else {
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 07:33:40 +00:00
ku->ku_mflags = fuword32(&ku->ku_mailbox->km_flags);
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
if (ku->ku_mflags == -1) {
PROC_LOCK(p);
sigexit(td, SIGSEGV);
}
}
Add a timeout parameter to kse_release.
2003-02-20 08:18:15 +00:00
PROC_LOCK(p);
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
if (ku->ku_mflags & KMF_WAITSIGEVENT) {
/* UTS wants to wait for signal event */
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
if (!(p->p_flag & P_SIGEVENT) &&
!(ku->ku_flags & KUF_DOUPCALL)) {
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
td->td_kflags |= TDK_KSERELSIG;
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
error = msleep(&p->p_siglist, &p->p_mtx, PPAUSE|PCATCH,
"ksesigwait", (uap->timeout ? tvtohz(&tv) : 0));
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
td->td_kflags &= ~(TDK_KSERELSIG | TDK_WAKEUP);
}
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
p->p_flag &= ~P_SIGEVENT;
sigset = p->p_siglist;
PROC_UNLOCK(p);
error = copyout(&sigset, &ku->ku_mailbox->km_sigscaught,
sizeof(sigset));
} else {
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
if ((ku->ku_flags & KUF_DOUPCALL) == 0 &&
((ku->ku_mflags & KMF_NOCOMPLETED) ||
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
(p->p_completed == NULL))) {
p->p_upsleeps++;
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
td->td_kflags |= TDK_KSEREL;
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
error = msleep(&p->p_completed, &p->p_mtx,
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
PPAUSE|PCATCH, "kserel",
(uap->timeout ? tvtohz(&tv) : 0));
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
td->td_kflags &= ~(TDK_KSEREL | TDK_WAKEUP);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
p->p_upsleeps--;
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
}
PROC_UNLOCK(p);
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
}
if (ku->ku_flags & KUF_DOUPCALL) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
ku->ku_flags &= ~KUF_DOUPCALL;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
Unbreak the KSE code. Keep track of zobie threads using the Per-CPU storage during the context switch. Rearrange thread cleanups to avoid problems with Giant. Clean threads when freed or when recycled. Approved by: re (jhb)
2002-12-10 02:33:45 +00:00
return (0);
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
#else /* !KSE */
return (EOPNOTSUPP);
#endif
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
/* struct kse_wakeup_args {
struct kse_mailbox *mbx;
}; */
int
kse_wakeup(struct thread *td, struct kse_wakeup_args *uap)
{
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
#ifdef KSE
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
struct proc *p;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
struct kse_upcall *ku;
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
struct thread *td2;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
p = td->td_proc;
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
td2 = NULL;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
ku = NULL;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
/* KSE-enabled processes only, please. */
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
PROC_LOCK(p);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
if (!(p->p_flag & P_SA)) {
PROC_UNLOCK(p);
return (EINVAL);
}
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
if (uap->mbx) {
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
FOREACH_UPCALL_IN_PROC(p, ku) {
if (ku->ku_mailbox == uap->mbx)
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
break;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
} else {
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if (p->p_upsleeps) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
wakeup(&p->p_completed);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
PROC_UNLOCK(p);
return (0);
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
}
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
ku = TAILQ_FIRST(&p->p_upcalls);
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
if (ku == NULL) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
PROC_UNLOCK(p);
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
return (ESRCH);
}
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
mtx_lock_spin(&kse_lock);
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
if ((td2 = ku->ku_owner) == NULL) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
mtx_unlock_spin(&kse_lock);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_UNLOCK(p);
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
panic("%s: no owner", __func__);
} else if (td2->td_kflags & (TDK_KSEREL | TDK_KSERELSIG)) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
mtx_unlock_spin(&kse_lock);
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
if (!(td2->td_kflags & TDK_WAKEUP)) {
td2->td_kflags |= TDK_WAKEUP;
if (td2->td_kflags & TDK_KSEREL)
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
sleepq_remove(td2, &p->p_completed);
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
else
sleepq_remove(td2, &p->p_siglist);
}
} else {
ku->ku_flags |= KUF_DOUPCALL;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
mtx_unlock_spin(&kse_lock);
Make an explicit flag to indicate that a KSE has a reason to upcall, and use that flag when there is a kse_wakeup() call. It will probably be used with signal delivery as well eventually. Submitted by: davidxu@
2003-01-03 20:41:49 +00:00
}
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
PROC_UNLOCK(p);
Keep track of threads waiting in kse_release() to avoid a race condition where kse_wakeup() doesn't yet see them in (interruptible) sleep queues. Also add an upcall check to sleepqueue_catch_signals() suggested by jhb. This commit should fix recent mysql hangs. Reviewed by: jhb, davidxu Mysql'd by: Robin P. Blanchard <robin.blanchard at gactr uga edu>
2004-04-28 20:36:53 +00:00
return (0);
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
#else /* !KSE */
return (EOPNOTSUPP);
#endif
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
/*
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* newgroup == 0: first call: use current KSE, don't schedule an upcall
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* All other situations, do allocate max new KSEs and schedule an upcall.
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
*
* XXX should be changed so that 'first' behaviour lasts for as long
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* as you have not made a thread in this proc. i.e. as long as we do not have
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
* a mailbox..
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
*/
/* struct kse_create_args {
struct kse_mailbox *mbx;
int newgroup;
}; */
int
kse_create(struct thread *td, struct kse_create_args *uap)
{
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
#ifdef KSE
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
struct proc *p;
struct kse_mailbox mbx;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
struct kse_upcall *newku;
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
int err, ncpus, sa = 0, first = 0;
struct thread *newtd;
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
p = td->td_proc;
light rearrangement of some code to get some locking more correct MFC after: 4 days
2004-10-05 20:48:16 +00:00
/*
* Processes using the other threading model can't
* suddenly start calling this one
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* XXX maybe...
light rearrangement of some code to get some locking more correct MFC after: 4 days
2004-10-05 20:48:16 +00:00
*/
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_LOCK(p);
light rearrangement of some code to get some locking more correct MFC after: 4 days
2004-10-05 20:48:16 +00:00
if ((p->p_flag & (P_SA|P_HADTHREADS)) == P_HADTHREADS) {
PROC_UNLOCK(p);
return (EINVAL);
}
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
if (!(p->p_flag & P_SA)) {
first = 1;
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
p->p_flag |= P_SA|P_HADTHREADS;
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
}
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_UNLOCK(p);
light rearrangement of some code to get some locking more correct MFC after: 4 days
2004-10-05 20:48:16 +00:00
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if ((err = copyin(uap->mbx, &mbx, sizeof(mbx))))
return (err);
ncpus = mp_ncpus;
if (virtual_cpu != 0)
ncpus = virtual_cpu;
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* If the new UTS mailbox says that this
* will be a BOUND lwp, then it had better
* have its thread mailbox already there.
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
*/
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if ((mbx.km_flags & KMF_BOUND) || uap->newgroup) {
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
/* It's a bound thread (1:1) */
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if (mbx.km_curthread == NULL)
return (EINVAL);
ncpus = 1;
if (!(uap->newgroup || first))
return (EINVAL);
} else {
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
/* It's an upcall capable thread */
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
sa = TDP_SA;
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
PROC_LOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
/*
* Limit it to NCPU upcall contexts per proc in any case.
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
* numupcalls will soon be numkse or something
* as it will represent the number of
* non-bound upcalls available. (i.e. ones that can
* actually call up).
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
*/
if (p->p_numupcalls >= ncpus) {
Let SA process work under ULE scheduler, originally it would panic kernel. Reviewed by: jeff
2003-08-26 11:33:15 +00:00
PROC_UNLOCK(p);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
return (EPROCLIM);
}
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
p->p_numupcalls++;
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
PROC_UNLOCK(p);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
}
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
/*
* For the first call this may not have been set.
* Of course nor may it actually be needed.
* thread_schedule_upcall() will look for it.
*/
if (td->td_standin == NULL) {
if (!thread_alloc_spare(td))
return (ENOMEM);
}
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
* Even bound LWPs get a mailbox and an upcall to hold it.
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
* XXX This should change.
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
*/
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
newku = upcall_alloc();
newku->ku_mailbox = uap->mbx;
newku->ku_func = mbx.km_func;
bcopy(&mbx.km_stack, &newku->ku_stack, sizeof(stack_t));
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
PROC_LOCK(p);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
/*
* If we are the first time, and a normal thread,
* then transfer all the signals back to the 'process'.
* SA threading will make a special thread to handle them.
*/
if (first) {
sigqueue_move_set(&td->td_sigqueue, &p->p_sigqueue,
&td->td_sigqueue.sq_signals);
SIGFILLSET(td->td_sigmask);
SIG_CANTMASK(td->td_sigmask);
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
}
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* Make the new upcall available to the process.
Various small style fixes.
2004-09-22 15:24:33 +00:00
* It may or may not use it, but it's available.
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
*/
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
TAILQ_INSERT_TAIL(&p->p_upcalls, newku, ku_link);
newku->ku_proc = p;
Fix a LOR between sched_lock and sleep queue lock.
2005-08-19 13:35:34 +00:00
PROC_UNLOCK(p);
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
if (mbx.km_quantum)
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
/* XXX should this be in the thread? */
p->p_upquantum = max(1, mbx.km_quantum / tick);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
* Each upcall structure has an owner thread, find which
* one owns it.
*/
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(td);
mtx_lock_spin(&kse_lock);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
if (uap->newgroup) {
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
/*
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* The newgroup parameter now means
* "bound, non SA, system scope"
* It is only used for the interrupt thread at the
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
* moment I think.. (or system scope threads dopey).
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* We'll rename it later.
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
*/
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
newtd = thread_schedule_upcall(td, newku);
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
} else {
/*
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
* If the current thread hasn't an upcall structure,
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* just assign the upcall to it.
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
* It'll just return.
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
*/
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
if (td->td_upcall == NULL) {
newku->ku_owner = td;
td->td_upcall = newku;
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
newtd = td;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
} else {
/*
* Create a new upcall thread to own it.
*/
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
newtd = thread_schedule_upcall(td, newku);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
}
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
}
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
mtx_unlock_spin(&kse_lock);
thread_unlock(td);
PROC_SUNLOCK(p);
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
* Let the UTS instance know its LWPID.
* It doesn't really care. But the debugger will.
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* XXX warning.. remember that this moves.
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +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 07:33:40 +00:00
suword32(&newku->ku_mailbox->km_lwp, newtd->td_tid);
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
* In the same manner, if the UTS has a current user thread,
* then it is also running on this LWP so set it as well.
* The library could do that of course.. but why not..
After the last change to KSE threading a bug was introduced where all threads were counted against the count of upcall capable threads. this changes the way we do this accounting.
2007-03-07 20:17:41 +00:00
* XXX I'm not sure this can ever happen but ...
* XXX does the UTS ever set this in the mailbox before calling this?
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +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 07:33:40 +00:00
if (mbx.km_curthread)
suword32(&mbx.km_curthread->tm_lwp, newtd->td_tid);
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
if (sa) {
newtd->td_pflags |= TDP_SA;
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
/*
* If we are starting a new thread, kick it off.
*/
if (newtd != td) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(newtd);
- Remove setrunqueue and replace it with direct calls to sched_add(). setrunqueue() was mostly empty. The few asserts and thread state setting were moved to the individual schedulers. sched_add() was chosen to displace it for naming consistency reasons. - Remove adjustrunqueue, it was 4 lines of code that was ifdef'd to be different on all three schedulers where it was only called in one place each. - Remove the long ifdef'd out remrunqueue code. - Remove the now redundant ts_state. Inspect the thread state directly. - Don't set TSF_* flags from kern_switch.c, we were only doing this to support a feature in one scheduler. - Change sched_choose() to return a thread rather than a td_sched. Also, rely on the schedulers to return the idlethread. This simplifies the logic in choosethread(). Aside from the run queue links kern_switch.c mostly does not care about the contents of td_sched. Discussed with: julian - Move the idle thread loop into the per scheduler area. ULE wants to do something different from the other schedulers. Suggested by: jhb Tested on: x86/amd64 sched_{4BSD, ULE, CORE}.
2007-01-23 08:46:51 +00:00
sched_add(newtd, SRQ_BORING);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(newtd);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
}
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
} else {
Move TDF_SA from td_flags to td_pflags (and rename it accordingly) so that it is no longer necessary to hold sched_lock while manipulating it. Reviewed by: davidxu
2004-06-02 07:52:36 +00:00
newtd->td_pflags &= ~TDP_SA;
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
* Since a library will use the mailbox pointer to
* identify even a bound thread, and the mailbox pointer
* will never be allowed to change after this syscall
* for a bound thread, set it here so the library can
* find the thread after the syscall returns.
*/
newtd->td_mailbox = mbx.km_curthread;
cpu_set_upcall_kse needs to access userspace, release schedule lock before calling it for bound thread. To avoid this problem, change thread_schedule_upcall to not put new thread on run queue, let caller do it, so we can tweak the new thread before setting it to run. Reported by: pho
2003-06-20 09:12:12 +00:00
if (newtd != td) {
Comment kse_create() and make a few minor code cleanups Reviewed by: davidxu
2004-08-01 23:02:00 +00:00
/*
* If we did create a new thread then
* make sure it goes to the right place
* when it starts up, and make sure that it runs
* at full speed when it gets there.
* thread_schedule_upcall() copies all cpu state
* to the new thread, so we should clear single step
* flag here.
*/
Change cpu_set_kse_upcall to more generic style, so we can reuse it in other codes. Add cpu_set_user_tls, use it to tweak user register and setup user TLS. I ever wanted to merge it into cpu_set_kse_upcall, but since cpu_set_kse_upcall is also used by M:N threads which may not need this feature, so I wrote a separated cpu_set_user_tls.
2005-04-23 02:32:32 +00:00
cpu_set_upcall_kse(newtd, newku->ku_func,
newku->ku_mailbox, &newku->ku_stack);
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
PROC_LOCK(p);
if (p->p_flag & P_TRACED) {
_PHOLD(p);
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 07:33:40 +00:00
ptrace_clear_single_step(newtd);
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
_PRELE(p);
}
PROC_UNLOCK(p);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(newtd);
- Remove setrunqueue and replace it with direct calls to sched_add(). setrunqueue() was mostly empty. The few asserts and thread state setting were moved to the individual schedulers. sched_add() was chosen to displace it for naming consistency reasons. - Remove adjustrunqueue, it was 4 lines of code that was ifdef'd to be different on all three schedulers where it was only called in one place each. - Remove the long ifdef'd out remrunqueue code. - Remove the now redundant ts_state. Inspect the thread state directly. - Don't set TSF_* flags from kern_switch.c, we were only doing this to support a feature in one scheduler. - Change sched_choose() to return a thread rather than a td_sched. Also, rely on the schedulers to return the idlethread. This simplifies the logic in choosethread(). Aside from the run queue links kern_switch.c mostly does not care about the contents of td_sched. Discussed with: julian - Move the idle thread loop into the per scheduler area. ULE wants to do something different from the other schedulers. Suggested by: jhb Tested on: x86/amd64 sched_{4BSD, ULE, CORE}.
2007-01-23 08:46:51 +00:00
sched_add(newtd, SRQ_BORING);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(newtd);
cpu_set_upcall_kse needs to access userspace, release schedule lock before calling it for bound thread. To avoid this problem, change thread_schedule_upcall to not put new thread on run queue, let caller do it, so we can tweak the new thread before setting it to run. Reported by: pho
2003-06-20 09:12:12 +00:00
}
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
}
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
return (0);
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
#else /* !KSE */
return (EOPNOTSUPP);
#endif
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +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
#ifdef KSE
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
/*
Shuffle some code around.
2004-06-11 17:48:20 +00:00
* Initialize global thread allocation resources.
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
*/
void
Move the KSE ABI specific code here and separate it from code that is generic to any threading system. This commit does not link this file to the build yet, nor does it remove these functions from their current location in kern_thread.c. (that commit coming up after further review)
2004-06-07 07:25:03 +00:00
kseinit(void)
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
{
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
upcall_zone = uma_zcreate("UPCALL", sizeof(struct kse_upcall),
NULL, NULL, NULL, NULL, UMA_ALIGN_CACHE, 0);
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
}
/*
* Store the thread context in the UTS's mailbox.
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
* then add the mailbox at the head of a list we are building in user space.
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* The list is anchored in the proc structure.
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
*/
int
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
thread_export_context(struct thread *td, int willexit)
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
{
Put an easy-to-miss assignment into the proper place. It was stray in the middle of a block of code, with no clear assignment. While here, move one nearby assignment out of declaration.
2002-10-05 04:49:46 +00:00
struct proc *p;
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
uintptr_t mbx;
void *addr;
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
int error = 0, sig;
Remove un-needed code. Don't copyin() data we are about to overwrite. Add a flag to tell userland that KSE is officially "DONE" with the mailbox and has gone away. Obtained from: davidxu@
2003-06-04 00:12:57 +00:00
mcontext_t mc;
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
Put an easy-to-miss assignment into the proper place. It was stray in the middle of a block of code, with no clear assignment. While here, move one nearby assignment out of declaration.
2002-10-05 04:49:46 +00:00
p = td->td_proc;
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
/*
* Post sync signal, or process SIGKILL and SIGSTOP.
* For sync signal, it is only possible when the signal is not
* caught by userland or process is being debugged.
*/
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
PROC_LOCK(p);
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
if (td->td_flags & TDF_NEEDSIGCHK) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(td);
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
td->td_flags &= ~TDF_NEEDSIGCHK;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td);
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
mtx_lock(&p->p_sigacts->ps_mtx);
while ((sig = cursig(td)) != 0)
postsig(sig);
mtx_unlock(&p->p_sigacts->ps_mtx);
}
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
if (willexit)
SIGFILLSET(td->td_sigmask);
PROC_UNLOCK(p);
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
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 07:33:40 +00:00
/* Export the user/machine context. */
get_mcontext(td, &mc, 0);
addr = (void *)(&td->td_mailbox->tm_context.uc_mcontext);
error = copyout(&mc, addr, sizeof(mcontext_t));
if (error)
goto bad;
addr = (caddr_t)(&td->td_mailbox->tm_lwp);
if (suword32(addr, 0)) {
error = EFAULT;
goto bad;
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/* Get address in latest mbox of list pointer */
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
addr = (void *)(&td->td_mailbox->tm_next);
/*
* Put the saved address of the previous first
* entry into this one
*/
for (;;) {
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
mbx = (uintptr_t)p->p_completed;
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
if (suword(addr, mbx)) {
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
error = EFAULT;
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
goto bad;
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
}
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_LOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if (mbx == (uintptr_t)p->p_completed) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
p->p_completed = td->td_mailbox;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
* The thread context may be taken away by
* other upcall threads when we unlock
* process lock. it's no longer valid to
* use it again in any other places.
*/
td->td_mailbox = NULL;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_UNLOCK(p);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
break;
}
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_UNLOCK(p);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td->td_usticks = 0;
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
return (0);
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
bad:
PROC_LOCK(p);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
sigexit(td, SIGILL);
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
return (error);
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
}
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
/*
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
* Take the list of completed mailboxes for this Process and put them on this
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* upcall's mailbox as it's the next one going up.
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
*/
static int
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
thread_link_mboxes(struct proc *p, struct kse_upcall *ku)
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
{
void *addr;
uintptr_t mbx;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
addr = (void *)(&ku->ku_mailbox->km_completed);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
for (;;) {
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
mbx = (uintptr_t)p->p_completed;
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
if (suword(addr, mbx)) {
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_LOCK(p);
psignal(p, SIGSEGV);
PROC_UNLOCK(p);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
return (EFAULT);
}
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_LOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if (mbx == (uintptr_t)p->p_completed) {
p->p_completed = NULL;
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_UNLOCK(p);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
break;
}
lock proc while calling psignal (plus related cleanups) Submitted by: davidxu
2002-09-29 02:48:37 +00:00
PROC_UNLOCK(p);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
}
return (0);
}
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
/*
* This function should be called at statclock interrupt time
*/
int
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
thread_statclock(int user)
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
{
struct thread *td = curthread;
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
if (!(td->td_pflags & TDP_SA))
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
return (0);
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
if (user) {
/* Current always do via ast() */
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(td);
- Move TDF_OWEPREEMPT, TDF_OWEUPC, and TDF_USTATCLOCK over to td_pflags since they are only accessed by curthread and thus do not need any locking. - Move pr_addr and pr_ticks out of struct uprof (which is per-process) and directly into struct thread as td_profil_addr and td_profil_ticks as these variables are really per-thread. (They are used to defer an addupc_intr() that was too "hard" until ast()).
2004-07-16 21:04:55 +00:00
td->td_flags |= TDF_ASTPENDING;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td->td_uuticks++;
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 07:33:40 +00:00
} else if (td->td_mailbox != NULL)
td->td_usticks++;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
return (0);
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
Add a simple facility to allow round roubin in userland. Reviewed by: julain
2003-02-26 00:58:23 +00:00
* Export state clock ticks for userland
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
*/
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
static int
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
thread_update_usr_ticks(struct thread *td)
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
{
struct proc *p = td->td_proc;
caddr_t addr;
Consistently use the BSD u_int and u_short instead of the SYSV uint and ushort. In most of these files, there was a mixture of both styles and this change just makes them self-consistent. Requested by: bde (kern_ktrace.c)
2003-08-07 15:04:27 +00:00
u_int uticks;
Reversion of commit by Davidxu plus fixes since applied. I'm not convinced there is anything major wrong with the patch but them's the rules.. I am using my "David's mentor" hat to revert this as he's offline for a while.
2003-02-01 12:17:09 +00:00
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
if (td->td_mailbox == NULL) {
thread_unlock(td);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
return (-1);
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
}
thread_unlock(td);
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
if ((uticks = td->td_uuticks) != 0) {
Add a simple facility to allow round roubin in userland. Reviewed by: julain
2003-02-26 00:58:23 +00:00
td->td_uuticks = 0;
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
addr = (caddr_t)&td->td_mailbox->tm_uticks;
if (suword32(addr, uticks+fuword32(addr)))
goto error;
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
}
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
if ((uticks = td->td_usticks) != 0) {
td->td_usticks = 0;
addr = (caddr_t)&td->td_mailbox->tm_sticks;
if (suword32(addr, uticks+fuword32(addr)))
goto error;
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
return (0);
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
error:
PROC_LOCK(p);
psignal(p, SIGSEGV);
PROC_UNLOCK(p);
return (-2);
1. Support versioning and wall clock in kse mailbox, also add rusage time in thread mailbox. 2. Minor change for thread limit code in thread_user_enter(), fix typo in kse_release() last I committed. Reviewed by: deischen, mini
2002-11-18 01:59:31 +00:00
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
* This function is intended to be used to initialize a spare thread
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
* for upcall. Initialize thread's large data area outside the thread lock
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
* for thread_schedule_upcall(). The crhold is also here to get it out
* from the schedlock as it has a mutex op itself.
light rearrangement of some code to get some locking more correct MFC after: 4 days
2004-10-05 20:48:16 +00:00
* XXX BUG.. we need to get the cr ref after the thread has
* checked and chenged its own, not 6 months before...
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
*/
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
int
Remove an argument that is never used.
2004-08-02 23:48:43 +00:00
thread_alloc_spare(struct thread *td)
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
{
Repeat after me: "Do not apply your tested patches to your commit tree by hand"
2004-08-03 01:43:29 +00:00
struct thread *spare;
Push Giant down a little further: - no longer serialize on Giant for thread_single*() and family in fork, exit and exec - thread_wait() is mpsafe, assert no Giant - reduce scope of Giant in exit to not cover thread_wait and just do vm_waitproc(). - assert that thread_single() family are not called with Giant - remove the DROP/PICKUP_GIANT macros from thread_single() family - assert that thread_suspend_check() s not called with Giant - remove manual drop_giant hack in thread_suspend_check since we know it isn't held. - remove the DROP/PICKUP_GIANT macros from thread_suspend_check() family - mark kse_create() mpsafe
2004-03-13 22:31:39 +00:00
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
if (td->td_standin)
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
return (1);
Remove an argument that is never used.
2004-08-02 23:48:43 +00:00
spare = thread_alloc();
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
if (spare == NULL)
return (0);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td->td_standin = spare;
bzero(&spare->td_startzero,
Remove local definitions of RANGEOF() and use __rangeof() instead. Also remove a few bogus casts.
2004-11-20 23:00:59 +00:00
__rangeof(struct thread, td_startzero, td_endzero));
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
spare->td_proc = td->td_proc;
spare->td_ucred = crhold(td->td_ucred);
- Move all of the PS_ flags into either p_flag or td_flags. - p_sflag was mostly protected by PROC_LOCK rather than the PROC_SLOCK or previously the sched_lock. These bugs have existed for some time. - Allow swapout to try each thread in a process individually and then swapin the whole process if any of these fail. This allows us to move most scheduler related swap flags into td_flags. - Keep ki_sflag for backwards compat but change all in source tools to use the new and more correct location of P_INMEM. Reported by: pho Reviewed by: attilio, kib Approved by: re (kensmith)
2007-09-17 05:31:39 +00:00
spare->td_flags = TDF_INMEM;
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
return (1);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
}
Move thread related code from kern_proc.c to kern_thread.c. Add code to free KSEs and KSEGRPs on exit. Sort KSE prototypes in proc.h. Add the missing kse_exit() syscall. ksetest now does not leak KSEs and KSEGRPS. Submitted by: (parts) davidxu
2002-10-24 08:46:34 +00:00
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
/*
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
* Create a thread and schedule it for upcall on the KSE given.
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
* Use our thread's standin so that we don't have to allocate one.
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
*/
struct thread *
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
thread_schedule_upcall(struct thread *td, struct kse_upcall *ku)
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
{
struct thread *td2;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
mtx_assert(&kse_lock, MA_OWNED);
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
/*
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* Schedule an upcall thread on specified kse_upcall,
* the kse_upcall must be free.
* td must have a spare thread.
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
*/
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
KASSERT(ku->ku_owner == NULL, ("%s: upcall has owner", __func__));
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
if ((td2 = td->td_standin) != NULL) {
td->td_standin = NULL;
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
} else {
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
panic("no reserve thread when scheduling an upcall");
Style fixes.
2002-10-30 03:01:28 +00:00
return (NULL);
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
}
CTR3(KTR_PROC, "thread_schedule_upcall: thread %p (pid %d, %s)",
generally we are interested in what thread did something as opposed to what process. Since threads by default have teh name of the process unless over-written with more useful information, just print the thread name instead.
2007-11-14 06:21:24 +00:00
td2, td->td_proc->p_pid, td->td_name);
Slight changes to comments and some whitespace changes.
2004-08-09 21:57:30 +00:00
/*
* Bzero already done in thread_alloc_spare() because we can't
* do the crhold here because we are in schedlock already.
*/
Use UMA as a complex object allocator. The process allocator now caches and hands out complete process structures *including substructures* . i.e. it get's the process structure with the first thread (and soon KSE) already allocated and attached, all in one hit. For the average non threaded program (non KSE that is) the allocated thread and its stack remain attached to the process, even when the process is unused and in the process cache. This saves having to allocate and attach it later, effectively bringing us (hopefully) close to the efficiency of pre-KSE systems where these were a single structure. Reviewed by: davidxu@freebsd.org, peter@freebsd.org
2002-09-06 07:00:37 +00:00
bcopy(&td->td_startcopy, &td2->td_startcopy,
Remove local definitions of RANGEOF() and use __rangeof() instead. Also remove a few bogus casts.
2004-11-20 23:00:59 +00:00
__rangeof(struct thread, td_startcopy, td_endcopy));
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
sched_fork_thread(td, td2);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
thread_link(td2, ku->ku_proc);
Fix build.
2007-11-15 14:16:20 +00:00
bcopy(ku->ku_proc->p_comm, td2->td_name, sizeof(td2->td_name));
Shuffle some code around.
2004-06-11 17:48:20 +00:00
/* inherit parts of blocked thread's context as a good template */
Change the second (and last) argument of cpu_set_upcall(). Previously we were passing in a void* representing the PCB of the parent thread. Now we pass a pointer to the parent thread itself. The prime reason for this change is to allow cpu_set_upcall() to copy (parts of) the trapframe instead of having it done in MI code in each caller of cpu_set_upcall(). Copying the trapframe cannot always be done with a simply bcopy() or may not always be optimal that way. On ia64 specifically the trapframe contains information that is specific to an entry into the kernel and can only be used by the corresponding exit from the kernel. A trapframe copied verbatim from another frame is in most cases useless without some additional normalization. Note that this change removes the assignment to td->td_frame in some implementations of cpu_set_upcall(). The assignment is redundant. A previous call to cpu_thread_setup() already did the exact same assignment. An added benefit of removing the redundant assignment is that we can now change td_pcb without nasty side-effects. This change officially marks the ability on ia64 for 1:1 threading. Not tested on: amd64, powerpc Compile & boot tested on: alpha, sparc64 Functionally tested on: i386, ia64
2003-06-04 21:13:21 +00:00
cpu_set_upcall(td2, td);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/* Let the new thread become owner of the upcall */
ku->ku_owner = td2;
td2->td_upcall = ku;
Move TDF_SA from td_flags to td_pflags (and rename it accordingly) so that it is no longer necessary to hold sched_lock while manipulating it. Reviewed by: davidxu
2004-06-02 07:52:36 +00:00
td2->td_pflags = TDP_SA|TDP_UPCALLING;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td2->td_state = TDS_CAN_RUN;
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
td2->td_inhibitors = 0;
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
SIGFILLSET(td2->td_sigmask);
SIG_CANTMASK(td2->td_sigmask);
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
return (td2); /* bogus.. should be a void function */
Add files that are new for KSE.
2002-06-29 07:04:59 +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
/*
* It is only used when thread generated a trap and process is being
* debugged.
*/
- Add a new function, thread_signal_add(), that is called from postsig to add a signal to a mailbox's pending set. - Add a new function, thread_signal_upcall(), this causes the current thread to upcall so that we can deliver pending signals. Reviewed by: mini
2003-02-17 09:58:11 +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
thread_signal_add(struct thread *td, ksiginfo_t *ksi)
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
{
- Add a new function, thread_signal_add(), that is called from postsig to add a signal to a mailbox's pending set. - Add a new function, thread_signal_upcall(), this causes the current thread to upcall so that we can deliver pending signals. Reviewed by: mini
2003-02-17 09:58:11 +00:00
struct proc *p;
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
struct sigacts *ps;
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
int error;
thread_signal_add now is called with ps_mtx held, unlock it before calling copyin.
2003-06-06 02:17:38 +00:00
p = td->td_proc;
PROC_LOCK_ASSERT(p, MA_OWNED);
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
ps = p->p_sigacts;
mtx_assert(&ps->ps_mtx, MA_OWNED);
- Add a new function, thread_signal_add(), that is called from postsig to add a signal to a mailbox's pending set. - Add a new function, thread_signal_upcall(), this causes the current thread to upcall so that we can deliver pending signals. Reviewed by: mini
2003-02-17 09:58:11 +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
mtx_unlock(&ps->ps_mtx);
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
SIGADDSET(td->td_sigmask, ksi->ksi_signo);
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
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
error = copyout(&ksi->ksi_info, &td->td_mailbox->tm_syncsig,
sizeof(siginfo_t));
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
if (error) {
PROC_LOCK(p);
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 07:33:40 +00:00
sigexit(td, SIGSEGV);
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
}
- Add a new function, thread_signal_add(), that is called from postsig to add a signal to a mailbox's pending set. - Add a new function, thread_signal_upcall(), this causes the current thread to upcall so that we can deliver pending signals. Reviewed by: mini
2003-02-17 09:58:11 +00:00
PROC_LOCK(p);
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
mtx_lock(&ps->ps_mtx);
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
}
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
#include "opt_sched.h"
struct thread *
thread_switchout(struct thread *td, int flags, struct thread *nextthread)
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
{
struct kse_upcall *ku;
cpu_set_upcall_kse needs to access userspace, release schedule lock before calling it for bound thread. To avoid this problem, change thread_schedule_upcall to not put new thread on run queue, let caller do it, so we can tweak the new thread before setting it to run. Reported by: pho
2003-06-20 09:12:12 +00:00
struct thread *td2;
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
THREAD_LOCK_ASSERT(td, MA_OWNED);
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
/*
* If the outgoing thread is in threaded group and has never
* scheduled an upcall, decide whether this is a short
* or long term event and thus whether or not to schedule
* an upcall.
* If it is a short term event, just suspend it in
* a way that takes its KSE with it.
* Select the events for which we want to schedule upcalls.
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
* For now it's just sleep or if thread is suspended but
* process wide suspending flag is not set (debugger
* suspends thread).
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
* XXXKSE eventually almost any inhibition could do.
*/
1.Add KSE_INTR_DBSUSPEND command for kse_thr_interrupt to suspend a bound thread, after the bound thread leaves critical region, the thread should check debug flag may suspend itself by using the command. 2.Schedule upcall after thread is suspended by debugger 3.Wakeup upcall thread after process suspension. Reviewed by: deischen
2004-08-08 22:32:20 +00:00
if (TD_CAN_UNBIND(td) && (td->td_standin) &&
(TD_ON_SLEEPQ(td) || (TD_IS_SUSPENDED(td) &&
!P_SHOULDSTOP(td->td_proc)))) {
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
/*
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
* Release ownership of upcall, and schedule an upcall
* thread, this new upcall thread becomes the owner of
Slight changes to comments and some whitespace changes.
2004-08-09 21:57:30 +00:00
* the upcall structure. It will be ahead of us in the
* run queue, so as we are stopping, it should either
* start up immediatly, or at least before us if
* we release our slot.
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
*/
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
mtx_lock_spin(&kse_lock);
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
ku = td->td_upcall;
ku->ku_owner = NULL;
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
td->td_upcall = NULL;
Move TDF_CAN_UNBIND to thread private flags td_pflags, this eliminates need of sched_lock in some places. Also in thread_userret, remove spare thread allocation code, it is already done in thread_user_enter. Reviewed by: julian
2004-08-28 04:08:05 +00:00
td->td_pflags &= ~TDP_CAN_UNBIND;
cpu_set_upcall_kse needs to access userspace, release schedule lock before calling it for bound thread. To avoid this problem, change thread_schedule_upcall to not put new thread on run queue, let caller do it, so we can tweak the new thread before setting it to run. Reported by: pho
2003-06-20 09:12:12 +00:00
td2 = thread_schedule_upcall(td, ku);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
mtx_unlock_spin(&kse_lock);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
if (flags & SW_INVOL || nextthread) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_lock(td2);
- Remove setrunqueue and replace it with direct calls to sched_add(). setrunqueue() was mostly empty. The few asserts and thread state setting were moved to the individual schedulers. sched_add() was chosen to displace it for naming consistency reasons. - Remove adjustrunqueue, it was 4 lines of code that was ifdef'd to be different on all three schedulers where it was only called in one place each. - Remove the long ifdef'd out remrunqueue code. - Remove the now redundant ts_state. Inspect the thread state directly. - Don't set TSF_* flags from kern_switch.c, we were only doing this to support a feature in one scheduler. - Change sched_choose() to return a thread rather than a td_sched. Also, rely on the schedulers to return the idlethread. This simplifies the logic in choosethread(). Aside from the run queue links kern_switch.c mostly does not care about the contents of td_sched. Discussed with: julian - Move the idle thread loop into the per scheduler area. ULE wants to do something different from the other schedulers. Suggested by: jhb Tested on: x86/amd64 sched_{4BSD, ULE, CORE}.
2007-01-23 08:46:51 +00:00
sched_add(td2, SRQ_YIELDING);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
thread_unlock(td2);
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
} else {
/* Keep up with reality.. we have one extra thread
* in the picture.. and it's 'running'.
*/
return td2;
}
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
}
Refactor a bunch of scheduler code to give basically the same behaviour but with slightly cleaned up interfaces. The KSE structure has become the same as the "per thread scheduler private data" structure. In order to not make the diffs too great one is #defined as the other at this time. The KSE (or td_sched) structure is now allocated per thread and has no allocation code of its own. Concurrency for a KSEGRP is now kept track of via a simple pair of counters rather than using KSE structures as tokens. Since the KSE structure is different in each scheduler, kern_switch.c is now included at the end of each scheduler. Nothing outside the scheduler knows the contents of the KSE (aka td_sched) structure. The fields in the ksegrp structure that are to do with the scheduler's queueing mechanisms are now moved to the kg_sched structure. (per ksegrp scheduler private data structure). In other words how the scheduler queues and keeps track of threads is no-one's business except the scheduler's. This should allow people to write experimental schedulers with completely different internal structuring. A scheduler call sched_set_concurrency(kg, N) has been added that notifies teh scheduler that no more than N threads from that ksegrp should be allowed to be on concurrently scheduled. This is also used to enforce 'fainess' at this time so that a ksegrp with 10000 threads can not swamp a the run queue and force out a process with 1 thread, since the current code will not set the concurrency above NCPU, and both schedulers will not allow more than that many onto the system run queue at a time. Each scheduler should eventualy develop their own methods to do this now that they are effectively separated. Rejig libthr's kernel interface to follow the same code paths as linkse for scope system threads. This has slightly hurt libthr's performance but I will work to recover as much of it as I can. Thread exit code has been cleaned up greatly. exit and exec code now transitions a process back to 'standard non-threaded mode' before taking the next step. Reviewed by: scottl, peter MFC after: 1 week
2004-09-05 02:09:54 +00:00
return (nextthread);
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
}
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +00:00
/*
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* Setup done on the thread when it enters the kernel.
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +00:00
*/
void
Remove an unneeded argument.. The removed argument could trivially be derived from the remaining one. That in turn should be the same as curthread, but it is possible that curthread could be expensive to derive on some syste,s so leave it as an argument. Having both proc and thread as an argumen tjust gives an opportunity for them to get out sync. MFC after: 3 days
2004-08-31 07:34:54 +00:00
thread_user_enter(struct thread *td)
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +00:00
{
Restore some code removed in revision 1.193 and 1.194, julian said he'd like to keep these code.
2004-10-06 00:49:41 +00:00
struct proc *p = td->td_proc;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
struct kse_upcall *ku;
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
struct kse_thr_mailbox *tmbx;
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 07:33:40 +00:00
uint32_t flags;
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
Restore some code removed in revision 1.193 and 1.194, julian said he'd like to keep these code.
2004-10-06 00:49:41 +00:00
/*
* First check that we shouldn't just abort. we
* can suspend it here or just exit.
*/
if (__predict_false(P_SHOULDSTOP(p))) {
PROC_LOCK(p);
thread_suspend_check(0);
PROC_UNLOCK(p);
}
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 07:33:40 +00:00
if (!(td->td_pflags & TDP_SA))
return;
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +00:00
/*
* If we are doing a syscall in a KSE environment,
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 07:33:40 +00:00
* note where our mailbox is.
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +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 07:33:40 +00:00
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
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 07:33:40 +00:00
ku = td->td_upcall;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
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 07:33:40 +00:00
KASSERT(ku != NULL, ("no upcall owned"));
KASSERT(ku->ku_owner == td, ("wrong owner"));
KASSERT(!TD_CAN_UNBIND(td), ("can unbind"));
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
if (td->td_standin == NULL) {
if (!thread_alloc_spare(td)) {
PROC_LOCK(p);
if (kern_logsigexit)
log(LOG_INFO,
"pid %d (%s), uid %d: thread_alloc_spare failed\n",
p->p_pid, p->p_comm,
td->td_ucred ? td->td_ucred->cr_uid : -1);
sigexit(td, SIGSEGV); /* XXX ? */
/* panic("thread_user_enter: thread_alloc_spare failed"); */
}
}
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 07:33:40 +00:00
ku->ku_mflags = fuword32((void *)&ku->ku_mailbox->km_flags);
tmbx = (void *)fuword((void *)&ku->ku_mailbox->km_curthread);
if ((tmbx == NULL) || (tmbx == (void *)-1L) ||
(ku->ku_mflags & KMF_NOUPCALL)) {
td->td_mailbox = NULL;
} else {
flags = fuword32(&tmbx->tm_flags);
/*
* On some architectures, TP register points to thread
* mailbox but not points to kse mailbox, and userland
* can not atomically clear km_curthread, but can
* use TP register, and set TMF_NOUPCALL in thread
* flag to indicate a critical region.
*/
if (flags & TMF_NOUPCALL) {
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td->td_mailbox = NULL;
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +00:00
} else {
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 07:33:40 +00:00
td->td_mailbox = tmbx;
Move TDF_CAN_UNBIND to thread private flags td_pflags, this eliminates need of sched_lock in some places. Also in thread_userret, remove spare thread allocation code, it is already done in thread_user_enter. Reviewed by: julian
2004-08-28 04:08:05 +00:00
td->td_pflags |= TDP_CAN_UNBIND;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_LOCK(p);
Restore some code removed in revision 1.193 and 1.194, julian said he'd like to keep these code.
2004-10-06 00:49:41 +00:00
if (__predict_false(p->p_flag & P_TRACED)) {
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 07:33:40 +00:00
flags = fuword32(&tmbx->tm_dflags);
s/TMDF_DONOTRUNUSER/TMDF_SUSPEND/g Dicussed with: deischen
2004-08-03 02:23:06 +00:00
if (flags & TMDF_SUSPEND) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
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 07:33:40 +00:00
/* fuword can block, check again */
if (td->td_upcall)
ku->ku_flags |= KUF_DOUPCALL;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
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 07:33:40 +00:00
}
Introduce a thread mailbox flag TMF_NOUPCALL. On some architectures other than i386 or AMD64, TP register points to thread mailbox, and they can not atomically clear km_curthread in kse mailbox, in this case, thread retrieves its thread pointer from TP register and sets flag TMF_NOUPCALL in its thread mailbox to indicate a critical region.
2003-08-05 12:00:55 +00:00
}
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_UNLOCK(p);
Extract out KSE specific code from machine specific code so that there is ony one copy of it. Fix that one copy so that KSEs with no mailbox in a KSE program are not a cause of page faults (this can legitmatly happen). Submitted by: (parts) davidxu
2002-10-24 23:09:48 +00:00
}
}
}
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
/*
* The extra work we go through if we are a threaded process when we
* return to userland.
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
*
* If we are a KSE process and returning to user mode, check for
* extra work to do before we return (e.g. for more syscalls
* to complete first). If we were in a critical section, we should
* just return to let it finish. Same if we were in the UTS (in
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
* which case the mailbox's context's busy indicator will be set).
* The only traps we suport will have set the mailbox.
* We will clear it here.
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
*/
int
slightly clean up the thread_userret() and thread_consider_upcall() calls. also some slight changes for TDF_BOUND testing and small style changes Should ONLY affect KSE programs Submitted by: davidxu
2002-09-23 06:14:30 +00:00
thread_userret(struct thread *td, struct trapframe *frame)
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
{
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
struct kse_upcall *ku;
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
struct proc *p;
Make sure only update wall clock at upcall time, slightly reformat code in kse_relase().
2002-11-18 12:28:15 +00:00
struct timespec ts;
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
int error = 0, uts_crit;
Add files that are new for KSE.
2002-06-29 07:04:59 +00:00
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
/* Nothing to do with bound thread */
Move TDF_SA from td_flags to td_pflags (and rename it accordingly) so that it is no longer necessary to hold sched_lock while manipulating it. Reviewed by: davidxu
2004-06-02 07:52:36 +00:00
if (!(td->td_pflags & TDP_SA))
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
return (0);
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
/*
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
* Update stat clock count for userland
Add kernel support needed for the KSE-aware libpthread: - Use ucontext_t's to store KSE thread state. - Synthesize state for the UTS upon each upcall, rather than saving and copying a trapframe. - Deliver signals to KSE-aware processes via upcall. - Rename kse mailbox structure fields to be more BSD-like. - Store the UTS's stack in struct proc in a stack_t. Reviewed by: bde, deischen, julian Approved by: -arch
2002-09-16 19:26:48 +00:00
*/
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
if (td->td_mailbox != NULL) {
thread_update_usr_ticks(td);
uts_crit = 0;
} else {
uts_crit = 1;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
}
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
p = td->td_proc;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_lock(td);
Remove TDP_USTATCLOCK, we no longer need it because we now always update tick count for userland in thread_userret. This change also removes a "no upcall owned" panic because fuword() schedules an upcall under heavily loaded, and code assumes there is no upcall can occur. Reported and Tested by: Peter Holm <peter@holm.cc>
2004-08-31 11:52:05 +00:00
ku = td->td_upcall;
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
/*
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* Optimisation:
* This thread has not started any upcall.
* If there is no work to report other than ourself,
* then it can return direct to userland.
*/
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
if (TD_CAN_UNBIND(td)) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
Move TDF_CAN_UNBIND to thread private flags td_pflags, this eliminates need of sched_lock in some places. Also in thread_userret, remove spare thread allocation code, it is already done in thread_user_enter. Reviewed by: julian
2004-08-28 04:08:05 +00:00
td->td_pflags &= ~TDP_CAN_UNBIND;
- 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 ((td->td_flags & TDF_NEEDSIGCHK) == 0 &&
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
(p->p_completed == NULL) &&
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
(ku->ku_flags & KUF_DOUPCALL) == 0 &&
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
(p->p_upquantum && ticks < p->p_nextupcall)) {
Export current time when returning from never blocked syscall.
2003-03-14 03:52:16 +00:00
nanotime(&ts);
Adjust code for userland preemptive. Userland can set a quantum in kse_mailbox to schedule an upcall, this is useful for userland timeout routine, for example pthread_cond_timedwait(). Also extract upcall scheduling code from kse_reassign and create a new function called thread_switchout to include these code. Reviewed by: julain
2003-03-19 05:49:38 +00:00
error = copyout(&ts,
Export current time when returning from never blocked syscall.
2003-03-14 03:52:16 +00:00
(caddr_t)&ku->ku_mailbox->km_timeofday,
sizeof(ts));
Fix signal delivering bug for threaded process.
2003-03-11 02:59:50 +00:00
td->td_mailbox = 0;
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
ku->ku_mflags = 0;
Export current time when returning from never blocked syscall.
2003-03-14 03:52:16 +00:00
if (error)
goto out;
Fix signal delivering bug for threaded process.
2003-03-11 02:59:50 +00:00
return (0);
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
}
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
thread_export_context(td, 0);
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
/*
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* There is something to report, and we own an upcall
Fix a typo in a comment. Approved by: re (scottl)
2005-06-23 21:55:43 +00:00
* structure, we can go to userland.
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* Turn ourself into an upcall thread.
Redo how completing threads pass their state to userland if they are not going to cross over themselves. Also change how the list of completed user threads is tracked and passed to the KSE. This is not a change in design but rather the implementation of what was originally envisionned.
2002-09-27 07:11:11 +00:00
*/
1. Migrate TDF_UPCALLING from td_flags to td_pflags. 2. Add a flag TDF_SA, it will be used to distinguish SA based thread from bound thread.
2003-06-15 03:18:58 +00:00
td->td_pflags |= TDP_UPCALLING;
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
} else if (td->td_mailbox && (ku == NULL)) {
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
thread_unlock(td);
o Refine kse_thr_interrupt to allow it to handle different commands. o Remove TDF_NOSIGPOST. o Add a member td_waitset to proc structure, it will be used for sigwait. Tested by: deischen
2003-07-17 22:45:33 +00:00
thread_export_context(td, 1);
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
PROC_LOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if (p->p_upsleeps)
wakeup(&p->p_completed);
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
WITNESS_WARN(WARN_PANIC, &p->p_mtx.lock_object,
Add witness warnings to panic if a thread tries to exit while holding any locks. Requested by: jeff MFC after: 3 days
2005-09-02 20:20:01 +00:00
"thread exiting in userret");
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_flush(&td->td_sigqueue);
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
Fix threaded process job control bug. SMP tested. Reviewed by: julian
2003-03-11 00:07:53 +00:00
thread_stopped(p);
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
thread_exit();
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/* NOTREACHED */
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
} else
thread_unlock(td);
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
1. try to use existing mailbox address in thread_update_usr_ticks. 2. remove '\n' in KASSERT.
2004-08-28 04:16:32 +00:00
KASSERT(ku != NULL, ("upcall is NULL"));
Move thread limits testing code up a bit. This let UPCALLING thread takes possible accumulated contexts away.
2003-02-20 01:11:17 +00:00
KASSERT(TD_CAN_UNBIND(td) == 0, ("can unbind"));
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_LOCK(p);
PROC_SLOCK(p);
Move thread limits testing code up a bit. This let UPCALLING thread takes possible accumulated contexts away.
2003-02-20 01:11:17 +00:00
if (p->p_numthreads > max_threads_per_proc) {
max_threads_hits++;
while (p->p_numthreads > max_threads_per_proc) {
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
if (p->p_numupcalls >= max_threads_per_proc)
Move thread limits testing code up a bit. This let UPCALLING thread takes possible accumulated contexts away.
2003-02-20 01:11:17 +00:00
break;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
If there are signals delivered to current thread, breaks out of loop, userret() will be called again by ast() and thread_userret() will be called again by userret(). Reported by: tegge
2003-06-10 02:21:32 +00:00
if (msleep(&p->p_numthreads, &p->p_mtx, PPAUSE|PCATCH,
Fox a LOR of sleep and sched_lock by using a timeout wait when process reaches maximum number of threads. MFC after: 3 days
2005-09-30 06:09:41 +00:00
"maxthreads", hz/10) != EWOULDBLOCK) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
Fix error in my last commit. Correctly maintain p_maxthrwaits and unlock sched_lock.
2003-06-11 01:08:33 +00:00
break;
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
} else
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
Move thread limits testing code up a bit. This let UPCALLING thread takes possible accumulated contexts away.
2003-02-20 01:11:17 +00:00
}
}
Actually, KSE kernel bits locking is broken and can lead likely to dangerous races. Fix this problems adding correct locking for the members of 'struct kse_upcall' and other struct proc/struct thread related members. For the moment, just leave ku_mflag and ku_flags "lazy" locked. While here, cleanup the code removing the function kse_GC() (unused), and merging upcall_link(), upcall_unlink(), upcall_stash() in their respective callers (static functions, very short and only called in one place). Reported by: pav Tested by: pav (on some pointyhat cluster nodes) Approved by: jeff Approved by: re Sponsorized by: NGX Italy (http://www.ngx.it)
2007-07-23 14:52:22 +00:00
PROC_SUNLOCK(p);
PROC_UNLOCK(p);
Move thread limits testing code up a bit. This let UPCALLING thread takes possible accumulated contexts away.
2003-02-20 01:11:17 +00:00
1. Migrate TDF_UPCALLING from td_flags to td_pflags. 2. Add a flag TDF_SA, it will be used to distinguish SA based thread from bound thread.
2003-06-15 03:18:58 +00:00
if (td->td_pflags & TDP_UPCALLING) {
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
uts_crit = 0;
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
p->p_nextupcall = ticks + p->p_upquantum;
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
/*
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
* There is no more work to do and we are going to ride
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
* this thread up to userland as an upcall.
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
* Do the last parts of the setup needed for the upcall.
*/
CTR3(KTR_PROC, "userret: upcall thread %p (pid %d, %s)",
generally we are interested in what thread did something as opposed to what process. Since threads by default have teh name of the process unless over-written with more useful information, just print the thread name instead.
2007-11-14 06:21:24 +00:00
td, td->td_proc->p_pid, td->td_name);
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
1. Migrate TDF_UPCALLING from td_flags to td_pflags. 2. Add a flag TDF_SA, it will be used to distinguish SA based thread from bound thread.
2003-06-15 03:18:58 +00:00
td->td_pflags &= ~TDP_UPCALLING;
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
if (ku->ku_flags & KUF_DOUPCALL) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
ku->ku_flags &= ~KUF_DOUPCALL;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
1. Add code to support bound thread. when blocked, a bound thread never schedules an upcall. Signal delivering to a bound thread is same as non-threaded process. This is intended to be used by libpthread to implement PTHREAD_SCOPE_SYSTEM thread. 2. Simplify kse_release() a bit, remove sleep loop.
2003-06-15 12:51:26 +00:00
}
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
/*
* Set user context to the UTS
*/
if (!(ku->ku_mflags & KMF_NOUPCALL)) {
Change cpu_set_kse_upcall to more generic style, so we can reuse it in other codes. Add cpu_set_user_tls, use it to tweak user register and setup user TLS. I ever wanted to merge it into cpu_set_kse_upcall, but since cpu_set_kse_upcall is also used by M:N threads which may not need this feature, so I wrote a separated cpu_set_user_tls.
2005-04-23 02:32:32 +00:00
cpu_set_upcall_kse(td, ku->ku_func, ku->ku_mailbox,
&ku->ku_stack);
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
PROC_LOCK(p);
if (p->p_flag & P_TRACED) {
_PHOLD(p);
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 07:33:40 +00:00
ptrace_clear_single_step(td);
Close some races between procfs/ptrace and exit(2): - Reorder the events in exit(2) slightly so that we trigger the S_EXIT stop event earlier. After we have signalled that, we set P_WEXIT and then wait for any processes with a hold on the vmspace via PHOLD to release it. PHOLD now KASSERT()'s that P_WEXIT is clear when it is invoked, and PRELE now does a wakeup if P_WEXIT is set and p_lock drops to zero. - Change proc_rwmem() to require that the processing read from has its vmspace held via PHOLD by the caller and get rid of all the junk to screw around with the vmspace reference count as we no longer need it. - In ptrace() and pseudofs(), treat a process with P_WEXIT set as if it doesn't exist. - Only do one PHOLD in kern_ptrace() now, and do it earlier so it covers FIX_SSTEP() (since on alpha at least this can end up calling proc_rwmem() to clear an earlier single-step simualted via a breakpoint). We only do one to avoid races. Also, by making the EINVAL error for unknown requests be part of the default: case in the switch, the various switch cases can now just break out to return which removes a _lot_ of duplicated PRELE and proc unlocks, etc. Also, it fixes at least one bug where a LWP ptrace command could return EINVAL with the proc lock still held. - Changed the locking for ptrace_single_step(), ptrace_set_pc(), and ptrace_clear_single_step() to always be called with the proc lock held (it was a mixed bag previously). Alpha and arm have to drop the lock while the mess around with breakpoints, but other archs avoid extra lock release/acquires in ptrace(). I did have to fix a couple of other consumers in kern_kse and a few other places to hold the proc lock and PHOLD. Tested by: ps (1 mostly, but some bits of 2-4 as well) MFC after: 1 week
2006-02-22 18:57:50 +00:00
_PRELE(p);
}
PROC_UNLOCK(p);
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 07:33:40 +00:00
error = suword32(&ku->ku_mailbox->km_lwp,
Slight changes to comments and some whitespace changes.
2004-08-09 21:57:30 +00:00
td->td_tid);
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 07:33:40 +00:00
if (error)
goto out;
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
error = suword(&ku->ku_mailbox->km_curthread, 0);
if (error)
goto out;
}
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
/*
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
* Unhook the list of completed threads.
s/Muliple/Multiple Removed whitespace at EOL and EOF.
2004-01-10 18:34:01 +00:00
* anything that completes after this gets to
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
* come in next time.
* Put the list of completed thread mailboxes on
* this KSE's mailbox.
*/
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
if (!(ku->ku_mflags & KMF_NOCOMPLETED) &&
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
(error = thread_link_mboxes(p, ku)) != 0)
Optimize the case when max threads number was hit.
2003-02-19 04:01:55 +00:00
goto out;
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
}
if (!uts_crit) {
Make sure only update wall clock at upcall time, slightly reformat code in kse_relase().
2002-11-18 12:28:15 +00:00
nanotime(&ts);
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
error = copyout(&ts, &ku->ku_mailbox->km_timeofday, sizeof(ts));
Optimize the case when max threads number was hit.
2003-02-19 04:01:55 +00:00
}
out:
if (error) {
/*
Count non-threaded group.
2003-02-19 13:40:24 +00:00
* Things are going to be so screwed we should just kill
* the process.
Optimize the case when max threads number was hit.
2003-02-19 04:01:55 +00:00
* how do we do that?
*/
Restore some code removed in revision 1.193 and 1.194, julian said he'd like to keep these code.
2004-10-06 00:49:41 +00:00
PROC_LOCK(p);
psignal(p, SIGSEGV);
PROC_UNLOCK(p);
} else {
/*
* Optimisation:
* Ensure that we have a spare thread available,
* for when we re-enter the kernel.
*/
if (td->td_standin == NULL)
Fix for the panic("vm_thread_new: kstack allocation failed") and silent NULL pointer dereference in the i386 and sparc64 pmap_pinit() when the kmem_alloc_nofault() failed to allocate address space. Both functions now return error instead of panicing or dereferencing NULL. As consequence, vmspace_exec() and vmspace_unshare() returns the errno int. struct vmspace arg was added to vm_forkproc() to avoid dealing with failed allocation when most of the fork1() job is already done. The kernel stack for the thread is now set up in the thread_alloc(), that itself may return NULL. Also, allocation of the first process thread is performed in the fork1() to properly deal with stack allocation failure. proc_linkup() is separated into proc_linkup() called from fork1(), and proc_linkup0(), that is used to set up the kernel process (was known as swapper). In collaboration with: Peter Holm Reviewed by: jhb
2007-11-05 11:36:16 +00:00
thread_alloc_spare(td); /* XXX care of failure ? */
Make sure only update wall clock at upcall time, slightly reformat code in kse_relase().
2002-11-18 12:28:15 +00:00
}
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
Introduce two flags to control upcall behaviour: o KMF_NOUPCALL Ask kse_release to not return to userland upcall entry, but instead direct returns to userland by using current thread's stack and return address on stack. This flags is intended to be used by UTS in critical region to wait another UTS thread to leave critical region, by using kse_release with this flag to avoid spinnng and burning CPU. Also this flags can be used by UTS to poll completed context when there is nothing to do in userland and needn't restart from its entry like normal upcall. o KMF_NOCOMPLETED Ask kernel to not bring completed thread contexts back to userland when doing upcall, this flags is intend to be used with above flag when an upcall thread is in critical region and can not process completed contexts at that time. Tested by: deischen
2003-04-21 07:27:59 +00:00
ku->ku_mflags = 0;
Add code to ddb to allow backtracing an arbitrary thread. (show thread {address}) Remove the IDLE kse state and replace it with a change in the way threads sahre KSEs. Every KSE now has a thread, which is considered its "owner" however a KSE may also be lent to other threads in the same group to allow completion of in-kernel work. n this case the owner remains the same and the KSE will revert to the owner when the other work has been completed. All creations of upcalls etc. is now done from kse_reassign() which in turn is called from mi_switch or thread_exit(). This means that special code can be removed from msleep() and cv_wait(). kse_release() does not leave a KSE with no thread any more but converts the existing thread into teh KSE's owner, and sets it up for doing an upcall. It is just inhibitted from being scheduled until there is some reason to do an upcall. Remove all trace of the kse_idle queue since it is no-longer needed. "Idle" KSEs are now on the loanable queue.
2002-12-28 01:23:07 +00:00
td->td_mailbox = NULL;
- Split the struct kse into struct upcall and struct kse. struct kse will soon be visible only to schedulers. This greatly simplifies much the KSE code. Submitted by: davidxu
2003-02-17 05:14:26 +00:00
td->td_usticks = 0;
Round out the facilty for a 'bound' thread to loan out its KSE in specific situations. The owner thread must be blocked, and the borrower can not proceed back to user space with the borrowed KSE. The borrower will return the KSE on the next context switch where teh owner wants it back. This removes a lot of possible race conditions and deadlocks. It is consceivable that the borrower should inherit the priority of the owner too. that's another discussion and would be simple to do. Also, as part of this, the "preallocatd spare thread" is attached to the thread doing a syscall rather than the KSE. This removes the need to lock the scheduler when we want to access it, as it's now "at hand". DDB now shows a lot mor info for threaded proceses though it may need some optimisation to squeeze it all back into 80 chars again. (possible JKH project) Upcalls are now "bound" threads, but "KSE Lending" now means that other completing syscalls can be completed using that KSE before the upcall finally makes it back to the UTS. (getting threads OUT OF THE KERNEL is one of the highest priorities in the KSE system.) The upcall when it happens will present all the completed syscalls to the KSE for selection.
2002-10-09 02:33:36 +00:00
return (error); /* go sync */
Add files that are new for KSE.
2002-06-29 07:04:59 +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 07:33:40 +00:00
/*
* called after ptrace resumed a process, force all
* virtual CPUs to schedule upcall for SA process,
* because debugger may have changed something in userland,
* we should notice UTS as soon as possible.
*/
void
thread_continued(struct proc *p)
{
struct kse_upcall *ku;
struct thread *td;
PROC_LOCK_ASSERT(p, MA_OWNED);
Add missing brackets. Noticed by: stefanf@
2005-08-19 22:30:13 +00:00
KASSERT(P_SHOULDSTOP(p), ("process not stopped"));
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 07:33:40 +00:00
if (!(p->p_flag & P_SA))
return;
if (p->p_flag & P_TRACED) {
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
td = TAILQ_FIRST(&p->p_threads);
if (td && (td->td_pflags & TDP_SA)) {
FOREACH_UPCALL_IN_PROC(p, ku) {
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SLOCK(p);
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 07:33:40 +00:00
ku->ku_flags |= KUF_DOUPCALL;
Commit 7/14 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. - Use a global kse spinlock to protect upcall and thread assignment. The per-process spinlock can not be used because this lock must be acquired via mi_switch() where we already hold a thread lock. The kse spinlock is a leaf lock ordered after the process and thread spinlocks. 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 23:54:27 +00:00
PROC_SUNLOCK(p);
Threading cleanup.. part 2 of several. Make part of John Birrell's KSE patch permanent.. Specifically, remove: Any reference of the ksegrp structure. This feature was never fully utilised and made things overly complicated. All code in the scheduler that tried to make threaded programs fair to unthreaded programs. Libpthread processes will already do this to some extent and libthr processes already disable it. Also: Since this makes such a big change to the scheduler(s), take the opportunity to rename some structures and elements that had to be moved anyhow. This makes the code a lot more readable. The ULE scheduler compiles again but I have no idea if it works. The 4bsd scheduler still reqires a little cleaning and some functions that now do ALMOST nothing will go away, but I thought I'd do that as a separate commit. Tested by David Xu, and Dan Eischen using libthr and libpthread.
2006-12-06 06:34:57 +00:00
wakeup(&p->p_completed);
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 07:33:40 +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
#endif
Reference in New Issue Copy Permalink