d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
5a19f6d9d4
freebsd-nq/sys/kern/sys_generic.c

1241 lines
26 KiB
C
Raw Normal View History

BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Copyright (c) 1982, 1986, 1989, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* @(#)sys_generic.c 8.5 (Berkeley) 1/21/94
*/
Use __FBSDID().
2003-06-11 00:56:59 +00:00
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
Converted two options over to the new scheme: USER_LDT and KTRACE.
1996-01-03 21:42:35 +00:00
#include "opt_ktrace.h"
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/param.h>
#include <sys/systm.h>
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#include <sys/sysproto.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/filedesc.h>
Don't include <sys/ioctl.h> in the kernel. Stage 4: include <sys/ttycom.h> and sometimes <sys/filio.h> instead of <sys/ioctl.h> in miscellaneous files. Most of these files have nothing to do with ttys but need to include <sys/ttycom.h> to get the definitions of TIOC[SG]PGRP which are (ab)used to convert F[SG]ETOWN fcntls into ioctls.
1997-03-24 11:52:29 +00:00
#include <sys/filio.h>
Don't #include <sys/fcntl.h> in <sys/file.h> if KERNEL is defined. Fixed everything that depended on getting fcntl.h stuff from the wrong place. Most things don't depend on file.h stuff at all.
1997-03-23 03:37:54 +00:00
#include <sys/fcntl.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/file.h>
#include <sys/proc.h>
All of this is cosmetic. prototypes, #includes, printfs and so on. Makes GCC a lot more silent.
1994-10-02 17:35:40 +00:00
#include <sys/signalvar.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/socketvar.h>
#include <sys/uio.h>
#include <sys/kernel.h>
Deprecate machine/limits.h in favor of new sys/limits.h. Change all in-tree consumers to include <sys/limits.h> Discussed on: standards@ Partially submitted by: Craig Rodrigues <rodrigc@attbi.com>
2003-04-29 13:36:06 +00:00
#include <sys/limits.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#include <sys/malloc.h>
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
#include <sys/poll.h>
The code I picked up from NetBSD in '97 had a nasty bug. It limited the index of the pollfd array to the number of fd's currently open, not the maximum number of fd's. ie: if you had 0,1,2 open, you could not use pollfd slots higher than 20. The specs say we only have to support OPEN_MAX [64] entries but we allow way more than that.
2001-02-07 23:28:01 +00:00
#include <sys/resourcevar.h>
select() DKI is now in <sys/selinfo.h>.
2001-01-09 04:33:49 +00:00
#include <sys/selinfo.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>
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
#include <sys/syscallsubr.h>
Export the nselcoll counter via the kern.nselcoll sysctl so we can see just how bad it gets in various situations. Reminded by: adrian
2000-01-05 19:40:17 +00:00
#include <sys/sysctl.h>
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
#include <sys/sysent.h>
pread/pwrite: follow lseek spirit - return EINVAL on negative offset for non-VCHR
2004-01-20 01:27:42 +00:00
#include <sys/vnode.h>
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
#include <sys/bio.h>
#include <sys/buf.h>
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
#include <sys/condvar.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#ifdef KTRACE
#include <sys/ktrace.h>
#endif
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
#include <vm/vm.h>
#include <vm/vm_page.h>
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Last major round (Unless Bruce thinks of somthing :-) of malloc changes. Distribute all but the most fundamental malloc types. This time I also remembered the trick to making things static: Put "static" in front of them. A couple of finer points by: bde
1997-10-12 20:26:33 +00:00
static MALLOC_DEFINE(M_IOCTLOPS, "ioctlops", "ioctl data buffer");
static MALLOC_DEFINE(M_SELECT, "select", "select() buffer");
MALLOC_DEFINE(M_IOV, "iov", "large iov's");
Distribute and statizice a lot of the malloc M_* types. Substantial input from: bde
1997-10-11 18:31:40 +00:00
Remove __P
2002-03-09 22:44:37 +00:00
static int pollscan(struct thread *, struct pollfd *, u_int);
static int selscan(struct thread *, fd_mask **, fd_mask **, int);
static int dofileread(struct thread *, struct file *, int, void *,
size_t, off_t, int);
static int dofilewrite(struct thread *, struct file *, int,
const void *, size_t, off_t, int);
- Implement selwakeuppri() which allows raising the priority of a thread being waken up. The thread waken up can run at a priority as high as after tsleep(). - Replace selwakeup()s with selwakeuppri()s and pass appropriate priorities. - Add cv_broadcastpri() which raises the priority of the broadcast threads. Used by selwakeuppri() if collision occurs. Not objected in: -arch, -current
2003-11-09 09:17:26 +00:00
static void doselwakeup(struct selinfo *, int);
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Read system call.
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct read_args {
int fd;
Fixed bogotification of pseudocode for syscall args by rev.1.53 of syscalls.master.
1998-09-05 14:30:11 +00:00
void *buf;
size_t nbyte;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
read(td, uap)
struct thread *td;
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct read_args *uap;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct file *fp;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
if ((error = fget_read(td, uap->fd, &fp)) == 0) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = dofileread(td, fp, uap->fd, uap->buf,
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
uap->nbyte, (off_t)-1, 0);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
}
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
/*
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
* Pread system call
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
*/
#ifndef _SYS_SYSPROTO_H_
struct pread_args {
int fd;
void *buf;
size_t nbyte;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
int pad;
off_t offset;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
};
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
pread(td, uap)
struct thread *td;
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct pread_args *uap;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
{
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct file *fp;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
int error;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
make pread use fget_read instead of holdfp.
2002-01-23 08:22:59 +00:00
if ((error = fget_read(td, uap->fd, &fp)) != 0)
return (error);
pread/pwrite: follow lseek spirit - return EINVAL on negative offset for non-VCHR
2004-01-20 01:27:42 +00:00
if (!(fp->f_ops->fo_flags & DFLAG_SEEKABLE))
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
error = ESPIPE;
pread/pwrite: follow lseek spirit - return EINVAL on negative offset for non-VCHR
2004-01-20 01:27:42 +00:00
else if (uap->offset < 0 && fp->f_vnode->v_type != VCHR)
error = EINVAL;
else {
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
error = dofileread(td, fp, uap->fd, uap->buf, uap->nbyte,
uap->offset, FOF_OFFSET);
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
fdrop(fp, td);
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
return(error);
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
}
/*
* Code common for read and pread
*/
Be consistent about "static" functions: if the function is marked static in its prototype, mark it static at the definition too. Inspired by: FlexeLint warning #512
2002-09-28 17:15:38 +00:00
static int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
dofileread(td, fp, fd, buf, nbyte, offset, flags)
struct thread *td;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
struct file *fp;
int fd, flags;
void *buf;
size_t nbyte;
off_t offset;
{
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
struct uio auio;
struct iovec aiov;
long cnt, error = 0;
#ifdef KTRACE
struct iovec ktriov;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
struct uio ktruio;
Distinguish between whether ktraceing was enabled before an IO operation or after it. If the ktrace operation was enabled while the process was blocked doing IO, the race would allow it to pass down invalid (uninitialized) data and panic later down the call stack.
2000-07-27 03:45:18 +00:00
int didktr = 0;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#endif
remove caddr_t
2002-07-22 15:44:27 +00:00
aiov.iov_base = buf;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
aiov.iov_len = nbyte;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
auio.uio_offset = offset;
if (nbyte > INT_MAX)
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
return (EINVAL);
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
auio.uio_resid = nbyte;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
auio.uio_rw = UIO_READ;
auio.uio_segflg = UIO_USERSPACE;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
auio.uio_td = td;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#ifdef KTRACE
/*
* if tracing, save a copy of iovec
*/
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
if (KTRPOINT(td, KTR_GENIO)) {
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
ktriov = aiov;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
ktruio = auio;
Distinguish between whether ktraceing was enabled before an IO operation or after it. If the ktrace operation was enabled while the process was blocked doing IO, the race would allow it to pass down invalid (uninitialized) data and panic later down the call stack.
2000-07-27 03:45:18 +00:00
didktr = 1;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#endif
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
cnt = nbyte;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
In order to better support flexible and extensible access control, make a series of modifications to the credential arguments relating to file read and write operations to cliarfy which credential is used for what: - Change fo_read() and fo_write() to accept "active_cred" instead of "cred", and change the semantics of consumers of fo_read() and fo_write() to pass the active credential of the thread requesting an operation rather than the cached file cred. The cached file cred is still available in fo_read() and fo_write() consumers via fp->f_cred. These changes largely in sys_generic.c. For each implementation of fo_read() and fo_write(), update cred usage to reflect this change and maintain current semantics: - badfo_readwrite() unchanged - kqueue_read/write() unchanged pipe_read/write() now authorize MAC using active_cred rather than td->td_ucred - soo_read/write() unchanged - vn_read/write() now authorize MAC using active_cred but VOP_READ/WRITE() with fp->f_cred Modify vn_rdwr() to accept two credential arguments instead of a single credential: active_cred and file_cred. Use active_cred for MAC authorization, and select a credential for use in VOP_READ/WRITE() based on whether file_cred is NULL or not. If file_cred is provided, authorize the VOP using that cred, otherwise the active credential, matching current semantics. Modify current vn_rdwr() consumers to pass a file_cred if used in the context of a struct file, and to always pass active_cred. When vn_rdwr() is used without a file_cred, pass NOCRED. These changes should maintain current semantics for read/write, but avoid a redundant passing of fp->f_cred, as well as making it more clear what the origin of each credential is in file descriptor read/write operations. Follow-up commits will make similar changes to other file descriptor operations, and modify the MAC framework to pass both credentials to MAC policy modules so they can implement either semantic for revocation. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-15 20:55:08 +00:00
if ((error = fo_read(fp, &auio, td->td_ucred, flags, td))) {
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
if (auio.uio_resid != cnt && (error == ERESTART ||
error == EINTR || error == EWOULDBLOCK))
error = 0;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
cnt -= auio.uio_resid;
#ifdef KTRACE
Distinguish between whether ktraceing was enabled before an IO operation or after it. If the ktrace operation was enabled while the process was blocked doing IO, the race would allow it to pass down invalid (uninitialized) data and panic later down the call stack.
2000-07-27 03:45:18 +00:00
if (didktr && error == 0) {
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
ktruio.uio_iov = &ktriov;
ktruio.uio_resid = cnt;
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
ktrgenio(fd, UIO_READ, &ktruio, error);
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#endif
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_retval[0] = cnt;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
return (error);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Scatter read system call.
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct readv_args {
Fixed the type of readv(). An args struct member name conflicted with the machine-generated one in <sys/sysproto.h>.
1995-11-11 06:57:34 +00:00
int fd;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct iovec *iovp;
u_int iovcnt;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
readv(td, uap)
struct thread *td;
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct readv_args *uap;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct file *fp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct uio auio;
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct iovec *iov;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct iovec *needfree;
struct iovec aiov[UIO_SMALLIOV];
o Correct an error made in revision 1.65: In readv(), if uap->iovcnt is out-of-range, drop the file reference before returning. (This error also exists in the RELENG_4 branch.) o Eliminate the acquisition and release of Giant in readv() now that malloc() and free() are callable without Giant.
2002-05-09 02:30:41 +00:00
long i, cnt;
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
u_int iovlen;
#ifdef KTRACE
struct iovec *ktriov = NULL;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
struct uio ktruio;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#endif
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
if ((error = fget_read(td, uap->fd, &fp)) != 0)
o Correct an error made in revision 1.65: In readv(), if uap->iovcnt is out-of-range, drop the file reference before returning. (This error also exists in the RELENG_4 branch.) o Eliminate the acquisition and release of Giant in readv() now that malloc() and free() are callable without Giant.
2002-05-09 02:30:41 +00:00
return (error);
needfree = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* note: can't use iovlen until iovcnt is validated */
iovlen = uap->iovcnt * sizeof (struct iovec);
if (uap->iovcnt > UIO_SMALLIOV) {
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
if (uap->iovcnt > UIO_MAXIOV) {
error = EINVAL;
o Correct an error made in revision 1.65: In readv(), if uap->iovcnt is out-of-range, drop the file reference before returning. (This error also exists in the RELENG_4 branch.) o Eliminate the acquisition and release of Giant in readv() now that malloc() and free() are callable without Giant.
2002-05-09 02:30:41 +00:00
goto done;
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
}
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
needfree = iov;
o Correct an error made in revision 1.65: In readv(), if uap->iovcnt is out-of-range, drop the file reference before returning. (This error also exists in the RELENG_4 branch.) o Eliminate the acquisition and release of Giant in readv() now that malloc() and free() are callable without Giant.
2002-05-09 02:30:41 +00:00
} else
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
iov = aiov;
auio.uio_iov = iov;
auio.uio_iovcnt = uap->iovcnt;
auio.uio_rw = UIO_READ;
auio.uio_segflg = UIO_USERSPACE;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
auio.uio_td = td;
Modifications to existing files to support the initial AIO/LIO and kernel based threading support.
1997-06-16 00:29:36 +00:00
auio.uio_offset = -1;
remove caddr_t
2002-07-22 15:44:27 +00:00
if ((error = copyin(uap->iovp, iov, iovlen)))
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto done;
auio.uio_resid = 0;
for (i = 0; i < uap->iovcnt; i++) {
Change various syscalls to use size_t arguments instead of u_int. Add some overflow checks to read/write (from bde). Change all modifications to vm_page::flags, vm_page::busy, vm_object::flags and vm_object::paging_in_progress to use operations which are not interruptable. Reviewed by: Bruce Evans <bde@zeta.org.au>
1998-08-24 08:39:39 +00:00
if (iov->iov_len > INT_MAX - auio.uio_resid) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = EINVAL;
goto done;
}
Change various syscalls to use size_t arguments instead of u_int. Add some overflow checks to read/write (from bde). Change all modifications to vm_page::flags, vm_page::busy, vm_object::flags and vm_object::paging_in_progress to use operations which are not interruptable. Reviewed by: Bruce Evans <bde@zeta.org.au>
1998-08-24 08:39:39 +00:00
auio.uio_resid += iov->iov_len;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
iov++;
}
#ifdef KTRACE
/*
* if tracing, save a copy of iovec
*/
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
if (KTRPOINT(td, KTR_GENIO)) {
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
MALLOC(ktriov, struct iovec *, iovlen, M_TEMP, M_WAITOK);
remove caddr_t
2002-07-22 15:44:27 +00:00
bcopy(auio.uio_iov, ktriov, iovlen);
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
ktruio = auio;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
#endif
cnt = auio.uio_resid;
In order to better support flexible and extensible access control, make a series of modifications to the credential arguments relating to file read and write operations to cliarfy which credential is used for what: - Change fo_read() and fo_write() to accept "active_cred" instead of "cred", and change the semantics of consumers of fo_read() and fo_write() to pass the active credential of the thread requesting an operation rather than the cached file cred. The cached file cred is still available in fo_read() and fo_write() consumers via fp->f_cred. These changes largely in sys_generic.c. For each implementation of fo_read() and fo_write(), update cred usage to reflect this change and maintain current semantics: - badfo_readwrite() unchanged - kqueue_read/write() unchanged pipe_read/write() now authorize MAC using active_cred rather than td->td_ucred - soo_read/write() unchanged - vn_read/write() now authorize MAC using active_cred but VOP_READ/WRITE() with fp->f_cred Modify vn_rdwr() to accept two credential arguments instead of a single credential: active_cred and file_cred. Use active_cred for MAC authorization, and select a credential for use in VOP_READ/WRITE() based on whether file_cred is NULL or not. If file_cred is provided, authorize the VOP using that cred, otherwise the active credential, matching current semantics. Modify current vn_rdwr() consumers to pass a file_cred if used in the context of a struct file, and to always pass active_cred. When vn_rdwr() is used without a file_cred, pass NOCRED. These changes should maintain current semantics for read/write, but avoid a redundant passing of fp->f_cred, as well as making it more clear what the origin of each credential is in file descriptor read/write operations. Follow-up commits will make similar changes to other file descriptor operations, and modify the MAC framework to pass both credentials to MAC policy modules so they can implement either semantic for revocation. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-15 20:55:08 +00:00
if ((error = fo_read(fp, &auio, td->td_ucred, 0, td))) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (auio.uio_resid != cnt && (error == ERESTART ||
error == EINTR || error == EWOULDBLOCK))
error = 0;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
cnt -= auio.uio_resid;
#ifdef KTRACE
if (ktriov != NULL) {
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
if (error == 0) {
ktruio.uio_iov = ktriov;
ktruio.uio_resid = cnt;
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
ktrgenio(uap->fd, UIO_READ, &ktruio, error);
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
FREE(ktriov, M_TEMP);
}
#endif
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_retval[0] = cnt;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
done:
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (needfree)
FREE(needfree, M_IOV);
return (error);
}
/*
* Write system call
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct write_args {
int fd;
Fixed bogotification of pseudocode for syscall args by rev.1.53 of syscalls.master.
1998-09-05 14:30:11 +00:00
const void *buf;
size_t nbyte;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
write(td, uap)
struct thread *td;
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct write_args *uap;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct file *fp;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
int error;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
if ((error = fget_write(td, uap->fd, &fp)) == 0) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = dofilewrite(td, fp, uap->fd, uap->buf, uap->nbyte,
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
(off_t)-1, 0);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
} else {
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
error = EBADF; /* XXX this can't be right */
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
}
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
return(error);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
/*
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
* Pwrite system call
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
*/
#ifndef _SYS_SYSPROTO_H_
struct pwrite_args {
int fd;
const void *buf;
size_t nbyte;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
int pad;
off_t offset;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
};
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
pwrite(td, uap)
struct thread *td;
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct pwrite_args *uap;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
{
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct file *fp;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
int error;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
if ((error = fget_write(td, uap->fd, &fp)) == 0) {
pread/pwrite: follow lseek spirit - return EINVAL on negative offset for non-VCHR
2004-01-20 01:27:42 +00:00
if (!(fp->f_ops->fo_flags & DFLAG_SEEKABLE))
Introduce a new flag on a file descriptor: DFLAG_SEEKABLE and use that rather than assume that only DTYPE_VNODE is seekable.
2003-06-18 19:53:59 +00:00
error = ESPIPE;
pread/pwrite: follow lseek spirit - return EINVAL on negative offset for non-VCHR
2004-01-20 01:27:42 +00:00
else if (uap->offset < 0 && fp->f_vnode->v_type != VCHR)
error = EINVAL;
else {
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
error = dofilewrite(td, fp, uap->fd, uap->buf,
uap->nbyte, uap->offset, FOF_OFFSET);
}
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
} else {
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
error = EBADF; /* this can't be right */
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
return(error);
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
}
static int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
dofilewrite(td, fp, fd, buf, nbyte, offset, flags)
struct thread *td;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
struct file *fp;
int fd, flags;
const void *buf;
size_t nbyte;
off_t offset;
{
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
struct uio auio;
struct iovec aiov;
long cnt, error = 0;
#ifdef KTRACE
struct iovec ktriov;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
struct uio ktruio;
Distinguish between whether ktraceing was enabled before an IO operation or after it. If the ktrace operation was enabled while the process was blocked doing IO, the race would allow it to pass down invalid (uninitialized) data and panic later down the call stack.
2000-07-27 03:45:18 +00:00
int didktr = 0;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#endif
Fix a warning that has been annoying me for some time: "kern/sys_generic.c:358: warning: cast discards qualifiers from pointer target type" The idea for using the uintptr_t intermediate cast for de-constifying a pointer was hinted at by bde some time ago.
2000-07-28 22:17:42 +00:00
aiov.iov_base = (void *)(uintptr_t)buf;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
aiov.iov_len = nbyte;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
auio.uio_iov = &aiov;
auio.uio_iovcnt = 1;
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
auio.uio_offset = offset;
if (nbyte > INT_MAX)
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
return (EINVAL);
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
auio.uio_resid = nbyte;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
auio.uio_rw = UIO_WRITE;
auio.uio_segflg = UIO_USERSPACE;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
auio.uio_td = td;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#ifdef KTRACE
/*
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
* if tracing, save a copy of iovec and uio
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
*/
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
if (KTRPOINT(td, KTR_GENIO)) {
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
ktriov = aiov;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
ktruio = auio;
Distinguish between whether ktraceing was enabled before an IO operation or after it. If the ktrace operation was enabled while the process was blocked doing IO, the race would allow it to pass down invalid (uninitialized) data and panic later down the call stack.
2000-07-27 03:45:18 +00:00
didktr = 1;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#endif
Add standard padding argument to pread and pwrite syscall. That should make them NetBSD compatible. Add parameter to fo_read and fo_write. (The only flag FOF_OFFSET mean that the offset is set in the struct uio). Factor out some common code from read/pread/write/pwrite syscalls.
1999-04-04 21:41:28 +00:00
cnt = nbyte;
only call bwillwrite() to stall on IO when dealing with VNODEs otherwise we will stall on non-disk IO for things like fifos and sockets
2000-11-30 20:23:14 +00:00
if (fp->f_type == DTYPE_VNODE)
bwillwrite();
In order to better support flexible and extensible access control, make a series of modifications to the credential arguments relating to file read and write operations to cliarfy which credential is used for what: - Change fo_read() and fo_write() to accept "active_cred" instead of "cred", and change the semantics of consumers of fo_read() and fo_write() to pass the active credential of the thread requesting an operation rather than the cached file cred. The cached file cred is still available in fo_read() and fo_write() consumers via fp->f_cred. These changes largely in sys_generic.c. For each implementation of fo_read() and fo_write(), update cred usage to reflect this change and maintain current semantics: - badfo_readwrite() unchanged - kqueue_read/write() unchanged pipe_read/write() now authorize MAC using active_cred rather than td->td_ucred - soo_read/write() unchanged - vn_read/write() now authorize MAC using active_cred but VOP_READ/WRITE() with fp->f_cred Modify vn_rdwr() to accept two credential arguments instead of a single credential: active_cred and file_cred. Use active_cred for MAC authorization, and select a credential for use in VOP_READ/WRITE() based on whether file_cred is NULL or not. If file_cred is provided, authorize the VOP using that cred, otherwise the active credential, matching current semantics. Modify current vn_rdwr() consumers to pass a file_cred if used in the context of a struct file, and to always pass active_cred. When vn_rdwr() is used without a file_cred, pass NOCRED. These changes should maintain current semantics for read/write, but avoid a redundant passing of fp->f_cred, as well as making it more clear what the origin of each credential is in file descriptor read/write operations. Follow-up commits will make similar changes to other file descriptor operations, and modify the MAC framework to pass both credentials to MAC policy modules so they can implement either semantic for revocation. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-15 20:55:08 +00:00
if ((error = fo_write(fp, &auio, td->td_ucred, flags, td))) {
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
if (auio.uio_resid != cnt && (error == ERESTART ||
error == EINTR || error == EWOULDBLOCK))
error = 0;
Implement SO_NOSIGPIPE option for sockets. This allows one to request that an EPIPE error return not generate SIGPIPE on sockets. Submitted by: lioux Inspired by: Darwin
2002-06-20 18:52:54 +00:00
/* Socket layer is responsible for issuing SIGPIPE. */
if (error == EPIPE && fp->f_type != DTYPE_SOCKET) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
PROC_LOCK(td->td_proc);
psignal(td->td_proc, SIGPIPE);
PROC_UNLOCK(td->td_proc);
Grab the process lock while calling psignal and before calling psignal.
2001-03-07 03:37:06 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
}
cnt -= auio.uio_resid;
#ifdef KTRACE
Distinguish between whether ktraceing was enabled before an IO operation or after it. If the ktrace operation was enabled while the process was blocked doing IO, the race would allow it to pass down invalid (uninitialized) data and panic later down the call stack.
2000-07-27 03:45:18 +00:00
if (didktr && error == 0) {
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
ktruio.uio_iov = &ktriov;
ktruio.uio_resid = cnt;
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
ktrgenio(fd, UIO_WRITE, &ktruio, error);
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
}
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
#endif
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_retval[0] = cnt;
Added pread and pwrite. These functions are defined by the X/Open Threads Extension. (Note: We use the same syscall numbers as NetBSD.) Submitted by: John Plevyak <jplevyak@inktomi.com>
1999-03-27 21:16:58 +00:00
return (error);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Gather write system call
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct writev_args {
int fd;
struct iovec *iovp;
u_int iovcnt;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
writev(td, uap)
struct thread *td;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
register struct writev_args *uap;
{
remove holdfp() Replace uses of holdfp() with fget*() or fgetvp*() calls as appropriate introduce fget(), fget_read(), fget_write() - these functions will take a thread and file descriptor and return a file pointer with its ref count bumped. introduce fgetvp(), fgetvp_read(), fgetvp_write() - these functions will take a thread and file descriptor and return a vref()'d vnode. *_read() requires that the file pointer be FREAD, *_write that it be FWRITE. This continues the cleanup of struct filedesc and struct file access routines which, when are all through with it, will allow us to then make the API calls MP safe and be able to move Giant down into the fo_* functions.
2001-11-14 06:30:36 +00:00
struct file *fp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct uio auio;
register struct iovec *iov;
struct iovec *needfree;
struct iovec aiov[UIO_SMALLIOV];
long i, cnt, error = 0;
u_int iovlen;
#ifdef KTRACE
struct iovec *ktriov = NULL;
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
struct uio ktruio;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#endif
Remove Giant from writev(2). Eliminate trivial style differences between writev(2) and readv(2).
2003-08-01 02:21:54 +00:00
if ((error = fget_write(td, uap->fd, &fp)) != 0)
return (EBADF);
needfree = NULL;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* note: can't use iovlen until iovcnt is validated */
iovlen = uap->iovcnt * sizeof (struct iovec);
if (uap->iovcnt > UIO_SMALLIOV) {
Fix a race condition with shared fd tables and writev(). It's still not safe to consider file table sharing secure. Submitted by: Ville-Pertti Keinonen <will@iki.fi>
1999-10-14 05:37:52 +00:00
if (uap->iovcnt > UIO_MAXIOV) {
error = EINVAL;
goto done;
}
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
MALLOC(iov, struct iovec *, iovlen, M_IOV, M_WAITOK);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
needfree = iov;
Remove Giant from writev(2). Eliminate trivial style differences between writev(2) and readv(2).
2003-08-01 02:21:54 +00:00
} else
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
iov = aiov;
auio.uio_iov = iov;
auio.uio_iovcnt = uap->iovcnt;
auio.uio_rw = UIO_WRITE;
auio.uio_segflg = UIO_USERSPACE;
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
auio.uio_td = td;
Modifications to existing files to support the initial AIO/LIO and kernel based threading support.
1997-06-16 00:29:36 +00:00
auio.uio_offset = -1;
remove caddr_t
2002-07-22 15:44:27 +00:00
if ((error = copyin(uap->iovp, iov, iovlen)))
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto done;
auio.uio_resid = 0;
for (i = 0; i < uap->iovcnt; i++) {
Change various syscalls to use size_t arguments instead of u_int. Add some overflow checks to read/write (from bde). Change all modifications to vm_page::flags, vm_page::busy, vm_object::flags and vm_object::paging_in_progress to use operations which are not interruptable. Reviewed by: Bruce Evans <bde@zeta.org.au>
1998-08-24 08:39:39 +00:00
if (iov->iov_len > INT_MAX - auio.uio_resid) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = EINVAL;
goto done;
}
Change various syscalls to use size_t arguments instead of u_int. Add some overflow checks to read/write (from bde). Change all modifications to vm_page::flags, vm_page::busy, vm_object::flags and vm_object::paging_in_progress to use operations which are not interruptable. Reviewed by: Bruce Evans <bde@zeta.org.au>
1998-08-24 08:39:39 +00:00
auio.uio_resid += iov->iov_len;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
iov++;
}
#ifdef KTRACE
/*
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
* if tracing, save a copy of iovec and uio
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*/
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
if (KTRPOINT(td, KTR_GENIO)) {
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
MALLOC(ktriov, struct iovec *, iovlen, M_TEMP, M_WAITOK);
remove caddr_t
2002-07-22 15:44:27 +00:00
bcopy(auio.uio_iov, ktriov, iovlen);
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
ktruio = auio;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
#endif
cnt = auio.uio_resid;
Only call bwillwrite() for vnodes. Do not penalize devices or pipes.
2000-12-07 23:45:57 +00:00
if (fp->f_type == DTYPE_VNODE)
bwillwrite();
In order to better support flexible and extensible access control, make a series of modifications to the credential arguments relating to file read and write operations to cliarfy which credential is used for what: - Change fo_read() and fo_write() to accept "active_cred" instead of "cred", and change the semantics of consumers of fo_read() and fo_write() to pass the active credential of the thread requesting an operation rather than the cached file cred. The cached file cred is still available in fo_read() and fo_write() consumers via fp->f_cred. These changes largely in sys_generic.c. For each implementation of fo_read() and fo_write(), update cred usage to reflect this change and maintain current semantics: - badfo_readwrite() unchanged - kqueue_read/write() unchanged pipe_read/write() now authorize MAC using active_cred rather than td->td_ucred - soo_read/write() unchanged - vn_read/write() now authorize MAC using active_cred but VOP_READ/WRITE() with fp->f_cred Modify vn_rdwr() to accept two credential arguments instead of a single credential: active_cred and file_cred. Use active_cred for MAC authorization, and select a credential for use in VOP_READ/WRITE() based on whether file_cred is NULL or not. If file_cred is provided, authorize the VOP using that cred, otherwise the active credential, matching current semantics. Modify current vn_rdwr() consumers to pass a file_cred if used in the context of a struct file, and to always pass active_cred. When vn_rdwr() is used without a file_cred, pass NOCRED. These changes should maintain current semantics for read/write, but avoid a redundant passing of fp->f_cred, as well as making it more clear what the origin of each credential is in file descriptor read/write operations. Follow-up commits will make similar changes to other file descriptor operations, and modify the MAC framework to pass both credentials to MAC policy modules so they can implement either semantic for revocation. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-15 20:55:08 +00:00
if ((error = fo_write(fp, &auio, td->td_ucred, 0, td))) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (auio.uio_resid != cnt && (error == ERESTART ||
error == EINTR || error == EWOULDBLOCK))
error = 0;
Grab the process lock while calling psignal and before calling psignal.
2001-03-07 03:37:06 +00:00
if (error == EPIPE) {
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
PROC_LOCK(td->td_proc);
psignal(td->td_proc, SIGPIPE);
PROC_UNLOCK(td->td_proc);
Grab the process lock while calling psignal and before calling psignal.
2001-03-07 03:37:06 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
cnt -= auio.uio_resid;
#ifdef KTRACE
if (ktriov != NULL) {
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
if (error == 0) {
ktruio.uio_iov = ktriov;
ktruio.uio_resid = cnt;
Catch up to changes in ktrace API.
2002-06-07 05:37:18 +00:00
ktrgenio(uap->fd, UIO_WRITE, &ktruio, error);
Modify ktrace's general I/O tracing, ktrgenio(), to use a struct uio * instead of a struct iovec * array and int len. Get rid of stupidly trying to allocate all of the memory and copyin()ing the entire iovec[], and instead just do the proper VOP_WRITE() in ktrwrite() using a copy of the struct uio that the syscall originally used. This solves the DoS which could easily be performed; to work around the DoS, one could also remove "options KTRACE" from the kernel. This is a very strong MFC candidate for 4.1. Found by: art@OpenBSD.org
2000-07-02 08:08:09 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
FREE(ktriov, M_TEMP);
}
#endif
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_retval[0] = cnt;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
done:
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (needfree)
FREE(needfree, M_IOV);
return (error);
}
/*
* Ioctl system call
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct ioctl_args {
int fd;
Change various syscalls to use size_t arguments instead of u_int. Add some overflow checks to read/write (from bde). Change all modifications to vm_page::flags, vm_page::busy, vm_object::flags and vm_object::paging_in_progress to use operations which are not interruptable. Reviewed by: Bruce Evans <bde@zeta.org.au>
1998-08-24 08:39:39 +00:00
u_long com;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
caddr_t data;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* ARGSUSED */
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
ioctl(td, uap)
struct thread *td;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
register struct ioctl_args *uap;
{
Replace ffind_* with fget calls. Make fget MPsafe. Make fgetvp and fgetsock use the fget subsystem to reduce code bloat. Push giant down in fpathconf().
2002-01-14 00:13:45 +00:00
struct file *fp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
register struct filedesc *fdp;
64bit fixes: use u_long not int for ioctl command.
1998-06-10 10:29:31 +00:00
register u_long com;
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
int error = 0;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
register u_int size;
caddr_t data, memp;
int tmp;
#define STK_PARAMS 128
Some ioctl routines assume that the ioctl buffer is aligned, but a char[] declaration makes no such guarentee. A union is used to force alignment of the char buffer.
2000-05-09 17:43:21 +00:00
union {
char stkbuf[STK_PARAMS];
long align;
} ubuf;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Replace ffind_* with fget calls. Make fget MPsafe. Make fgetvp and fgetsock use the fget subsystem to reduce code bloat. Push giant down in fpathconf().
2002-01-14 00:13:45 +00:00
if ((error = fget(td, uap->fd, &fp)) != 0)
return (error);
undo a bit of the Giant pushdown. fdrop isn't SMP safe as it may call into the file's close routine which definetly is not SMP safe right now, so we hold Giant over calls to fdrop now.
2002-01-19 01:03:54 +00:00
mtx_lock(&Giant);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
if ((fp->f_flag & (FREAD | FWRITE)) == 0) {
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
fdrop(fp, td);
undo a bit of the Giant pushdown. fdrop isn't SMP safe as it may call into the file's close routine which definetly is not SMP safe right now, so we hold Giant over calls to fdrop now.
2002-01-19 01:03:54 +00:00
mtx_unlock(&Giant);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
return (EBADF);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
}
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
fdp = td->td_proc->p_fd;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
switch (com = uap->com) {
case FIONCLEX:
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILEDESC_LOCK(fdp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fdp->fd_ofileflags[uap->fd] &= ~UF_EXCLOSE;
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILEDESC_UNLOCK(fdp);
fdrop(fp, td);
undo a bit of the Giant pushdown. fdrop isn't SMP safe as it may call into the file's close routine which definetly is not SMP safe right now, so we hold Giant over calls to fdrop now.
2002-01-19 01:03:54 +00:00
mtx_unlock(&Giant);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
return (0);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
case FIOCLEX:
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILEDESC_LOCK(fdp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fdp->fd_ofileflags[uap->fd] |= UF_EXCLOSE;
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILEDESC_UNLOCK(fdp);
fdrop(fp, td);
undo a bit of the Giant pushdown. fdrop isn't SMP safe as it may call into the file's close routine which definetly is not SMP safe right now, so we hold Giant over calls to fdrop now.
2002-01-19 01:03:54 +00:00
mtx_unlock(&Giant);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
return (0);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
/*
* Interpret high order word to find amount of data to be
* copied to/from the user's address space.
*/
size = IOCPARM_LEN(com);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
if (size > IOCPARM_MAX) {
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
fdrop(fp, td);
undo a bit of the Giant pushdown. fdrop isn't SMP safe as it may call into the file's close routine which definetly is not SMP safe right now, so we hold Giant over calls to fdrop now.
2002-01-19 01:03:54 +00:00
mtx_unlock(&Giant);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
return (ENOTTY);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
}
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
memp = NULL;
Some ioctl routines assume that the ioctl buffer is aligned, but a char[] declaration makes no such guarentee. A union is used to force alignment of the char buffer.
2000-05-09 17:43:21 +00:00
if (size > sizeof (ubuf.stkbuf)) {
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
memp = malloc((u_long)size, M_IOCTLOPS, M_WAITOK);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
data = memp;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
} else {
Some ioctl routines assume that the ioctl buffer is aligned, but a char[] declaration makes no such guarentee. A union is used to force alignment of the char buffer.
2000-05-09 17:43:21 +00:00
data = ubuf.stkbuf;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (com&IOC_IN) {
if (size) {
error = copyin(uap->data, data, (u_int)size);
if (error) {
if (memp)
free(memp, M_IOCTLOPS);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
goto done;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
} else {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*(caddr_t *)data = uap->data;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
} else if ((com&IOC_OUT) && size) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Zero the buffer so the user always
* gets back something deterministic.
*/
bzero(data, size);
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
} else if (com&IOC_VOID) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
*(caddr_t *)data = uap->data;
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
switch (com) {
case FIONBIO:
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILE_LOCK(fp);
While in the real world, I had a bad case of being swapped out for a lot of cycles. While waiting there I added a lot of the extra ()'s I have, (I have never used LISP to any extent). So I compiled the kernel with -Wall and shut up a lot of "suggest you add ()'s", removed a bunch of unused var's and added a couple of declarations here and there. Having a lap-top is highly recommended. My kernel still runs, yell at me if you kernel breaks.
1994-09-25 19:34:02 +00:00
if ((tmp = *(int *)data))
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fp->f_flag |= FNONBLOCK;
else
fp->f_flag &= ~FNONBLOCK;
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILE_UNLOCK(fp);
In continuation of early fileop credential changes, modify fo_ioctl() to accept an 'active_cred' argument reflecting the credential of the thread initiating the ioctl operation. - Change fo_ioctl() to accept active_cred; change consumers of the fo_ioctl() interface to generally pass active_cred from td->td_ucred. - In fifofs, initialize filetmp.f_cred to ap->a_cred so that the invocations of soo_ioctl() are provided access to the calling f_cred. Pass ap->a_td->td_ucred as the active_cred, but note that this is required because we don't yet distinguish file_cred and active_cred in invoking VOP's. - Update kqueue_ioctl() for its new argument. - Update pipe_ioctl() for its new argument, pass active_cred rather than td_ucred to MAC for authorization. - Update soo_ioctl() for its new argument. - Update vn_ioctl() for its new argument, use active_cred rather than td->td_ucred to authorize VOP_IOCTL() and the associated VOP_GETATTR(). Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-17 02:36:16 +00:00
error = fo_ioctl(fp, FIONBIO, &tmp, td->td_ucred, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
case FIOASYNC:
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILE_LOCK(fp);
While in the real world, I had a bad case of being swapped out for a lot of cycles. While waiting there I added a lot of the extra ()'s I have, (I have never used LISP to any extent). So I compiled the kernel with -Wall and shut up a lot of "suggest you add ()'s", removed a bunch of unused var's and added a couple of declarations here and there. Having a lap-top is highly recommended. My kernel still runs, yell at me if you kernel breaks.
1994-09-25 19:34:02 +00:00
if ((tmp = *(int *)data))
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fp->f_flag |= FASYNC;
else
fp->f_flag &= ~FASYNC;
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILE_UNLOCK(fp);
In continuation of early fileop credential changes, modify fo_ioctl() to accept an 'active_cred' argument reflecting the credential of the thread initiating the ioctl operation. - Change fo_ioctl() to accept active_cred; change consumers of the fo_ioctl() interface to generally pass active_cred from td->td_ucred. - In fifofs, initialize filetmp.f_cred to ap->a_cred so that the invocations of soo_ioctl() are provided access to the calling f_cred. Pass ap->a_td->td_ucred as the active_cred, but note that this is required because we don't yet distinguish file_cred and active_cred in invoking VOP's. - Update kqueue_ioctl() for its new argument. - Update pipe_ioctl() for its new argument, pass active_cred rather than td_ucred to MAC for authorization. - Update soo_ioctl() for its new argument. - Update vn_ioctl() for its new argument, use active_cred rather than td->td_ucred to authorize VOP_IOCTL() and the associated VOP_GETATTR(). Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-17 02:36:16 +00:00
error = fo_ioctl(fp, FIOASYNC, &tmp, td->td_ucred, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
break;
default:
In continuation of early fileop credential changes, modify fo_ioctl() to accept an 'active_cred' argument reflecting the credential of the thread initiating the ioctl operation. - Change fo_ioctl() to accept active_cred; change consumers of the fo_ioctl() interface to generally pass active_cred from td->td_ucred. - In fifofs, initialize filetmp.f_cred to ap->a_cred so that the invocations of soo_ioctl() are provided access to the calling f_cred. Pass ap->a_td->td_ucred as the active_cred, but note that this is required because we don't yet distinguish file_cred and active_cred in invoking VOP's. - Update kqueue_ioctl() for its new argument. - Update pipe_ioctl() for its new argument, pass active_cred rather than td_ucred to MAC for authorization. - Update soo_ioctl() for its new argument. - Update vn_ioctl() for its new argument, use active_cred rather than td->td_ucred to authorize VOP_IOCTL() and the associated VOP_GETATTR(). Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-17 02:36:16 +00:00
error = fo_ioctl(fp, com, data, td->td_ucred, td);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Copy any data to user, size was
* already set and checked above.
*/
if (error == 0 && (com&IOC_OUT) && size)
error = copyout(data, uap->data, (u_int)size);
break;
}
if (memp)
free(memp, M_IOCTLOPS);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
fdrop(fp, td);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
done:
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
mtx_unlock(&Giant);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
/*
* sellock and selwait are initialized in selectinit() via SYSINIT.
*/
struct mtx sellock;
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
struct cv selwait;
Make nselcol, the number of select collisions since boot, unsigned as negative collisions simply doesn't make sense. PR: (one small part of) 19720 Approved by: alfred
2002-06-12 02:08:18 +00:00
u_int nselcoll; /* Select collisions since boot */
SYSCTL_UINT(_kern, OID_AUTO, nselcoll, CTLFLAG_RD, &nselcoll, 0, "");
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Select system call.
*/
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#ifndef _SYS_SYSPROTO_H_
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct select_args {
Remove the kernel FD_SETSIZE limit for select(). Make select()'s first argument 'int' not 'u_int'. Reviewed by: bde
1996-08-20 07:18:10 +00:00
int nd;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
fd_set *in, *ou, *ex;
struct timeval *tv;
};
Included <sys/sysproto.h> to get central declarations for syscall args structs and prototypes for syscalls. Ifdefed duplicated decentralized declarations of args structs. It's convenient to have this visible but they are hard to maintain. Some are already different from the central declarations. 4.4lite2 puts them in comments in the function headers but I wanted to avoid the large changes for that.
1995-11-12 06:43:28 +00:00
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch. Reviewed by: Rodney W. Grimes Submitted by: John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
select(td, uap)
register struct thread *td;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
register struct select_args *uap;
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
{
struct timeval tv, *tvp;
int error;
if (uap->tv != NULL) {
error = copyin(uap->tv, &tv, sizeof(tv));
if (error)
return (error);
tvp = &tv;
} else
tvp = NULL;
return (kern_select(td, uap->nd, uap->in, uap->ou, uap->ex, tvp));
}
int
kern_select(struct thread *td, int nd, fd_set *fd_in, fd_set *fd_ou,
fd_set *fd_ex, struct timeval *tvp)
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
struct filedesc *fdp;
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
/*
* The magic 2048 here is chosen to be just enough for FD_SETSIZE
* infds with the new FD_SETSIZE of 1024, and more than enough for
* FD_SETSIZE infds, outfds and exceptfds with the old FD_SETSIZE
* of 256.
*/
fd_mask s_selbits[howmany(2048, NFDBITS)];
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
fd_mask *ibits[3], *obits[3], *selbits, *sbp;
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
struct timeval atv, rtv, ttv;
Make nselcol, the number of select collisions since boot, unsigned as negative collisions simply doesn't make sense. PR: (one small part of) 19720 Approved by: alfred
2002-06-12 02:08:18 +00:00
int error, timo;
u_int ncoll, nbufbytes, ncpbytes, nfdbits;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (nd < 0)
Fix a minor style error in my code.
1996-08-20 15:03:41 +00:00
return (EINVAL);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
fdp = td->td_proc->p_fd;
Add annotations to mtx_lock(&Giant) in kern_select() and poll() that we always grab Giant, even if we're actually only polling objects that don't require giant. Once socket locking is merged, there will be strong motivation to fix this.
2004-03-13 05:58:57 +00:00
/*
* XXX: kern_select() currently requires that we acquire Giant
* even if none of the file descriptors we poll requires Giant.
*/
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
mtx_lock(&Giant);
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILEDESC_LOCK(fdp);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (nd > td->td_proc->p_fd->fd_nfiles)
nd = td->td_proc->p_fd->fd_nfiles; /* forgiving; slightly wrong */
SMP Lock struct file, filedesc and the global file list. Seigo Tanimura (tanimura) posted the initial delta. I've polished it quite a bit reducing the need for locking and adapting it for KSE. Locks: 1 mutex in each filedesc protects all the fields. protects "struct file" initialization, while a struct file is being changed from &badfileops -> &pipeops or something the filedesc should be locked. 1 mutex in each struct file protects the refcount fields. doesn't protect anything else. the flags used for garbage collection have been moved to f_gcflag which was the FILLER short, this doesn't need locking because the garbage collection is a single threaded container. could likely be made to use a pool mutex. 1 sx lock for the global filelist. struct file * fhold(struct file *fp); /* increments reference count on a file */ struct file * fhold_locked(struct file *fp); /* like fhold but expects file to locked */ struct file * ffind_hold(struct thread *, int fd); /* finds the struct file in thread, adds one reference and returns it unlocked */ struct file * ffind_lock(struct thread *, int fd); /* ffind_hold, but returns file locked */ I still have to smp-safe the fget cruft, I'll get to that asap.
2002-01-13 11:58:06 +00:00
FILEDESC_UNLOCK(fdp);
Remove the kernel FD_SETSIZE limit for select(). Make select()'s first argument 'int' not 'u_int'. Reviewed by: bde
1996-08-20 07:18:10 +00:00
/*
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
* Allocate just enough bits for the non-null fd_sets. Use the
* preallocated auto buffer if possible.
Remove the kernel FD_SETSIZE limit for select(). Make select()'s first argument 'int' not 'u_int'. Reviewed by: bde
1996-08-20 07:18:10 +00:00
*/
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
nfdbits = roundup(nd, NFDBITS);
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
ncpbytes = nfdbits / NBBY;
nbufbytes = 0;
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (fd_in != NULL)
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
nbufbytes += 2 * ncpbytes;
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (fd_ou != NULL)
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
nbufbytes += 2 * ncpbytes;
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (fd_ex != NULL)
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
nbufbytes += 2 * ncpbytes;
if (nbufbytes <= sizeof s_selbits)
selbits = &s_selbits[0];
else
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
selbits = malloc(nbufbytes, M_SELECT, M_WAITOK);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
/*
* Assign pointers into the bit buffers and fetch the input bits.
* Put the output buffers together so that they can be bzeroed
* together.
*/
sbp = selbits;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#define getbits(name, x) \
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
do { \
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (name == NULL) \
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
ibits[x] = NULL; \
else { \
ibits[x] = sbp + nbufbytes / 2 / sizeof *sbp; \
obits[x] = sbp; \
sbp += ncpbytes / sizeof *sbp; \
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
error = copyin(name, ibits[x], ncpbytes); \
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
if (error != 0) \
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
goto done_nosellock; \
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
} \
} while (0)
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
getbits(fd_in, 0);
getbits(fd_ou, 1);
getbits(fd_ex, 2);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#undef getbits
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
if (nbufbytes != 0)
bzero(selbits, nbufbytes / 2);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (tvp != NULL) {
atv = *tvp;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (itimerfix(&atv)) {
error = EINVAL;
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
goto done_nosellock;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
s/nanoruntime/nanouptime/g s/microruntime/microuptime/g Reviewed by: bde
1998-05-17 11:53:46 +00:00
getmicrouptime(&rtv);
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
timevaladd(&atv, &rtv);
For infinite timeouts, set both the tv_sec and tv_usec fields to zero in poll() and select(). Noticed by: Wesley Morgan <morganw@chemicals.tacorp.com>
2000-07-13 02:12:25 +00:00
} else {
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
atv.tv_sec = 0;
For infinite timeouts, set both the tv_sec and tv_usec fields to zero in poll() and select(). Noticed by: Wesley Morgan <morganw@chemicals.tacorp.com>
2000-07-13 02:12:25 +00:00
atv.tv_usec = 0;
}
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
timo = 0;
Move the TAILQ_INIT(&td->td_selq) before the retry: label. Otherwise in some circumstances when we get a select collision, we can end up with cases where we do not clear some sip->si_thread on the way out, leading to page faults in selwakeup(). This should solve the problem where postfix can crash the kernel during select collisions. Reviewed by: alfred
2002-08-23 22:43:28 +00:00
TAILQ_INIT(&td->td_selq);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_lock(&sellock);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
retry:
ncoll = nselcoll;
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_lock_spin(&sched_lock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_flags |= TDF_SELECT;
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_unlock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock(&sellock);
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
error = selscan(td, ibits, obits, nd);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_lock(&sellock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
if (error || td->td_retval[0])
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
goto done;
Fix a very obscure bug in select() and poll() where the timeout would never expire if poll() or select() was called before the system had been in multiuser for 1 second. This was caused by only checking to see if tv_sec was zero rather than checking both tv_sec and tv_usec.
2000-07-12 22:46:40 +00:00
if (atv.tv_sec || atv.tv_usec) {
s/nanoruntime/nanouptime/g s/microruntime/microuptime/g Reviewed by: bde
1998-05-17 11:53:46 +00:00
getmicrouptime(&rtv);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
if (timevalcmp(&rtv, &atv, >=))
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
goto done;
ttv = atv;
timevalsub(&ttv, &rtv);
timo = ttv.tv_sec > 24 * 60 * 60 ?
24 * 60 * 60 * hz : tvtohz(&ttv);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
/*
* An event of interest may occur while we do not hold
* sellock, so check TDF_SELECT and the number of
* collisions and rescan the file descriptors if
* necessary.
*/
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_lock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
if ((td->td_flags & TDF_SELECT) == 0 || nselcoll != ncoll) {
mtx_unlock_spin(&sched_lock);
goto retry;
}
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_unlock_spin(&sched_lock);
Add aio_waitcomplete(). Make aio work correctly for socket descriptors. Make gratuitous style(9) fixes (me, not the submitter) to make the aio code more readable. PR: kern/12053 Submitted by: Chris Sedore <cmsedore@maxwell.syr.edu>
2000-01-14 02:53:29 +00:00
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
if (timo > 0)
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
error = cv_timedwait_sig(&selwait, &sellock, timo);
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
else
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
error = cv_wait_sig(&selwait, &sellock);
Add aio_waitcomplete(). Make aio work correctly for socket descriptors. Make gratuitous style(9) fixes (me, not the submitter) to make the aio code more readable. PR: kern/12053 Submitted by: Chris Sedore <cmsedore@maxwell.syr.edu>
2000-01-14 02:53:29 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
if (error == 0)
goto retry;
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
done:
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
clear_selinfo_list(td);
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_lock_spin(&sched_lock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_flags &= ~TDF_SELECT;
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_unlock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock(&sellock);
done_nosellock:
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/* select is not restarted after signals... */
if (error == ERESTART)
error = EINTR;
if (error == EWOULDBLOCK)
error = 0;
#define putbits(name, x) \
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
if (name && (error2 = copyout(obits[x], name, ncpbytes))) \
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
error = error2;
if (error == 0) {
int error2;
Split out a number of mostly VFS and signal related syscalls into a kernel-internal kern_*() version and a wrapper that is called via the syscall vector table. For paths and structure pointers, the internal version either takes a uio_seg parameter or requires the caller to copyin() the data to kernel memory as appropiate. This will permit emulation layers to use these syscalls without having to copy out translated arguments to the stack gap. Discussed on: -arch Review/suggestions: bde, jhb, peter, marcel
2002-09-01 20:37:28 +00:00
putbits(fd_in, 0);
putbits(fd_ou, 1);
putbits(fd_ex, 2);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
#undef putbits
}
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
if (selbits != &s_selbits[0])
free(selbits, M_SELECT);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
mtx_unlock(&Giant);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (error);
}
A Major staticize sweep. Generates a couple of warnings that I'll deal with later. A number of unused vars removed. A number of unused procs removed or #ifdefed.
1995-12-14 08:32:45 +00:00
static int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
selscan(td, ibits, obits, nfd)
struct thread *td;
Remove the kernel FD_SETSIZE limit for select(). Make select()'s first argument 'int' not 'u_int'. Reviewed by: bde
1996-08-20 07:18:10 +00:00
fd_mask **ibits, **obits;
Move the "retval" (3rd) parameter from all syscall functions and put it in struct proc instead. This fixes a boatload of compiler warning, and removes a lot of cruft from the sources. I have not removed the /*ARGSUSED*/, they will require some looking at. libkvm, ps and other userland struct proc frobbing programs will need recompiled.
1997-11-06 19:29:57 +00:00
int nfd;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
Fix select(2) for the Alpha. (!!) It was never returning true for fd's in the range of 32-63, 96-127 etc. The first problem was the FD_*() macros were shifting a 32 bit integer "1" left by more than 32 bits. The same problem happened in selscan(). ffs() also takes an int argument and causes failure. For cases where int == long (ie: the usual case for x86, but not always as gcc can have long being a 64 bit quantity) ffs() could be used. Reported by: Marian Stagarescu <marian@bile.skycache.com> Reviewed by: dfr, gallatin (sys/types.h only) Approved by: jkh
2000-02-20 13:36:26 +00:00
int msk, i, fd;
fd_mask bits;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct file *fp;
int n = 0;
Fixed some style bugs in the poll() code. Removed dead code to "Avoid inadvertently sleeping forever". hzto() never returns 0.
1997-11-23 10:30:50 +00:00
/* Note: backend also returns POLLHUP/POLLERR if appropriate. */
static int flag[3] = { POLLRDNORM, POLLWRNORM, POLLRDBAND };
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
struct filedesc *fdp = td->td_proc->p_fd;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
FILEDESC_LOCK(fdp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
for (msk = 0; msk < 3; msk++) {
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
if (ibits[msk] == NULL)
continue;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
for (i = 0; i < nfd; i += NFDBITS) {
Remove the kernel FD_SETSIZE limit for select(). Make select()'s first argument 'int' not 'u_int'. Reviewed by: bde
1996-08-20 07:18:10 +00:00
bits = ibits[msk][i/NFDBITS];
Fix select(2) for the Alpha. (!!) It was never returning true for fd's in the range of 32-63, 96-127 etc. The first problem was the FD_*() macros were shifting a 32 bit integer "1" left by more than 32 bits. The same problem happened in selscan(). ffs() also takes an int argument and causes failure. For cases where int == long (ie: the usual case for x86, but not always as gcc can have long being a 64 bit quantity) ffs() could be used. Reported by: Marian Stagarescu <marian@bile.skycache.com> Reviewed by: dfr, gallatin (sys/types.h only) Approved by: jkh
2000-02-20 13:36:26 +00:00
/* ffs(int mask) not portable, fd_mask is long */
for (fd = i; bits && fd < nfd; fd++, bits >>= 1) {
if (!(bits & 1))
continue;
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
if ((fp = fget_locked(fdp, fd)) == NULL) {
FILEDESC_UNLOCK(fdp);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (EBADF);
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
}
Make similar changes to fo_stat() and fo_poll() as made earlier to fo_read() and fo_write(): explicitly use the cred argument to fo_poll() as "active_cred" using the passed file descriptor's f_cred reference to provide access to the file credential. Add an active_cred argument to fo_stat() so that implementers have access to the active credential as well as the file credential. Generally modify callers of fo_stat() to pass in td->td_ucred rather than fp->f_cred, which was redundantly provided via the fp argument. This set of modifications also permits threads to perform these operations on behalf of another thread without modifying their credential. Trickle this change down into fo_stat/poll() implementations: - badfo_poll(), badfo_stat(): modify/add arguments. - kqueue_poll(), kqueue_stat(): modify arguments. - pipe_poll(), pipe_stat(): modify/add arguments, pass active_cred to MAC checks rather than td->td_ucred. - soo_poll(), soo_stat(): modify/add arguments, pass fp->f_cred rather than cred to pru_sopoll() to maintain current semantics. - sopoll(): moidfy arguments. - vn_poll(), vn_statfile(): modify/add arguments, pass new arguments to vn_stat(). Pass active_cred to MAC and fp->f_cred to VOP_POLL() to maintian current semantics. - vn_close(): rename cred to file_cred to reflect reality while I'm here. - vn_stat(): Add active_cred and file_cred arguments to vn_stat() and consumers so that this distinction is maintained at the VFS as well as 'struct file' layer. Pass active_cred instead of td->td_ucred to MAC and to VOP_GETATTR() to maintain current semantics. - fifofs: modify the creation of a "filetemp" so that the file credential is properly initialized and can be used in the socket code if desired. Pass ap->a_td->td_ucred as the active credential to soo_poll(). If we teach the vnop interface about the distinction between file and active credentials, we would use the active credential here. Note that current inconsistent passing of active_cred vs. file_cred to VOP's is maintained. It's not clear why GETATTR would be authorized using active_cred while POLL would be authorized using file_cred at the file system level. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 12:52:03 +00:00
if (fo_poll(fp, flag[msk], td->td_ucred,
td)) {
Improved select(): - avoid malloc() if the number of fds is small. - pack the bits better so that `small' is quite large. - don't waste time generating zero bits for null fd_set pointers or scanning these bits. Possibly improved select(): - free malloc()ed storage before returning. This is simpler and I think huge select()s aren't worth optimizing since they are rare, relative gain would be small and there would be tiny costs for all selects(). Reviewed by: ache (first version by him too)
1997-02-20 11:51:52 +00:00
obits[msk][(fd)/NFDBITS] |=
Fix select(2) for the Alpha. (!!) It was never returning true for fd's in the range of 32-63, 96-127 etc. The first problem was the FD_*() macros were shifting a 32 bit integer "1" left by more than 32 bits. The same problem happened in selscan(). ffs() also takes an int argument and causes failure. For cases where int == long (ie: the usual case for x86, but not always as gcc can have long being a 64 bit quantity) ffs() could be used. Reported by: Marian Stagarescu <marian@bile.skycache.com> Reviewed by: dfr, gallatin (sys/types.h only) Approved by: jkh
2000-02-20 13:36:26 +00:00
((fd_mask)1 << ((fd) % NFDBITS));
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
n++;
}
}
}
}
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
FILEDESC_UNLOCK(fdp);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_retval[0] = n;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
return (0);
}
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
/*
* Poll system call.
*/
#ifndef _SYS_SYSPROTO_H_
struct poll_args {
struct pollfd *fds;
u_int nfds;
int timeout;
};
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
poll(td, uap)
struct thread *td;
Correctly declare variables as u_int rather than doing typecasts. Kill some register declarations while I'm here. Submitted by: bde (1)
2001-02-27 15:11:31 +00:00
struct poll_args *uap;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
{
caddr_t bits;
char smallbits[32 * sizeof(struct pollfd)];
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
struct timeval atv, rtv, ttv;
Make nselcol, the number of select collisions since boot, unsigned as negative collisions simply doesn't make sense. PR: (one small part of) 19720 Approved by: alfred
2002-06-12 02:08:18 +00:00
int error = 0, timo;
u_int ncoll, nfds;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
size_t ni;
SCARGS removal take II.
2002-12-14 01:56:26 +00:00
nfds = uap->nfds;
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
Add annotations to mtx_lock(&Giant) in kern_select() and poll() that we always grab Giant, even if we're actually only polling objects that don't require giant. Once socket locking is merged, there will be strong motivation to fix this.
2004-03-13 05:58:57 +00:00
/*
* XXX: poll() currently requires that we acquire Giant even if
* none of the file descriptors we poll requires Giant.
*/
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
mtx_lock(&Giant);
The code I picked up from NetBSD in '97 had a nasty bug. It limited the index of the pollfd array to the number of fd's currently open, not the maximum number of fd's. ie: if you had 0,1,2 open, you could not use pollfd slots higher than 20. The specs say we only have to support OPEN_MAX [64] entries but we allow way more than that.
2001-02-07 23:28:01 +00:00
/*
poll(2) array limits (take 2) - after some input from bde.
2001-02-09 08:10:22 +00:00
* This is kinda bogus. We have fd limits, but that is not
* really related to the size of the pollfd array. Make sure
* we let the process use at least FD_SETSIZE entries and at
* least enough for the current limits. We want to be reasonably
* safe, but not overly restrictive.
The code I picked up from NetBSD in '97 had a nasty bug. It limited the index of the pollfd array to the number of fd's currently open, not the maximum number of fd's. ie: if you had 0,1,2 open, you could not use pollfd slots higher than 20. The specs say we only have to support OPEN_MAX [64] entries but we allow way more than that.
2001-02-07 23:28:01 +00:00
*/
Locking for the per-process resource limits structure. - struct plimit includes a mutex to protect a reference count. The plimit structure is treated similarly to struct ucred in that is is always copy on write, so having a reference to a structure is sufficient to read from it without needing a further lock. - The proc lock protects the p_limit pointer and must be held while reading limits from a process to keep the limit structure from changing out from under you while reading from it. - Various global limits that are ints are not protected by a lock since int writes are atomic on all the archs we support and thus a lock wouldn't buy us anything. - All accesses to individual resource limits from a process are abstracted behind a simple lim_rlimit(), lim_max(), and lim_cur() API that return either an rlimit, or the current or max individual limit of the specified resource from a process. - dosetrlimit() was renamed to kern_setrlimit() to match existing style of other similar syscall helper functions. - The alpha OSF/1 compat layer no longer calls getrlimit() and setrlimit() (it didn't used the stackgap when it should have) but uses lim_rlimit() and kern_setrlimit() instead. - The svr4 compat no longer uses the stackgap for resource limits calls, but uses lim_rlimit() and kern_setrlimit() instead. - The ibcs2 compat no longer uses the stackgap for resource limits. It also no longer uses the stackgap for accessing sysctl's for the ibcs2_sysconf() syscall but uses kernel_sysctl() instead. As a result, ibcs2_sysconf() no longer needs Giant. - The p_rlimit macro no longer exists. Submitted by: mtm (mostly, I only did a few cleanups and catchups) Tested on: i386 Compiled on: alpha, amd64
2004-02-04 21:52:57 +00:00
PROC_LOCK(td->td_proc);
if ((nfds > lim_cur(td->td_proc, RLIMIT_NOFILE)) &&
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
(nfds > FD_SETSIZE)) {
Locking for the per-process resource limits structure. - struct plimit includes a mutex to protect a reference count. The plimit structure is treated similarly to struct ucred in that is is always copy on write, so having a reference to a structure is sufficient to read from it without needing a further lock. - The proc lock protects the p_limit pointer and must be held while reading limits from a process to keep the limit structure from changing out from under you while reading from it. - Various global limits that are ints are not protected by a lock since int writes are atomic on all the archs we support and thus a lock wouldn't buy us anything. - All accesses to individual resource limits from a process are abstracted behind a simple lim_rlimit(), lim_max(), and lim_cur() API that return either an rlimit, or the current or max individual limit of the specified resource from a process. - dosetrlimit() was renamed to kern_setrlimit() to match existing style of other similar syscall helper functions. - The alpha OSF/1 compat layer no longer calls getrlimit() and setrlimit() (it didn't used the stackgap when it should have) but uses lim_rlimit() and kern_setrlimit() instead. - The svr4 compat no longer uses the stackgap for resource limits calls, but uses lim_rlimit() and kern_setrlimit() instead. - The ibcs2 compat no longer uses the stackgap for resource limits. It also no longer uses the stackgap for accessing sysctl's for the ibcs2_sysconf() syscall but uses kernel_sysctl() instead. As a result, ibcs2_sysconf() no longer needs Giant. - The p_rlimit macro no longer exists. Submitted by: mtm (mostly, I only did a few cleanups and catchups) Tested on: i386 Compiled on: alpha, amd64
2004-02-04 21:52:57 +00:00
PROC_UNLOCK(td->td_proc);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
error = EINVAL;
goto done2;
}
Locking for the per-process resource limits structure. - struct plimit includes a mutex to protect a reference count. The plimit structure is treated similarly to struct ucred in that is is always copy on write, so having a reference to a structure is sufficient to read from it without needing a further lock. - The proc lock protects the p_limit pointer and must be held while reading limits from a process to keep the limit structure from changing out from under you while reading from it. - Various global limits that are ints are not protected by a lock since int writes are atomic on all the archs we support and thus a lock wouldn't buy us anything. - All accesses to individual resource limits from a process are abstracted behind a simple lim_rlimit(), lim_max(), and lim_cur() API that return either an rlimit, or the current or max individual limit of the specified resource from a process. - dosetrlimit() was renamed to kern_setrlimit() to match existing style of other similar syscall helper functions. - The alpha OSF/1 compat layer no longer calls getrlimit() and setrlimit() (it didn't used the stackgap when it should have) but uses lim_rlimit() and kern_setrlimit() instead. - The svr4 compat no longer uses the stackgap for resource limits calls, but uses lim_rlimit() and kern_setrlimit() instead. - The ibcs2 compat no longer uses the stackgap for resource limits. It also no longer uses the stackgap for accessing sysctl's for the ibcs2_sysconf() syscall but uses kernel_sysctl() instead. As a result, ibcs2_sysconf() no longer needs Giant. - The p_rlimit macro no longer exists. Submitted by: mtm (mostly, I only did a few cleanups and catchups) Tested on: i386 Compiled on: alpha, amd64
2004-02-04 21:52:57 +00:00
PROC_UNLOCK(td->td_proc);
The code I picked up from NetBSD in '97 had a nasty bug. It limited the index of the pollfd array to the number of fd's currently open, not the maximum number of fd's. ie: if you had 0,1,2 open, you could not use pollfd slots higher than 20. The specs say we only have to support OPEN_MAX [64] entries but we allow way more than that.
2001-02-07 23:28:01 +00:00
ni = nfds * sizeof(struct pollfd);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
if (ni > sizeof(smallbits))
Back out M_* changes, per decision of the TRB. Approved by: trb
2003-02-19 05:47:46 +00:00
bits = malloc(ni, M_TEMP, M_WAITOK);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
else
bits = smallbits;
SCARGS removal take II.
2002-12-14 01:56:26 +00:00
error = copyin(uap->fds, bits, ni);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
if (error)
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
goto done_nosellock;
SCARGS removal take II.
2002-12-14 01:56:26 +00:00
if (uap->timeout != INFTIM) {
atv.tv_sec = uap->timeout / 1000;
atv.tv_usec = (uap->timeout % 1000) * 1000;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
if (itimerfix(&atv)) {
error = EINVAL;
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
goto done_nosellock;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
}
s/nanoruntime/nanouptime/g s/microruntime/microuptime/g Reviewed by: bde
1998-05-17 11:53:46 +00:00
getmicrouptime(&rtv);
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
timevaladd(&atv, &rtv);
For infinite timeouts, set both the tv_sec and tv_usec fields to zero in poll() and select(). Noticed by: Wesley Morgan <morganw@chemicals.tacorp.com>
2000-07-13 02:12:25 +00:00
} else {
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
atv.tv_sec = 0;
For infinite timeouts, set both the tv_sec and tv_usec fields to zero in poll() and select(). Noticed by: Wesley Morgan <morganw@chemicals.tacorp.com>
2000-07-13 02:12:25 +00:00
atv.tv_usec = 0;
}
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
timo = 0;
Move the TAILQ_INIT(&td->td_selq) before the retry: label. Otherwise in some circumstances when we get a select collision, we can end up with cases where we do not clear some sip->si_thread on the way out, leading to page faults in selwakeup(). This should solve the problem where postfix can crash the kernel during select collisions. Reviewed by: alfred
2002-08-23 22:43:28 +00:00
TAILQ_INIT(&td->td_selq);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_lock(&sellock);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
retry:
ncoll = nselcoll;
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_lock_spin(&sched_lock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_flags |= TDF_SELECT;
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_unlock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock(&sellock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
error = pollscan(td, (struct pollfd *)bits, nfds);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_lock(&sellock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
if (error || td->td_retval[0])
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
goto done;
Fix a very obscure bug in select() and poll() where the timeout would never expire if poll() or select() was called before the system had been in multiuser for 1 second. This was caused by only checking to see if tv_sec was zero rather than checking both tv_sec and tv_usec.
2000-07-12 22:46:40 +00:00
if (atv.tv_sec || atv.tv_usec) {
s/nanoruntime/nanouptime/g s/microruntime/microuptime/g Reviewed by: bde
1998-05-17 11:53:46 +00:00
getmicrouptime(&rtv);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
if (timevalcmp(&rtv, &atv, >=))
Time changes mark 2: * Figure out UTC relative to boottime. Four new functions provide time relative to boottime. * move "runtime" into struct proc. This helps fix the calcru() problem in SMP. * kill mono_time. * add timespec{add|sub|cmp} macros to time.h. (XXX: These may change!) * nanosleep, select & poll takes long sleeps one day at a time Reviewed by: bde Tested by: ache and others
1998-04-04 13:26:20 +00:00
goto done;
ttv = atv;
timevalsub(&ttv, &rtv);
timo = ttv.tv_sec > 24 * 60 * 60 ?
24 * 60 * 60 * hz : tvtohz(&ttv);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
/*
* An event of interest may occur while we do not hold
* sellock, so check TDF_SELECT and the number of collisions
* and rescan the file descriptors if necessary.
*/
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_lock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
if ((td->td_flags & TDF_SELECT) == 0 || nselcoll != ncoll) {
mtx_unlock_spin(&sched_lock);
goto retry;
}
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_unlock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
if (timo > 0)
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
error = cv_timedwait_sig(&selwait, &sellock, timo);
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
else
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
error = cv_wait_sig(&selwait, &sellock);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
if (error == 0)
goto retry;
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
done:
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
clear_selinfo_list(td);
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_lock_spin(&sched_lock);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_flags &= ~TDF_SELECT;
The P_SELECT flag was moved from p->p_flag to td->td_flags, but p_flag was locked by the proc lock and td_flags is locked by the sched_lock. The places that read, set, and cleared TDF_SELECT weren't updated, so they read and modified td_flags w/o holding the sched_lock, meaning that they could corrupt the per-thread flags field. As an immediate band-aid, grab sched_lock while reading and manipulating td_flags in relation to TDF_SELECT. This will probably be cleaned up some later on.
2001-09-21 22:06:22 +00:00
mtx_unlock_spin(&sched_lock);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock(&sellock);
done_nosellock:
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
/* poll is not restarted after signals... */
if (error == ERESTART)
error = EINTR;
if (error == EWOULDBLOCK)
error = 0;
if (error == 0) {
SCARGS removal take II.
2002-12-14 01:56:26 +00:00
error = copyout(bits, uap->fds, ni);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
if (error)
goto out;
}
out:
if (ni > sizeof(smallbits))
free(bits, M_TEMP);
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
done2:
mtx_unlock(&Giant);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
return (error);
}
static int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
pollscan(td, fds, nfd)
struct thread *td;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
struct pollfd *fds;
Correctly declare variables as u_int rather than doing typecasts. Kill some register declarations while I'm here. Submitted by: bde (1)
2001-02-27 15:11:31 +00:00
u_int nfd;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
register struct filedesc *fdp = td->td_proc->p_fd;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
int i;
struct file *fp;
int n = 0;
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
FILEDESC_LOCK(fdp);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
for (i = 0; i < nfd; i++, fds++) {
poll(2) sets POLLNVAL for descriptors passed in that are less than 0. This makes it difficult to do efficient manipulation of the struct pollfd since you can't leave a slot empty. PR: 8599 Submitted-by: Marc Slemko <marcs@znep.com>
1998-12-10 01:53:26 +00:00
if (fds->fd >= fdp->fd_nfiles) {
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
fds->revents = POLLNVAL;
n++;
poll(2) sets POLLNVAL for descriptors passed in that are less than 0. This makes it difficult to do efficient manipulation of the struct pollfd since you can't leave a slot empty. PR: 8599 Submitted-by: Marc Slemko <marcs@znep.com>
1998-12-10 01:53:26 +00:00
} else if (fds->fd < 0) {
fds->revents = 0;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
} else {
fp = fdp->fd_ofiles[fds->fd];
This patchset fixes a large number of file descriptor race conditions. Pre-rfork code assumed inherent locking of a process's file descriptor array. However, with the advent of rfork() the file descriptor table could be shared between processes. This patch closes over a dozen serious race conditions related to one thread manipulating the table (e.g. closing or dup()ing a descriptor) while another is blocked in an open(), close(), fcntl(), read(), write(), etc... PR: kern/11629 Discussed with: Alexander Viro <viro@math.psu.edu>
2000-11-18 21:01:04 +00:00
if (fp == NULL) {
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
fds->revents = POLLNVAL;
n++;
} else {
Fixed some style bugs in the poll() code. Removed dead code to "Avoid inadvertently sleeping forever". hzto() never returns 0.
1997-11-23 10:30:50 +00:00
/*
* Note: backend also returns POLLHUP and
* POLLERR if appropriate.
*/
This is what was "fdfix2.patch," a fix for fd sharing. It's pretty far-reaching in fd-land, so you'll want to consult the code for changes. The biggest change is that now, you don't use fp->f_ops->fo_foo(fp, bar) but instead fo_foo(fp, bar), which increments and decrements the fp refcount upon entry and exit. Two new calls, fhold() and fdrop(), are provided. Each does what it seems like it should, and if fdrop() brings the refcount to zero, the fd is freed as well. Thanks to peter ("to hell with it, it looks ok to me.") for his review. Thanks to msmith for keeping me from putting locks everywhere :) Reviewed by: peter
1999-09-19 17:00:25 +00:00
fds->revents = fo_poll(fp, fds->events,
Make similar changes to fo_stat() and fo_poll() as made earlier to fo_read() and fo_write(): explicitly use the cred argument to fo_poll() as "active_cred" using the passed file descriptor's f_cred reference to provide access to the file credential. Add an active_cred argument to fo_stat() so that implementers have access to the active credential as well as the file credential. Generally modify callers of fo_stat() to pass in td->td_ucred rather than fp->f_cred, which was redundantly provided via the fp argument. This set of modifications also permits threads to perform these operations on behalf of another thread without modifying their credential. Trickle this change down into fo_stat/poll() implementations: - badfo_poll(), badfo_stat(): modify/add arguments. - kqueue_poll(), kqueue_stat(): modify arguments. - pipe_poll(), pipe_stat(): modify/add arguments, pass active_cred to MAC checks rather than td->td_ucred. - soo_poll(), soo_stat(): modify/add arguments, pass fp->f_cred rather than cred to pru_sopoll() to maintain current semantics. - sopoll(): moidfy arguments. - vn_poll(), vn_statfile(): modify/add arguments, pass new arguments to vn_stat(). Pass active_cred to MAC and fp->f_cred to VOP_POLL() to maintian current semantics. - vn_close(): rename cred to file_cred to reflect reality while I'm here. - vn_stat(): Add active_cred and file_cred arguments to vn_stat() and consumers so that this distinction is maintained at the VFS as well as 'struct file' layer. Pass active_cred instead of td->td_ucred to MAC and to VOP_GETATTR() to maintain current semantics. - fifofs: modify the creation of a "filetemp" so that the file credential is properly initialized and can be used in the socket code if desired. Pass ap->a_td->td_ucred as the active credential to soo_poll(). If we teach the vnop interface about the distinction between file and active credentials, we would use the active credential here. Note that current inconsistent passing of active_cred vs. file_cred to VOP's is maintained. It's not clear why GETATTR would be authorized using active_cred while POLL would be authorized using file_cred at the file system level. Obtained from: TrustedBSD Project Sponsored by: DARPA, NAI Labs
2002-08-16 12:52:03 +00:00
td->td_ucred, td);
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
if (fds->revents != 0)
n++;
}
}
}
Attempt to fixup select(2) and poll(2), this should fix some races with other threads as well as speed up the interfaces. To fix the race and accomplish the speedup, remove selholddrop and pollholddrop. The entire concept is somewhat bogus because holding the individual struct file pointers offers us no guarantees that another thread context won't close it on us thereby removing our access to our own reference. Selholddrop and pollholddrop also would do multiple locks and unlocks of mutexes _per-file_ in the fd arrays to be scanned, this needed to be sped up. Instead of using selholddrop and pollholddrop, simply hold the filedesc lock over the selscan and pollscan functions. This should protect us against close(2)'s on the files as reduce the multiple lock/unlock pairs per fd into a single lock over the filedesc.
2002-01-29 22:54:19 +00:00
FILEDESC_UNLOCK(fdp);
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
td->td_retval[0] = n;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
return (0);
}
/*
* OpenBSD poll system call.
* XXX this isn't quite a true representation.. OpenBSD uses select ops.
*/
#ifndef _SYS_SYSPROTO_H_
struct openbsd_poll_args {
struct pollfd *fds;
u_int nfds;
int timeout;
};
#endif
Giant Pushdown: read() pread() readv() write () pwrite() writev() ioctl() select () poll() openbsd_poll()
2001-09-01 19:34:23 +00:00
/*
* MPSAFE
*/
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
int
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
openbsd_poll(td, uap)
register struct thread *td;
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
register struct openbsd_poll_args *uap;
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
return (poll(td, (struct poll_args *)uap));
Implement poll(2). This is mostly taken from the NetBSD implementation (from some time ago) but with a few tweaks along the way. Obtained from: NetBSD
1997-09-14 02:30:32 +00:00
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
/*
* Remove the references to the thread from all of the objects
* we were polling.
*
* This code assumes that the underlying owner of the selinfo
* structure will hold sellock before it changes it, and that
* it will unlink itself from our list if it goes away.
*/
void
clear_selinfo_list(td)
struct thread *td;
{
struct selinfo *si;
mtx_assert(&sellock, MA_OWNED);
TAILQ_FOREACH(si, &td->td_selq, si_thrlist)
si->si_thread = NULL;
TAILQ_INIT(&td->td_selq);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Record a select request.
*/
void
selrecord(selector, sip)
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct thread *selector;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
struct selinfo *sip;
{
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_lock(&sellock);
/*
Attempt to clarify comment in selrecord.
2002-07-24 00:29:22 +00:00
* If the selinfo's thread pointer is NULL then take ownership of it.
*
* If the thread pointer is not NULL and it points to another
* thread, then we have a collision.
*
* If the thread pointer is not NULL and points back to us then leave
* it alone as we've already added pointed it at us and added it to
* our list.
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
*/
if (sip->si_thread == NULL) {
sip->si_thread = selector;
TAILQ_INSERT_TAIL(&selector->td_selq, sip, si_thrlist);
} else if (sip->si_thread != selector) {
sip->si_flags |= SI_COLL;
- Catch up to proc flag changes. - Add proc locking for selwakeup() and selrecord().
2001-01-24 11:12:37 +00:00
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock(&sellock);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Implement selwakeuppri() which allows raising the priority of a thread being waken up. The thread waken up can run at a priority as high as after tsleep(). - Replace selwakeup()s with selwakeuppri()s and pass appropriate priorities. - Add cv_broadcastpri() which raises the priority of the broadcast threads. Used by selwakeuppri() if collision occurs. Not objected in: -arch, -current
2003-11-09 09:17:26 +00:00
/* Wake up a selecting thread. */
void
selwakeup(sip)
struct selinfo *sip;
{
doselwakeup(sip, -1);
}
/* Wake up a selecting thread, and set its priority. */
void
selwakeuppri(sip, pri)
struct selinfo *sip;
int pri;
{
doselwakeup(sip, pri);
}
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
/*
* Do a wakeup when a selectable event occurs.
*/
- Implement selwakeuppri() which allows raising the priority of a thread being waken up. The thread waken up can run at a priority as high as after tsleep(). - Replace selwakeup()s with selwakeuppri()s and pass appropriate priorities. - Add cv_broadcastpri() which raises the priority of the broadcast threads. Used by selwakeuppri() if collision occurs. Not objected in: -arch, -current
2003-11-09 09:17:26 +00:00
static void
doselwakeup(sip, pri)
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
struct selinfo *sip;
- Implement selwakeuppri() which allows raising the priority of a thread being waken up. The thread waken up can run at a priority as high as after tsleep(). - Replace selwakeup()s with selwakeuppri()s and pass appropriate priorities. - Add cv_broadcastpri() which raises the priority of the broadcast threads. Used by selwakeuppri() if collision occurs. Not objected in: -arch, -current
2003-11-09 09:17:26 +00:00
int pri;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
{
KSE Milestone 2 Note ALL MODULES MUST BE RECOMPILED make the kernel aware that there are smaller units of scheduling than the process. (but only allow one thread per process at this time). This is functionally equivalent to teh previousl -current except that there is a thread associated with each process. Sorry john! (your next MFC will be a doosie!) Reviewed by: peter@freebsd.org, dillon@freebsd.org X-MFC after: ha ha ha ha
2001-09-12 08:38:13 +00:00
struct thread *td;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_lock(&sellock);
td = sip->si_thread;
if ((sip->si_flags & SI_COLL) != 0) {
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
nselcoll++;
sip->si_flags &= ~SI_COLL;
- Implement selwakeuppri() which allows raising the priority of a thread being waken up. The thread waken up can run at a priority as high as after tsleep(). - Replace selwakeup()s with selwakeuppri()s and pass appropriate priorities. - Add cv_broadcastpri() which raises the priority of the broadcast threads. Used by selwakeuppri() if collision occurs. Not objected in: -arch, -current
2003-11-09 09:17:26 +00:00
cv_broadcastpri(&selwait, pri);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
if (td == NULL) {
mtx_unlock(&sellock);
return;
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
TAILQ_REMOVE(&td->td_selq, sip, si_thrlist);
sip->si_thread = NULL;
mtx_lock_spin(&sched_lock);
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
td->td_flags &= ~TDF_SELECT;
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock_spin(&sched_lock);
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
sleepq_remove(td, &selwait);
Fixes to make select/poll mpsafe. Problem: selwakeup required calling pfind which would cause lock order reversals with the allproc_lock and the per-process filedesc lock. Solution: Instead of recording the pid of the select()'ing process into the selinfo structure, actually record a pointer to the thread. To avoid dereferencing a bad address all the selinfo structures that are in use by a thread are kept in a list hung off the thread (protected by sellock). When a selwakeup occurs the selinfo is removed from that threads list, it is also removed on the way out of select or poll where the thread will traverse its list removing all the selinfos from its own list. Problem: Previously the PROC_LOCK was used to provide the mutual exclusion needed to ensure proper locking, this couldn't work because there was a single condvar used for select and poll and condvars can only be used with a single mutex. Solution: Introduce a global mutex 'sellock' which is used to provide mutual exclusion when recording events to wait on as well as performing notification when an event occurs. Interesting note: schedlock is required to manipulate the per-thread TDF_SELECT flag, however if given its own field it would not need schedlock, also because TDF_SELECT is only manipulated under sellock one doesn't actually use schedlock for syncronization, only to protect against corruption. Proc locks are no longer used in select/poll. Portions contributed by: davidc
2002-03-14 01:32:30 +00:00
mtx_unlock(&sellock);
BSD 4.4 Lite Kernel Sources
1994-05-24 10:09:53 +00:00
}
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
Remove __P.
2002-03-19 21:25:46 +00:00
static void selectinit(void *);
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
SYSINIT(select, SI_SUB_LOCK, SI_ORDER_FIRST, selectinit, NULL)
/* ARGSUSED*/
static void
selectinit(dummy)
void *dummy;
{
cv_init(&selwait, "select");
Change callers of mtx_init() to pass in an appropriate lock type name. In most cases NULL is passed, but in some cases such as network driver locks (which use the MTX_NETWORK_LOCK macro) and UMA zone locks, a name is used. Tested on: i386, alpha, sparc64
2002-04-04 21:03:38 +00:00
mtx_init(&sellock, "sellck", NULL, MTX_DEF);
- Convert msleep(9) in select(2) and poll(2) to cv_*wait*(9). - Since polling should not involve sleeping, keep holding a process lock upon scanning file descriptors. - Hold a reference to every file descriptor prior to entering polling loop in order to avoid lock order reversal between lockmgr and p_mtx upon calling fdrop() in fo_poll(). (NOTE: this work has not been done for netncp and netsmb yet because a socket itself has no reference counts.) Reviewed by: jhb
2001-05-14 05:26:48 +00:00
}
Reference in New Issue Copy Permalink