]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoThis commit was generated by cvs2svn to compensate for changes in r125601,
Ruslan Ermilov [Sun, 8 Feb 2004 21:32:21 +0000 (21:32 +0000)]
This commit was generated by cvs2svn to compensate for changes in r125601,
which included commits to RCS files with non-trunk default branches.

20 years agoVendor import of bwk's 7-Feb-2004 release.
Ruslan Ermilov [Sun, 8 Feb 2004 21:32:21 +0000 (21:32 +0000)]
Vendor import of bwk's 7-Feb-2004 release.

20 years agoRewrite fabs.S to use pure SSE2 operations. I got the clues how to do
Peter Wemm [Sun, 8 Feb 2004 21:21:45 +0000 (21:21 +0000)]
Rewrite fabs.S to use pure SSE2 operations.  I got the clues how to do
this from looking at code generated by gcc.

20 years agoMake NdisMMapIoSpace() guard against NULL/uninitialized resource pointers too.
Bill Paul [Sun, 8 Feb 2004 20:39:35 +0000 (20:39 +0000)]
Make NdisMMapIoSpace() guard against NULL/uninitialized resource pointers too.

20 years agoMake NdisMMapIoSpace() handle the case where a device has both mem
Bill Paul [Sun, 8 Feb 2004 20:32:41 +0000 (20:32 +0000)]
Make NdisMMapIoSpace() handle the case where a device has both mem
and altmem ranges mapped.

20 years agoRemove condition variables and status associated with target mode
Matt Jacob [Sun, 8 Feb 2004 19:17:56 +0000 (19:17 +0000)]
Remove condition variables and status associated with target mode
enabling. Instead, go to an interrupt/polled model.

Fix get_lun_statep so we don't panic if there are no wildcard luns enabled.

MFC after: 6 days

20 years agoRemove condition variables and status associated with target mode
Matt Jacob [Sun, 8 Feb 2004 19:16:01 +0000 (19:16 +0000)]
Remove condition variables and status associated with target mode
enabling. Instead, go to an interrupt/polled model.

MFC after: 6 days

20 years ago- obey ip6po_minmtu.
Hajimu UMEMOTO [Sun, 8 Feb 2004 18:22:27 +0000 (18:22 +0000)]
- obey ip6po_minmtu.
- notify a proper path MTU to applications.

Obtained from: KAME

20 years agoAllow amr(4) to get a dynamic major number instead of a static one.
Scott Long [Sun, 8 Feb 2004 16:07:22 +0000 (16:07 +0000)]
Allow amr(4) to get a dynamic major number instead of a static one.

Submitted by: Andre Guibert de Bruet

20 years agoDon't free ressources that haven't been allocated. This should fix
Lukas Ertl [Sun, 8 Feb 2004 15:27:58 +0000 (15:27 +0000)]
Don't free ressources that haven't been allocated.  This should fix
the "disappearing subdisks" problem when new subdisks can't be created
due to some errors.

This is in fact an ugly hack, but a more elegant solution would probably
require a redesign of vinum in several places.

Approved by: joerg (mentor)

20 years agoUpdate the documentation for setpgrp(2) to reflect the access control
David Schultz [Sun, 8 Feb 2004 12:38:30 +0000 (12:38 +0000)]
Update the documentation for setpgrp(2) to reflect the access control
checks that the code actually performs.  Judging from the 4.2BSD
release notes, the docs have only been out of date for 20 years.

PR: 29844

20 years agoPolish the work/state engine in preparation for HW-crypto support.
Poul-Henning Kamp [Sun, 8 Feb 2004 10:19:18 +0000 (10:19 +0000)]
Polish the work/state engine in preparation for HW-crypto support.

20 years agoAdd a missing error case return.
Poul-Henning Kamp [Sun, 8 Feb 2004 09:39:02 +0000 (09:39 +0000)]
Add a missing error case return.

Problem reported by: Flemming Jacobsen <fj@batmule.dk>

20 years agoDocument the SF_NODISKIO flag, and fix a small typo.
Mike Silbersack [Sun, 8 Feb 2004 07:38:35 +0000 (07:38 +0000)]
Document the SF_NODISKIO flag, and fix a small typo.

20 years agoAdd the SF_NODISKIO flag to sendfile. This flag causes sendfile to be
Mike Silbersack [Sun, 8 Feb 2004 07:35:48 +0000 (07:35 +0000)]
Add the SF_NODISKIO flag to sendfile.  This flag causes sendfile to be
mindful of blocking on disk I/O and instead return EBUSY when such
blocking would occur.

Results from the DeBox project indicate that blocking on disk I/O
can slow the performance of a kqueue/poll based webserver.  Using
a flag such as SF_NODISKIO and throwing connections that would block
to helper processes/threads helped increase performance.

Currently, only the Flash webserver uses this flag, although it could
probably be applied to thttpd with relative ease.

Idea by: Yaoping Ruan & Vivek Pai

20 years agoMove m_uiotombuf from uio.h to mbuf.h, where I should have put it in
Mike Silbersack [Sun, 8 Feb 2004 03:19:08 +0000 (03:19 +0000)]
Move m_uiotombuf from uio.h to mbuf.h, where I should have put it in
the first place.

Suggested by: bde

20 years agoI forgot to add the NO_MODULES override for NOTES
Peter Wemm [Sun, 8 Feb 2004 00:39:38 +0000 (00:39 +0000)]
I forgot to add the NO_MODULES override for NOTES

20 years agoDon't hold NDIS_LOCK() around call to ndis_getstate_80211() since it
Bill Paul [Sat, 7 Feb 2004 23:52:58 +0000 (23:52 +0000)]
Don't hold NDIS_LOCK() around call to ndis_getstate_80211() since it
may block on ndis_get_info().

20 years agoArgh. kthread_suspend() when in P_KTHREAD context, tsleep() when not,
Bill Paul [Sat, 7 Feb 2004 23:47:10 +0000 (23:47 +0000)]
Argh. kthread_suspend() when in P_KTHREAD context, tsleep() when not,
not the other way around.

20 years agoArgh, unbreak "make depend" for AMD64.
Ruslan Ermilov [Sat, 7 Feb 2004 23:30:45 +0000 (23:30 +0000)]
Argh, unbreak "make depend" for AMD64.

Reported by: kris

20 years agoFix typo in comment, s/reebots/reboots/
Johan Karlsson [Sat, 7 Feb 2004 23:13:28 +0000 (23:13 +0000)]
Fix typo in comment, s/reebots/reboots/

PR: 62481
Submitted by: Ulrich Spoerlein <q at uni.de>

20 years agoWe don't need to hold Giant to create the worker kthread.
Poul-Henning Kamp [Sat, 7 Feb 2004 23:01:17 +0000 (23:01 +0000)]
We don't need to hold Giant to create the worker kthread.

20 years agoFix the last and most important bit of the test case to test the same
Poul-Henning Kamp [Sat, 7 Feb 2004 22:58:39 +0000 (22:58 +0000)]
Fix the last and most important bit of the test case to test the same
binary as the rest of it.

Add MD5 check that the md(4) device gets set up correctly.

20 years agoCorrect an intance of mtx_pool_lock() that should have been mtx_pool_unlock().
Bill Paul [Sat, 7 Feb 2004 22:19:20 +0000 (22:19 +0000)]
Correct an intance of mtx_pool_lock() that should have been mtx_pool_unlock().

20 years agoRemove the use of AACQ_COMPLETE here since there is no longer a completion
Scott Long [Sat, 7 Feb 2004 20:54:29 +0000 (20:54 +0000)]
Remove the use of AACQ_COMPLETE here since there is no longer a completion
queue.

20 years agoI guess nobody has needed to use the SVR4olator to create device
Poul-Henning Kamp [Sat, 7 Feb 2004 18:54:34 +0000 (18:54 +0000)]
I guess nobody has needed to use the SVR4olator to create device
nodes, or if they did, they're now locked away on the Kurt Gdel
memorial home for the numerically confused:

Don't cast a kernel pointer (from makedev(9)) to an integer (maj+minor combo).

20 years agoDo some small cleanups to comments and remove AACQ_COMPLETE definitions since
Scott Long [Sat, 7 Feb 2004 17:40:38 +0000 (17:40 +0000)]
Do some small cleanups to comments and remove AACQ_COMPLETE definitions since
the completion queue is long-gone.

20 years agoConvert to plain bsd.prog.mk makefile.
Ruslan Ermilov [Sat, 7 Feb 2004 16:58:23 +0000 (16:58 +0000)]
Convert to plain bsd.prog.mk makefile.

Reviewed by: scottl

20 years agoUpdate for the 2003/07/31 import.
Ruslan Ermilov [Sat, 7 Feb 2004 16:46:05 +0000 (16:46 +0000)]
Update for the 2003/07/31 import.

20 years agottys is gone, kill it here too
Warner Losh [Sat, 7 Feb 2004 15:46:20 +0000 (15:46 +0000)]
ttys is gone, kill it here too

20 years agoNULL looks better than (char *)0 unless we're passing
Yaroslav Tykhiy [Sat, 7 Feb 2004 14:59:11 +0000 (14:59 +0000)]
NULL looks better than (char *)0 unless we're passing
an unprototyped argument to a function.

20 years agoDeny attempts to rename a file from guest users if the policy
Yaroslav Tykhiy [Sat, 7 Feb 2004 14:54:30 +0000 (14:54 +0000)]
Deny attempts to rename a file from guest users if the policy
says they may not modify existing files through FTP.

Renaming a file is effectively a way to modify it.
For instance, if a malicious party is unable to delete or overwrite
a sensitive file, they can nevertheless rename it to a hidden name
and then upload a troyan horse under the guise of the old file name.

20 years agoperror_reply() should not be used where errno isn't meaningful.
Yaroslav Tykhiy [Sat, 7 Feb 2004 14:38:04 +0000 (14:38 +0000)]
perror_reply() should not be used where errno isn't meaningful.

20 years agoMFi386: Removed -elf and duplicate -ffreestanding.
Ruslan Ermilov [Sat, 7 Feb 2004 14:35:33 +0000 (14:35 +0000)]
MFi386: Removed -elf and duplicate -ffreestanding.

20 years ago- Removed -elf which is not a valid gcc(1) option anymore.
Ruslan Ermilov [Sat, 7 Feb 2004 14:32:06 +0000 (14:32 +0000)]
- Removed -elf which is not a valid gcc(1) option anymore.
- Removed -ffreestanding; it's enforced by ../Makefile.inc.

20 years agoWork around a bug in some clients by never returning raw directory
Yaroslav Tykhiy [Sat, 7 Feb 2004 14:11:38 +0000 (14:11 +0000)]
Work around a bug in some clients by never returning raw directory
contents in reply to a RETR command.  Such clients consider RETR
as a way to tell a file from a directory.  Mozilla is an example.

PR: bin/62232
Submitted by: Bob Finch <bob+freebsd <at> nas <dot> com>
MFC after: 1 week

20 years agoGenerate .depend file.
Ruslan Ermilov [Sat, 7 Feb 2004 13:38:47 +0000 (13:38 +0000)]
Generate .depend file.

20 years agoMFi386: Cleanups to sys/boot makefiles.
Yoshihiro Takahashi [Sat, 7 Feb 2004 12:19:44 +0000 (12:19 +0000)]
MFi386: Cleanups to sys/boot makefiles.

20 years agoUnbreak world.
Ruslan Ermilov [Sat, 7 Feb 2004 11:13:47 +0000 (11:13 +0000)]
Unbreak world.

20 years agoStyle: use the defined() expression explicitly.
Ruslan Ermilov [Sat, 7 Feb 2004 11:05:10 +0000 (11:05 +0000)]
Style: use the defined() expression explicitly.

20 years agoRemove the hack of lowering AAC_MAX_FIB now that the root cause of the
Scott Long [Sat, 7 Feb 2004 10:30:59 +0000 (10:30 +0000)]
Remove the hack of lowering AAC_MAX_FIB now that the root cause of the
problem was found.

20 years agoIf a command has to be deferred because there are no more resources for it
Scott Long [Sat, 7 Feb 2004 10:30:22 +0000 (10:30 +0000)]
If a command has to be deferred because there are no more resources for it
on the card, unmap it first.  This allows it to be picked up properly when
the queue gets kicked again.  This was the root problem for the lost command
(i.e. stuck in getblk/vinvalb) problem.  While here, panic if commands don't
map correctly instead of just silently ignoring the problem and dropping
command.  Also slow down the dynamic allocation of new commands.

It should be safe to go back into the aac waters.  Thanks to everyone who
suffered through this and provided good feedback.

20 years agoswp_pager_async_iodone() no longer requires Giant. Modify bufdone()
Alan Cox [Sat, 7 Feb 2004 08:54:50 +0000 (08:54 +0000)]
swp_pager_async_iodone() no longer requires Giant.  Modify bufdone()
and swapgeom_done() to perform swp_pager_async_iodone() without Giant.

Reviewed by: tegge

20 years agoUse the default threading library if requested.
Ruslan Ermilov [Sat, 7 Feb 2004 08:23:48 +0000 (08:23 +0000)]
Use the default threading library if requested.

Reviewed by: des, deischen

20 years agoUntangle building of AMD64 boot code.
Ruslan Ermilov [Sat, 7 Feb 2004 08:10:07 +0000 (08:10 +0000)]
Untangle building of AMD64 boot code.

Tested on: amd64 (sledge)

20 years agogetnetbyname fixes:
Diomidis Spinellis [Sat, 7 Feb 2004 07:30:02 +0000 (07:30 +0000)]
getnetbyname fixes:
Do not choke on malformed network addresses.
Return n_name in static space, not on the function's stack.

MFC after: 1 week

20 years agoAdd an entry about the removal of etc/rc.d/ttys.
Mike Makonnen [Sat, 7 Feb 2004 07:25:47 +0000 (07:25 +0000)]
Add an entry about the removal of etc/rc.d/ttys.

20 years ago- Demangle usage().
Ruslan Ermilov [Sat, 7 Feb 2004 07:22:11 +0000 (07:22 +0000)]
- Demangle usage().
- Require the "directory" argument with the -d option.

20 years agoNow that devfs is mandatory, there is no need to muck around
Mike Makonnen [Sat, 7 Feb 2004 07:10:25 +0000 (07:10 +0000)]
Now that devfs is mandatory, there is no need to muck around
with tty/pty permissions.

Noticed by: brooks
OKed: phk

20 years agoAdd a whole bunch of new stuff to make the driver for the AMD Am1771/Am1772
Bill Paul [Sat, 7 Feb 2004 06:44:13 +0000 (06:44 +0000)]
Add a whole bunch of new stuff to make the driver for the AMD Am1771/Am1772
802.11b chipset work. This chip is present on the SMC2602W version 3
NIC, which is what was used for testing. This driver creates kernel
threads (12 of them!) for various purposes, and required the following
routines:

PsCreateSystemThread()
PsTerminateSystemThread()
KeInitializeEvent()
KeSetEvent()
KeResetEvent()
KeInitializeMutex()
KeReleaseMutex()
KeWaitForSingleObject()
KeWaitForMultipleObjects()
IoGetDeviceProperty()

and several more. Also, this driver abuses the fact that NDIS events
and timers are actually Windows events and timers, and uses NDIS events
with KeWaitForSingleObject(). The NDIS event routines have been rewritten
to interface with the ntoskrnl module. Many routines with incorrect
prototypes have been cleaned up.

Also, this driver puts jobs on the NDIS taskqueue (via NdisScheduleWorkItem())
which block on events, and this interferes with the operation of
NdisMAllocateSharedMemoryAsync(), which was also being put on the
NDIS taskqueue. To avoid the deadlock, NdisMAllocateSharedMemoryAsync()
is now performed in the NDIS SWI thread instead.

There's still room for some cleanups here, and I really should implement
KeInitializeTimer() and friends.

20 years agoAdd a simple tool for checking the queue stats in the aac(4) driver. The
Scott Long [Sat, 7 Feb 2004 03:51:20 +0000 (03:51 +0000)]
Add a simple tool for checking the queue stats in the aac(4) driver.  The
ioctl interface for this has been here for years, so it might as well get
used.

20 years agoCheckpoint of work in progress in cleaning up target mode. It actually
Matt Jacob [Sat, 7 Feb 2004 03:47:33 +0000 (03:47 +0000)]
Checkpoint of work in progress in cleaning up target mode. It actually
seems to work well in RELENG_4. However, 5.X locking foo means that I'll
have to do some quick redesign.

Add ioctl handlers for ISP_GETROLE and ISP_SETROLE ioctls.

20 years agoadd a count for inotifies as well as atios.
Matt Jacob [Sat, 7 Feb 2004 03:44:43 +0000 (03:44 +0000)]
add a count for inotifies as well as atios.

MFC after: 1 week

20 years agoReverse role defines for initiator and target mode to better match the
Matt Jacob [Sat, 7 Feb 2004 03:43:27 +0000 (03:43 +0000)]
Reverse role defines for initiator and target mode to better match the
class 3 service parameters we'd get.

Steal 8 bits out of the portid u_int32_t for role information (port ids
are 24 bits anyway).

MFC after: 1 week

20 years agoAdd case to handle ISPCTL_GET_PDB.
Matt Jacob [Sat, 7 Feb 2004 03:42:17 +0000 (03:42 +0000)]
Add case to handle ISPCTL_GET_PDB.

MFC after: 1 week

20 years agoIf we're defined to have a default role for target mode, make it
Matt Jacob [Sat, 7 Feb 2004 03:39:42 +0000 (03:39 +0000)]
If we're defined to have a default role for target mode, make it
just ISP_ROLE_TARGET- not both.

MFC after: 1 week

20 years agoChange role defines so that they better match class 3 service parameters.
Matt Jacob [Sat, 7 Feb 2004 03:38:39 +0000 (03:38 +0000)]
Change role defines so that they better match class 3 service parameters.
Add ISPCTL_GET_PDB isp_control operation.

MFC after: 1 week

20 years agoReduce AAC_MAX_FIBS to work around some yet-unidentified bugs in the
Scott Long [Sat, 7 Feb 2004 03:30:32 +0000 (03:30 +0000)]
Reduce AAC_MAX_FIBS to work around some yet-unidentified bugs in the
handling of resources shortages.  The driver is now so fast that it can
completely fill all 512 slots on the card, but for some reason only 511
slots are being allocated.  Anything that tries to go into the 512th
slot gets silently lost.   Both bugs need to be fixed at a later date,
but this should fix the reports of hangs in getblk and vinvalb.

20 years ago- Broaden the scope of locking in aac_command_thread() again to catch some
Scott Long [Sat, 7 Feb 2004 03:26:38 +0000 (03:26 +0000)]
- Broaden the scope of locking in aac_command_thread() again to catch some
  edge cases in the loop.

- Try to grab a command before dequeueing the bio from the bioq.  The old
  behaviour of requeuing deferred bios to the end of the bioq is arguably
  wrong.  This should be fixed in the future to check the bioq head without
  automatically dequeueing the bio.

20 years agoAdd an #ifdef _KERNEL so that this file can be used from userland.
Scott Long [Sat, 7 Feb 2004 03:20:07 +0000 (03:20 +0000)]
Add an #ifdef _KERNEL so that this file can be used from userland.

20 years agoThe "bin" distribution was renamed to "base" in CURRENT.
Jimmy Olgeni [Sat, 7 Feb 2004 00:34:11 +0000 (00:34 +0000)]
The "bin" distribution was renamed to "base" in CURRENT.

20 years agoAllow decreasing access count even if there is no disk anymore.
Pawel Jakub Dawidek [Fri, 6 Feb 2004 23:10:49 +0000 (23:10 +0000)]
Allow decreasing access count even if there is no disk anymore.
This will allow closing disks that were removed while opened.

Approved by: phk, scottl (mentor)

20 years agoFix memory leak.
Lukas Ertl [Fri, 6 Feb 2004 22:51:04 +0000 (22:51 +0000)]
Fix memory leak.

PR:            kern/58634
Submitted by:  le
Approved by:   phk

20 years agoFirst round of cleanups to sys/boot/ makefiles:
Ruslan Ermilov [Fri, 6 Feb 2004 21:58:32 +0000 (21:58 +0000)]
First round of cleanups to sys/boot/ makefiles:

- do not use PROG for what's not a real C program,
- use sys.mk transformation rules where possible,
- only create the "machine" symlink on AMD64,
- removed MAINTAINER lines in individual makefiles,
- added the LIBSTAND defitinion to <bsd.libnames.mk>,
- somewhat better contents in .depend files.

Tested on: i386, amd64
Prodded by: bde

20 years agoNew release note: SA-04:02.
Bruce A. Mah [Fri, 6 Feb 2004 21:36:48 +0000 (21:36 +0000)]
New release note: SA-04:02.

20 years agoCorrect order of arguments given to checkparity and rebuildparity.
Lukas Ertl [Fri, 6 Feb 2004 21:07:17 +0000 (21:07 +0000)]
Correct order of arguments given to checkparity and rebuildparity.

Approved by: joerg (mentor)

20 years agoCorrect typo.
Lukas Ertl [Fri, 6 Feb 2004 21:05:42 +0000 (21:05 +0000)]
Correct typo.

20 years agoReduce the inline limit from 20000 to 8000 after the previous changes
Peter Wemm [Fri, 6 Feb 2004 20:40:04 +0000 (20:40 +0000)]
Reduce the inline limit from 20000 to 8000 after the previous changes
to simplify the curthread expansion some months ago.

20 years agoRemove the badsw* INVARIANTS checks. The events that this attempts
Peter Wemm [Fri, 6 Feb 2004 20:38:39 +0000 (20:38 +0000)]
Remove the badsw* INVARIANTS checks.  The events that this attempts
to catch are already nicely caught by trapping the null pointer derefs.
Remove no-longer-used noswitch/nothrow strings.  They were referenced
by the stub cpu_switch() etc functions before they were implemented.
Try something a little different for the lock prefixes.

Prompted by:  bde (the first two items anyway)

20 years agoRegen.
John Baldwin [Fri, 6 Feb 2004 20:21:18 +0000 (20:21 +0000)]
Regen.

20 years agoSync up MP safe flags with global syscalls.master. This includes write(),
John Baldwin [Fri, 6 Feb 2004 20:20:07 +0000 (20:20 +0000)]
Sync up MP safe flags with global syscalls.master.  This includes write(),
close(), getpid(), ibcs2_setuid(), getuid(), ptrace(), ibcs2_pause(),
ibcs2_nice(), ibcs2_kill(), ibcs2_pgrpsys(), dup(), pipe(), ibcs2_setgid(),
getgid(), umask(), and sigreturn().

20 years agoSync up MP safe flags with global syscalls.master. This includes read(),
John Baldwin [Fri, 6 Feb 2004 20:07:33 +0000 (20:07 +0000)]
Sync up MP safe flags with global syscalls.master.  This includes read(),
write(), close(), getpid(), setuid(), getuid(), svr4_sys_pause(),
svr4_sys_nice(), svr4_sys_kill(), svr4_sys_pgrpsys(), dup(), pipe(),
setgid(), getgid(), svr4_sys_signal(), umask(), getgroups(), setgroups(),
svr4_sys_sigprocmask(), svr4_sys_sigsuspend(), svr4_sys_sigaltstack(),
svr4_sys_sigaction(), svr4_sys_sigpending(), mprotect(), munmap(),
setegid(), seteuid(), setreuid(), setregid().

20 years agoMake all ACPI debug layers unique again. This makes debugging a more
Philip Paeps [Fri, 6 Feb 2004 19:38:05 +0000 (19:38 +0000)]
Make all ACPI debug layers unique again.  This makes debugging a more
pleasant experience (for certain definition of 'pleasant').

Submitted by: Mark Santcroos <marks@ripe.net>
Approved by: njl (mentor)

20 years ago- Convert the plimit lock to a pool mutex lock.
John Baldwin [Fri, 6 Feb 2004 19:35:14 +0000 (19:35 +0000)]
- Convert the plimit lock to a pool mutex lock.
- Hide struct plimit from userland.

Submitted by: bde (2)

20 years ago- Correct the translation of old rlimit values to properly handle the old
John Baldwin [Fri, 6 Feb 2004 19:30:12 +0000 (19:30 +0000)]
- Correct the translation of old rlimit values to properly handle the old
  RLIM_INFINITY case for ogetrlimit().
- Use %jd and intmax_t to output negative time in usec in calcru().
- Rework getrusage() to make a copy of the rusage struct into a local
  variable while holding Giant and then do the copyout from the local
  variable to avoid having to have the original process rusage struct
  locked while doing the copyout (which would not be safe).  This also
  includes a few style fixes from Bruce to getrusage().

Submitted by: bde (1, parts of 3)
Suggested by: bde (2)

20 years agoA few more style fixes from Bruce including a few I missed last time.
John Baldwin [Fri, 6 Feb 2004 19:25:34 +0000 (19:25 +0000)]
A few more style fixes from Bruce including a few I missed last time.

Submitted by: bde

20 years agoAdd cancellation point to sem_wait() and sem_timedwait() for pshared
Daniel Eischen [Fri, 6 Feb 2004 15:20:56 +0000 (15:20 +0000)]
Add cancellation point to sem_wait() and sem_timedwait() for pshared
semaphores.  Also add cancellation cleanup handlers to keep semaphores
in a consistent state.

Submitted in part by: davidxu
Reviewed by: davidxu

20 years agoPrevent memory leak on sem_destroy() by destroying the semaphore's
Daniel Eischen [Fri, 6 Feb 2004 15:15:44 +0000 (15:15 +0000)]
Prevent memory leak on sem_destroy() by destroying the semaphore's
internal mutex and CV.

20 years agoMFNetBSD:
MIHIRA Sanpei Yoshiro [Fri, 6 Feb 2004 14:38:14 +0000 (14:38 +0000)]
MFNetBSD:
netbsd.org -> NetBSD.org(rev.1.109)
fix typo in comment(rev.1.90)

add list of ``already merged''

20 years agoOnly allow the semaphore to be taken when the value is positive.
Daniel Eischen [Fri, 6 Feb 2004 13:54:38 +0000 (13:54 +0000)]
Only allow the semaphore to be taken when the value is positive.
Don't decrement it twice when it is taken.

Reported by: kris

20 years agoInherit BINDIR from a parent Makefile.inc.
Ruslan Ermilov [Fri, 6 Feb 2004 12:58:32 +0000 (12:58 +0000)]
Inherit BINDIR from a parent Makefile.inc.

20 years agoOnly include ../Makefile.inc once in loader/Makefile.
Ruslan Ermilov [Fri, 6 Feb 2004 12:45:27 +0000 (12:45 +0000)]
Only include ../Makefile.inc once in loader/Makefile.

20 years agoCommit simple workarounf for the "LiteOn" hang on boot problem.
Søren Schmidt [Fri, 6 Feb 2004 11:36:50 +0000 (11:36 +0000)]
Commit simple workarounf for the "LiteOn" hang on boot problem.

This allows the system to boot so I can get the world out of
my mailbox and get some work done to figure out what this mess
is all about.

20 years agoFrom what I can tell, install.sh supported all of the install(1)
Ruslan Ermilov [Fri, 6 Feb 2004 11:00:13 +0000 (11:00 +0000)]
From what I can tell, install.sh supported all of the install(1)
options since revision 1.2, so removed the XXX comment now.

Fixed bogus test(1) expression that would be evaluated as a null
string test if script is run without arguments, compacted option
lists.

20 years agoRemoved all the complexities when installing the strip(1) binary.
Ruslan Ermilov [Fri, 6 Feb 2004 09:25:59 +0000 (09:25 +0000)]
Removed all the complexities when installing the strip(1) binary.

Submitted by: bde

20 years agoTurn of ath since it causes a link failure without the hal till sam's
Peter Wemm [Fri, 6 Feb 2004 01:51:58 +0000 (01:51 +0000)]
Turn of ath since it causes a link failure without the hal till sam's
set up with a cross compiler and has the time to port the hal.

20 years agoMake LINT compile on amd64
Peter Wemm [Fri, 6 Feb 2004 01:47:52 +0000 (01:47 +0000)]
Make LINT compile on amd64

20 years agoMake this compile on amd64.
Peter Wemm [Fri, 6 Feb 2004 00:10:54 +0000 (00:10 +0000)]
Make this compile on amd64.

"I'll cope" by:  sam

20 years agoCorrect check for invalid semaphore on sem_destroy() (s/==/!=/).
Daniel Eischen [Thu, 5 Feb 2004 23:32:45 +0000 (23:32 +0000)]
Correct check for invalid semaphore on sem_destroy() (s/==/!=/).

Reported by: kris

20 years agomust convert protocol to sa type when preparing a DELETE message
Sam Leffler [Thu, 5 Feb 2004 23:19:17 +0000 (23:19 +0000)]
must convert protocol to sa type when preparing a DELETE message

Submitted by: Roselyn Lee <rosel@verniernetworks.com>
MFC after: 1 week

20 years agoFixes from 31-Jul-2003; make it 8-bit clean.
Ruslan Ermilov [Thu, 5 Feb 2004 23:16:31 +0000 (23:16 +0000)]
Fixes from 31-Jul-2003; make it 8-bit clean.

Forgotten by: ru (for far too long)

20 years agoThis commit was generated by cvs2svn to compensate for changes in r125505,
Ruslan Ermilov [Thu, 5 Feb 2004 23:16:31 +0000 (23:16 +0000)]
This commit was generated by cvs2svn to compensate for changes in r125505,
which included commits to RCS files with non-trunk default branches.

20 years agoAttempt to clean up the emu10k1-alsa.h stuff so that config doesn't
Peter Wemm [Thu, 5 Feb 2004 22:51:16 +0000 (22:51 +0000)]
Attempt to clean up the emu10k1-alsa.h stuff so that config doesn't
delete it each time its run and have it regenerated each time by make.
I used a quick hackish script rather than putting it in the files file
and used the before-depend rule to avoid the depend/no-depend hacks.

20 years agoFixed style of DPADD and LDADD assignments as per style.Makefile(5).
Ruslan Ermilov [Thu, 5 Feb 2004 22:44:25 +0000 (22:44 +0000)]
Fixed style of DPADD and LDADD assignments as per style.Makefile(5).

20 years agoCorrect the weak reference for sem_unlink.
Daniel Eischen [Thu, 5 Feb 2004 22:36:27 +0000 (22:36 +0000)]
Correct the weak reference for sem_unlink.

20 years agoCommit fix sent by Tor Egge <Tor.Egge@cvsup.no.freebsd.org>
Martin Cracauer [Thu, 5 Feb 2004 22:34:05 +0000 (22:34 +0000)]
Commit fix sent by Tor Egge <Tor.Egge@cvsup.no.freebsd.org>

Only use return value from system call if system call succeeded.

Tested with `make world` and some of my own scripts.

This should be MFCed soon.  While /bin/sh is hard to test the fix is
obviously correct and can be assumed not to break something else
(famous last words...).

20 years agoAdded missing DPADD.
Ruslan Ermilov [Thu, 5 Feb 2004 22:01:28 +0000 (22:01 +0000)]
Added missing DPADD.

20 years agoFixed ``make checkdpadd''. Removed redundant dependency.
Ruslan Ermilov [Thu, 5 Feb 2004 21:52:33 +0000 (21:52 +0000)]
Fixed ``make checkdpadd''.  Removed redundant dependency.

20 years agoRequire initrandom rather than random, since random requires
Dag-Erling Smørgrav [Thu, 5 Feb 2004 21:40:37 +0000 (21:40 +0000)]
Require initrandom rather than random, since random requires
mountcritlocal, which requires disks, which gbde_swap provides.

Noticed by: brooks

20 years agoDon't print the oldest line in the message buffer if the buffer is
Ian Dowse [Thu, 5 Feb 2004 21:07:50 +0000 (21:07 +0000)]
Don't print the oldest line in the message buffer if the buffer is
full, since that line is almost always incomplete. Make the parsing
of <%d> lines more strict.

Also simplify the logic a little:
 - Start off by making the buffer linear so that we don't have to
   deal with it wrapping around (suggested by bde).
 - Process line by line rather than byte at a time.