]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoFix fd leak. Threads people: does the call above to `access' need to be
Garrett Wollman [Fri, 3 May 2002 17:28:28 +0000 (17:28 +0000)]
Fix fd leak.  Threads people: does the call above to `access' need to be
underscored as well?

PR: 37717
Submitted by: fred@clift.org (slightly modified by me)

22 years agoGet rid of bogus holder for realloc(3).
Juli Mallett [Fri, 3 May 2002 17:13:29 +0000 (17:13 +0000)]
Get rid of bogus holder for realloc(3).

22 years agoRewrite the loop that handles substitution in the -I case, and try to make
Juli Mallett [Fri, 3 May 2002 17:05:25 +0000 (17:05 +0000)]
Rewrite the loop that handles substitution in the -I case, and try to make
it easier to understand.  Making it easy to understand isn't all that easy,
so litter the code with comments so some other poor soul can come along some
day and work on this if they see fit to do so.  Avoid calling strlen(3) to
check for a nil-string, when we can just check for *str=='\0'.

Approved by:    src/tools/regression/usr.bin/xargs/

22 years agoRename `lflag' to `Lflag' to match the option's case.
Juli Mallett [Fri, 3 May 2002 16:32:03 +0000 (16:32 +0000)]
Rename `lflag' to `Lflag' to match the option's case.

22 years agos/pfs_badop/vop_eopnotsupp/
Dag-Erling Smørgrav [Fri, 3 May 2002 14:58:25 +0000 (14:58 +0000)]
s/pfs_badop/vop_eopnotsupp/

Submitted by: phk

22 years agoRevert revision 1.22. I incorrectly modified the Berkeley identifiers.
Juli Mallett [Fri, 3 May 2002 14:50:31 +0000 (14:50 +0000)]
Revert revision 1.22.  I incorrectly modified the Berkeley identifiers.

Requested by: bde, mike

22 years agoRevert vendor identifiers, and #if 0 what is #if 0/#else for OpenBSD.
Juli Mallett [Fri, 3 May 2002 14:48:21 +0000 (14:48 +0000)]
Revert vendor identifiers, and #if 0 what is #if 0/#else for OpenBSD.

Add __FBSDID() properly.

Requested by: bde, mike

22 years agoRegenerate -I output since an argument bogon in the -I code was fixed, among
Juli Mallett [Fri, 3 May 2002 14:31:28 +0000 (14:31 +0000)]
Regenerate -I output since an argument bogon in the -I code was fixed, among
other things.

22 years agoUnbreak static build and remove usage() that isn't usage().
Dag-Erling Smørgrav [Fri, 3 May 2002 13:12:06 +0000 (13:12 +0000)]
Unbreak static build and remove usage() that isn't usage().

Reviewed by: bde

22 years agoThis commit was generated by cvs2svn to compensate for changes in r95978,
Dag-Erling Smørgrav [Fri, 3 May 2002 12:56:29 +0000 (12:56 +0000)]
This commit was generated by cvs2svn to compensate for changes in r95978,
which included commits to RCS files with non-trunk default branches.

22 years agoInitialize other to all-zeroes before using it (perforce change 10738)
Dag-Erling Smørgrav [Fri, 3 May 2002 12:56:29 +0000 (12:56 +0000)]
Initialize other to all-zeroes before using it (perforce change 10738)

22 years agoInitialize time_second to 1 instead of zero to pacify slightly bogus arp code.
Poul-Henning Kamp [Fri, 3 May 2002 08:46:03 +0000 (08:46 +0000)]
Initialize time_second to 1 instead of zero to pacify slightly bogus arp code.

Various minor style fixes from BDE.

22 years agoDon't grab the lock until somewhat later in attach to avoid a lock
Poul-Henning Kamp [Fri, 3 May 2002 08:44:53 +0000 (08:44 +0000)]
Don't grab the lock until somewhat later in attach to avoid a lock
reversal.

22 years agoName ufs_vop_[gs]etextattr() consistently with the rest of our VOPs and
Poul-Henning Kamp [Fri, 3 May 2002 08:40:33 +0000 (08:40 +0000)]
Name ufs_vop_[gs]etextattr() consistently with the rest of our VOPs and
put then in the ufs_vnops where they belong, rather than in the ffs_vnops.

Ok'ed by: rwatson
Sponsored by: DARPA & NAI Labs.

22 years agoAs malloc(9) and free(9) are now Giant-free, remove the Giant lock
Seigo Tanimura [Fri, 3 May 2002 07:46:59 +0000 (07:46 +0000)]
As malloc(9) and free(9) are now Giant-free, remove the Giant lock
across malloc(9) and free(9) of a pgrp or a session.

22 years agoThe first argument of pgsigio() is now struct sigio **.
Seigo Tanimura [Fri, 3 May 2002 07:45:26 +0000 (07:45 +0000)]
The first argument of pgsigio() is now struct sigio **.

22 years agoRemove Apple Airport card, since it seems to be mac-only.
Warner Losh [Fri, 3 May 2002 05:47:38 +0000 (05:47 +0000)]
Remove Apple Airport card, since it seems to be mac-only.
Change date to today.
Add socket low power WLAN CF card
Add SMC 2602W which I use all the time
Comment out the create-ibss example.  FreeBSD doesn't yet have that
media option.  Also change it to master-ibss since that's what OpenBSD
really uses (and that is a change in their man page too)

22 years agoAdditional patches from OpenBSD. Mostly grammar fixes and the like,
Warner Losh [Fri, 3 May 2002 05:39:16 +0000 (05:39 +0000)]
Additional patches from OpenBSD.  Mostly grammar fixes and the like,
however I'm adding the Dlink DWL520 as supported from OpenBSD.  Also
adding Dlink DWL650 since I have one in my hot little hands and it
works great.

# I suspect that OpenBSD needs this too :-)

22 years agoFix the lock order reversal between the sigio lock and a process/pgrp lock in
Seigo Tanimura [Fri, 3 May 2002 05:32:25 +0000 (05:32 +0000)]
Fix the lock order reversal between the sigio lock and a process/pgrp lock in
funsetownlst() by locking the sigio lock across funsetownlst().

22 years agoAdjust KINFO_PROC_SIZE due to segsz_t being changed from a 32-bit to
Marcel Moolenaar [Fri, 3 May 2002 01:41:37 +0000 (01:41 +0000)]
Adjust KINFO_PROC_SIZE due to segsz_t being changed from a 32-bit to
a 64-bit integral.

22 years agoPre-generate the optimized x86 crypto code and check it in rather than
Peter Wemm [Fri, 3 May 2002 00:14:39 +0000 (00:14 +0000)]
Pre-generate the optimized x86 crypto code and check it in rather than
depending on perl at build time.  Makefile.asm is a helper for after the
next import.

With my cvs@ hat on, the relatively small repo cost of this is acceptable,
especially given that we have other (much bigger) things like
lib*.so.gz.uu checked in under src/lib/compat/*.

Reviewed by: kris (maintainer)

22 years agobzero out the sync command buffer when sending commands. This was causing
Scott Long [Fri, 3 May 2002 00:07:50 +0000 (00:07 +0000)]
bzero out the sync command buffer when sending commands.  This was causing
problems when enumerating multiple arrays.

This is an MFC candidate.

22 years agoBring the wi man page into the 21st centry. This man page is based
Warner Losh [Thu, 2 May 2002 22:51:40 +0000 (22:51 +0000)]
Bring the wi man page into the 21st centry.  This man page is based
heavily on OpenBSD's wi man page, with OpenBSDisms replaced by
FreeBSDisms.  I also added a note about where the man page came from.
I hope that I've not broken anything that ru cleaned up.  We now run
this through tbl, but that appears to be automatic.

Obtained From:  OpenBSD

22 years agoMFNetBSD:
Josef Karthauser [Thu, 2 May 2002 22:47:37 +0000 (22:47 +0000)]
MFNetBSD:
    revision 1.94
    date: 2001/12/30 20:26:59;  author: augustss;  state: Exp;  lines: +5 -2
    Make sure we don't have any pending softintrs when entering polling mode.
    Thanks to Darrin for finding and fixing this problem when using USB
    keyboards in DDB.

22 years agoRetire makeobjops.pl - replaced by ../tools/makeobjops.awk.
Peter Wemm [Thu, 2 May 2002 22:21:59 +0000 (22:21 +0000)]
Retire makeobjops.pl - replaced by ../tools/makeobjops.awk.

22 years agoMake em driver compilable on IA64/alpha.
Prafulla Deuskar [Thu, 2 May 2002 22:21:01 +0000 (22:21 +0000)]
Make em driver compilable on IA64/alpha.

Submitted by: peter
MFC after: 3 days

22 years agoRetire makeLINT.pl
Peter Wemm [Thu, 2 May 2002 22:20:40 +0000 (22:20 +0000)]
Retire makeLINT.pl

22 years agoMakefile uses genwakecode.sh now.
Peter Wemm [Thu, 2 May 2002 22:17:35 +0000 (22:17 +0000)]
Makefile uses genwakecode.sh now.

22 years agoCleanup, quote:
Alfred Perlstein [Thu, 2 May 2002 22:03:19 +0000 (22:03 +0000)]
Cleanup, quote:

  This leaves some vestiges of the old locking, including style
  bugs in it.  I've only noticed anachronisms in socketvar.h so far
  (I've merged net* but not kern or all of sys).  The patch also
  has old fixes for style bugs in accf stuff and namespace pollution
  in uma...  The largest style bugs are line continued backslashes
  in column 80 and (these are fixed), and starting the do-while
  code for the new macros in column 40, which is quite unlike the
  usual indentation (see sys/queue.h) and not even like the indentation
  for the old macros (column 32) (this is not fixed).

Submitted by: bde

22 years agoAs promised make the hack for sizeof(struct disklabel) on alpha annoying.
Poul-Henning Kamp [Thu, 2 May 2002 21:53:39 +0000 (21:53 +0000)]
As promised make the hack for sizeof(struct disklabel) on alpha annoying.

Run make world (or recompile whatever program whines) to get rid of warning.

Compat bits will be removed entirely in about two weeks.

22 years agoSwap the order of mount and nmount calls. We now call nmount
Maxime Henrion [Thu, 2 May 2002 20:52:04 +0000 (20:52 +0000)]
Swap the order of mount and nmount calls.  We now call nmount
first, since all the consumers of mount_std in the tree have
been converted to nmount.

22 years agoConvert devfs to nmount.
Maxime Henrion [Thu, 2 May 2002 20:27:42 +0000 (20:27 +0000)]
Convert devfs to nmount.

Reviewed by: phk

22 years agoConvert the pseudofs framework to nmount (thus procfs and linprocfs).
Maxime Henrion [Thu, 2 May 2002 20:25:55 +0000 (20:25 +0000)]
Convert the pseudofs framework to nmount (thus procfs and linprocfs).

Reviewed by: des (some time ago), phk

22 years agoConvert fdescfs to nmount.
Maxime Henrion [Thu, 2 May 2002 20:24:50 +0000 (20:24 +0000)]
Convert fdescfs to nmount.

Reviewed by: phk

22 years agoDon't reference vop_std* since they are already implicitly
Scott Long [Thu, 2 May 2002 20:23:47 +0000 (20:23 +0000)]
Don't reference vop_std* since they are already implicitly
referenced through the VOP_DEFAULT vector

Submitted by: phk

22 years agoModernize
Poul-Henning Kamp [Thu, 2 May 2002 20:09:14 +0000 (20:09 +0000)]
Modernize

22 years agoMake these functions cancellation points like they should be:
Archie Cobbs [Thu, 2 May 2002 19:58:43 +0000 (19:58 +0000)]
Make these functions cancellation points like they should be:
poll(2), readv(2), select(2), wait4(2), writev(2).

PR: bin/37658
Reviewed by: deischen
MFC after: 1 week

22 years agoMake sure calls to pthread_cancel() do not take effect if the target
Archie Cobbs [Thu, 2 May 2002 19:57:59 +0000 (19:57 +0000)]
Make sure calls to pthread_cancel() do not take effect if the target
thread is already exiting.

PR: bin/37614
Reviewed by: deischen
MFC after: 1 week

22 years agoRevert rev 1.1018. rp(4) and dgb(4) are deemed MI, while the rest of the
David E. O'Brien [Thu, 2 May 2002 19:42:38 +0000 (19:42 +0000)]
Revert rev 1.1018.  rp(4) and dgb(4) are deemed MI, while the rest of the
drivers for simular hardware are i386-specific.  That is why I did not
find the information here.

22 years agoUse vop_panic() instead of our home-rolled version.
Poul-Henning Kamp [Thu, 2 May 2002 19:15:52 +0000 (19:15 +0000)]
Use vop_panic() instead of our home-rolled version.

22 years agoUse vop_panic() instead of rolling our own.
Poul-Henning Kamp [Thu, 2 May 2002 19:13:44 +0000 (19:13 +0000)]
Use vop_panic() instead of rolling our own.

22 years agoSet the permissions on restored symbolic links.
Ian Dowse [Thu, 2 May 2002 17:39:19 +0000 (17:39 +0000)]
Set the permissions on restored symbolic links.

PR: bin/37665
Submitted by: "Michael C. Adler" <mad1@tapil.com>

22 years ago o Change the implementation of vm_map locking to use exclusive locks
Alan Cox [Thu, 2 May 2002 17:32:27 +0000 (17:32 +0000)]
 o Change the implementation of vm_map locking to use exclusive locks
   exclusively.  The interface still, however, distinguishes
   between a shared lock and an exclusive lock.

22 years agoJoin the pissing contest: generate LINT with a single sed(1) command.
Dag-Erling Smørgrav [Thu, 2 May 2002 16:34:47 +0000 (16:34 +0000)]
Join the pissing contest: generate LINT with a single sed(1) command.
Smaller script, smaller (though equivalent) output.

22 years agoAdd missing UNLOCKs
Julian Elischer [Thu, 2 May 2002 15:58:04 +0000 (15:58 +0000)]
Add missing UNLOCKs

Submitted by: dirkx@covalent.net

22 years ago- Protect randompid and nprocs with the allproc_lock.
John Baldwin [Thu, 2 May 2002 15:13:45 +0000 (15:13 +0000)]
- Protect randompid and nprocs with the allproc_lock.
- Reorder fork1() to do malloc() and other blocking operations prior to
  acquiring the needed process locks.
- The new process inherit's the credentials of curthread, not the
  credentials of the old process.
- Document a really weird race that will come up with KSE allows multiple
  kernel threads per process.

22 years ago- Reorder a few things so that when we lock the process at the end of
John Baldwin [Thu, 2 May 2002 15:09:58 +0000 (15:09 +0000)]
- Reorder a few things so that when we lock the process at the end of
  exit1() we don't have to release it until we acquire schd_lock to
  call cpu_throw().
- Since we can switch at any time due to preemption or a lock release
  prior to acquiring sched_lock, don't update switchtime and switchticks
  until the very end of exit1() after we have acquired sched_lock.
- Interlock the proctree_lock and proc lock in wait1() and exit1() to
  avoid lost wakeups when a parent blocks waiting for a child to exit at
  the bottom of wait1().  In exit1() the proc lock interlocked with
  proctree_lock (and released after acquiring sched_lock) is that of
  the parent process.
- In wait1() use an exclusive lock of proctree lock while we are
  looking for a process to harvest.  This allows us to completely
  remove all references to the process once we've found one (i.e.,
  disconnect it from pgrp's, session's, zombproc list, and it's parent's
  children list) "atomically" without needing to worry about a lock
  upgrade.
- We don't need sched_lock to test if p_stat is SZOMB or SSTOP when holding
  the proc lock since the proc lock is always held with p_stat is set to
  SZOMB or SSTOP.
- Protect nprocs with an xlock of the allproc_lock.

22 years ago- Reorder execve() so that it performs blocking operations before it
John Baldwin [Thu, 2 May 2002 15:00:14 +0000 (15:00 +0000)]
- Reorder execve() so that it performs blocking operations before it
  locks the process.
- Defer other blocking operations such as vrele()'s until after we
  release locks.
- execsigs() now requires the proc lock to be held when it is called
  rather than locking the process internally.

22 years agoUn-rot the VOP_ table generator.
Poul-Henning Kamp [Thu, 2 May 2002 13:51:04 +0000 (13:51 +0000)]
Un-rot the VOP_ table generator.

22 years agoMake pkg_info working again when argument is a file not in the current
Maxim Sobolev [Thu, 2 May 2002 13:09:33 +0000 (13:09 +0000)]
Make pkg_info working again when argument is a file not in the current
directory.

PR: 37571
Submitted by: Dirk Meyer <dirk.meyer@dinoex.sub.org>
MFC after: 1 week

22 years agoMake `-g' working for symlinks.
Maxim Sobolev [Thu, 2 May 2002 12:52:42 +0000 (12:52 +0000)]
Make `-g' working for symlinks.

PR: 33857
Submitted by: darren <igla@batterybackups.net>
MFC after: 1 week

22 years agoKill local variable which shadows global one. This fixes creation of bzip2
Maxim Sobolev [Thu, 2 May 2002 10:19:13 +0000 (10:19 +0000)]
Kill local variable which shadows global one. This fixes creation of bzip2
compressed packages broken in the previous commit.

Pointy hat to: obrien

22 years agoHide a pointer to the malloc_type bucket at the end of the freed memory. If
Jeff Roberson [Thu, 2 May 2002 09:07:04 +0000 (09:07 +0000)]
Hide a pointer to the malloc_type bucket at the end of the freed memory.  If
this memory is modified after it has been freed we can now report it's
previous owner.

22 years agoMove around the dbg code a bit so it's always under a lock. This stops a
Jeff Roberson [Thu, 2 May 2002 09:05:36 +0000 (09:05 +0000)]
Move around the dbg code a bit so it's always under a lock.  This stops a
weird potential race if we were preempted right as we were doing the dbg
checks.

22 years agoThe width of segsz_t should be 64, not 32 on ia64.
Doug Rabson [Thu, 2 May 2002 09:04:29 +0000 (09:04 +0000)]
The width of segsz_t should be 64, not 32 on ia64.

22 years agoAdd the csplit(1) utility, which splits files based on context, as specified
Tim J. Robbins [Thu, 2 May 2002 07:54:04 +0000 (07:54 +0000)]
Add the csplit(1) utility, which splits files based on context, as specified
by SUSv3.

This Makefile got left out of the previous commit.

PR:             36191
Reviewed by:    mike

22 years agoAdd the csplit(1) utility, which splits files based on context, as specified
Tim J. Robbins [Thu, 2 May 2002 07:46:36 +0000 (07:46 +0000)]
Add the csplit(1) utility, which splits files based on context, as specified
by SUSv3.

PR: 36191
Reviewed by: mike

22 years ago- Changed the size element of uma_zctor_args to be size_t instead of int.
Andrew R. Reiter [Thu, 2 May 2002 07:36:30 +0000 (07:36 +0000)]
- Changed the size element of uma_zctor_args to be size_t instead of int.
- Changed uma_zcreate to accept the size argument as a size_t intead of
  int.

Approved by: jeff

22 years ago- Replaced the Atm_connection storage pool with an uma_zone of
Andrew R. Reiter [Thu, 2 May 2002 07:27:29 +0000 (07:27 +0000)]
- Replaced the Atm_connection storage pool with an uma_zone of
  Atm_connection items.
- Replaced the Atm_connvc storage pool with an uma_zone of Atm_connvc
  items.
- Created void atm_cm_init(void *) and added it to the netatm init code.
  I'm thinking that there will definetly be more ``subsystem'' init
  functions to be added so I'll probably change these calls to be a for
  loop through init routines (or something).

22 years agomalloc/free(9) no longer require Giant. Use the malloc_mtx to protect the
Jeff Roberson [Thu, 2 May 2002 07:22:19 +0000 (07:22 +0000)]
malloc/free(9) no longer require Giant.  Use the malloc_mtx to protect the
mallochash.  Mallochash is going to go away as soon as I introduce the
kfree/kmalloc api and partially overhaul the malloc wrapper.  This can't happen
until all users of the malloc api that expect memory to be aligned on the size
of the allocation are fixed.

22 years agoUse shell script version (using awk and sed) of makeLINT.pl.
Jun Kuriyama [Thu, 2 May 2002 06:10:09 +0000 (06:10 +0000)]
Use shell script version (using awk and sed) of makeLINT.pl.

22 years agoDon't cast incorrectly to malloc(3), and don't use errx(3) if malloc(3)
Juli Mallett [Thu, 2 May 2002 05:46:37 +0000 (05:46 +0000)]
Don't cast incorrectly to malloc(3), and don't use errx(3) if malloc(3)
returns NULL, as malloc(3) sets errno.  Use err(3).

22 years agoIn pmap_pinit0, remove duplicate initialization.
Marcel Moolenaar [Thu, 2 May 2002 05:35:02 +0000 (05:35 +0000)]
In pmap_pinit0, remove duplicate initialization.

22 years agoPCPU(current_pmap) is initialized in pmap_bootstrap. No need to
Marcel Moolenaar [Thu, 2 May 2002 05:30:42 +0000 (05:30 +0000)]
PCPU(current_pmap) is initialized in pmap_bootstrap. No need to
do it again.

22 years agoRemove redundant declarations of getopt(3) externals (since <unistd.h> does
Juli Mallett [Thu, 2 May 2002 05:25:23 +0000 (05:25 +0000)]
Remove redundant declarations of getopt(3) externals (since <unistd.h> does
take care of them), and add __FreeBSD__ to the defined() checks for the
_PATH_DIVNAME.

22 years ago__COPYRIGHT() and __SCCSID().
Juli Mallett [Thu, 2 May 2002 05:21:35 +0000 (05:21 +0000)]
__COPYRIGHT() and __SCCSID().

22 years agoAn explicit cast to size_t for an inline integer, since the two are not the
Juli Mallett [Thu, 2 May 2002 05:07:13 +0000 (05:07 +0000)]
An explicit cast to size_t for an inline integer, since the two are not the
same on Alpha and lint(1) pointed that out.

lint(1) on the same architecture pointed out how silly a cast to (u_int) to
malloc(3)'s argument was.  Change that to size_t.

22 years agoPAMify rexecd(8).
Dag-Erling Smørgrav [Thu, 2 May 2002 05:06:32 +0000 (05:06 +0000)]
PAMify rexecd(8).

Sponsored by: DARPA, NAI Labs

22 years agoAdd a PAM policy for rexecd(8).
Dag-Erling Smørgrav [Thu, 2 May 2002 05:05:28 +0000 (05:05 +0000)]
Add a PAM policy for rexecd(8).

Sponsored by: DARPA, NAI Labs

22 years agoIn udf_bmap(), return the physical block number, not the logical
Scott Long [Thu, 2 May 2002 05:01:14 +0000 (05:01 +0000)]
In udf_bmap(), return the physical block number, not the logical
block number.  This fixes things like cp (ouch!) which use mmap.

22 years agoxdm plays horrid tricks with PAM, and dumps core if it's allowed to call
Dag-Erling Smørgrav [Thu, 2 May 2002 05:00:40 +0000 (05:00 +0000)]
xdm plays horrid tricks with PAM, and dumps core if it's allowed to call
pam_lastlog, so add a dummy session chain to avoid using the one from
pam.d/other.  I assume gdm does something similar, so give it a dummy
session chain as well.

Sponsored by: DARPA, NAI Labs.

22 years agoAdd openpam_nullconv.c to SRCS.
Dag-Erling Smørgrav [Thu, 2 May 2002 04:42:59 +0000 (04:42 +0000)]
Add openpam_nullconv.c to SRCS.

22 years agoVendor import of OpenPAM Cinnamon.
Dag-Erling Smørgrav [Thu, 2 May 2002 04:40:21 +0000 (04:40 +0000)]
Vendor import of OpenPAM Cinnamon.

22 years agoThis commit was generated by cvs2svn to compensate for changes in r95908,
Dag-Erling Smørgrav [Thu, 2 May 2002 04:40:21 +0000 (04:40 +0000)]
This commit was generated by cvs2svn to compensate for changes in r95908,
which included commits to RCS files with non-trunk default branches.

22 years agoA markup fix, and document -R as non-standard.
Juli Mallett [Thu, 2 May 2002 03:29:38 +0000 (03:29 +0000)]
A markup fix, and document -R as non-standard.

22 years agoMake -J a proper flag internal to the code (rather than just check for
Juli Mallett [Thu, 2 May 2002 02:42:34 +0000 (02:42 +0000)]
Make -J a proper flag internal to the code (rather than just check for
use of replstr and lack of Iflag), and add -R, which when given with
-I controls the number of arguments on which replacement will be done.

Some people happen to think it's idiotic to limit to 5 arguments, so
let the user override it if they like.

22 years ago o Remove dead and lockmgr()-specific debugging code.
Alan Cox [Thu, 2 May 2002 02:32:09 +0000 (02:32 +0000)]
 o Remove dead and lockmgr()-specific debugging code.

22 years ago*str is spelled 's1' inside the body of the code.
Juli Mallett [Thu, 2 May 2002 02:15:06 +0000 (02:15 +0000)]
*str is spelled 's1' inside the body of the code.

Make 'this' not a local variable, since it isn't necessary or complex enough
to warrant such.

22 years agoRemove the temporary alignment check in free().
Jeff Roberson [Thu, 2 May 2002 02:08:48 +0000 (02:08 +0000)]
Remove the temporary alignment check in free().

Implement the following checks on freed memory in the bucket path:
- Slab membership
- Alignment
- Duplicate free

This previously was only done if we skipped the buckets.  This code will slow
down INVARIANTS a bit, but it is smp safe.  The checks were moved out of the
normal path and into hooks supplied in uma_dbg.

22 years agoRework some of the -I support so I can't find a way to make xargs(1) core or
Juli Mallett [Thu, 2 May 2002 02:06:03 +0000 (02:06 +0000)]
Rework some of the -I support so I can't find a way to make xargs(1) core or
corrupt memory.  Simplifies the code in one or two places, also removes some
code that looks like it was bogus or incomplete.

Update strnsubst to have one or two extra conditions which maybe would make
it more efficient, or at least more versatile.  This is likely a no-op.

22 years agoMinor bug fixes uncovered while working on UFS2.
Kirk McKusick [Thu, 2 May 2002 01:03:06 +0000 (01:03 +0000)]
Minor bug fixes uncovered while working on UFS2.
They should have no effect on existing systems.

22 years agoAdd some NOTES on the Comtrol Rocketport and the Digiboard drivers.
David E. O'Brien [Wed, 1 May 2002 23:03:31 +0000 (23:03 +0000)]
Add some NOTES on the Comtrol Rocketport and the Digiboard drivers.

22 years agoUsual after-import fixup of SCM IDs.
David E. O'Brien [Wed, 1 May 2002 22:39:53 +0000 (22:39 +0000)]
Usual after-import fixup of SCM IDs.

22 years agoSave the MCA info specific to the AP as part of the AP launch.
Marcel Moolenaar [Wed, 1 May 2002 22:19:03 +0000 (22:19 +0000)]
Save the MCA info specific to the AP as part of the AP launch.

22 years agoMake ia64_mca_save_state MP safe. Protect access to the info block,
Marcel Moolenaar [Wed, 1 May 2002 22:10:03 +0000 (22:10 +0000)]
Make ia64_mca_save_state MP safe. Protect access to the info block,
updating the sysctl tree and clearing the SAL state by a spin lock.

22 years agoIf we get a DATA UNDERRUN error from QLogic FC cards, but the RQCS_RU bit
Matt Jacob [Wed, 1 May 2002 21:58:36 +0000 (21:58 +0000)]
If we get a DATA UNDERRUN error from QLogic FC cards, but the RQCS_RU bit
is not set in the scsi completion status, or if the residual is clearly
nonsense, then this was a command that suffered the loss of one or more
FC frames in the middle of the exchange.

Set HBA_BOTCH and hope it will get retried. It's the only thing we can do.

MFC after: 1 day

22 years agoTry using genwakecode.sh instead of the perl version.
Peter Wemm [Wed, 1 May 2002 21:54:07 +0000 (21:54 +0000)]
Try using genwakecode.sh instead of the perl version.

22 years agoAdd a shell script to do what genwakecode.pl does.
Peter Wemm [Wed, 1 May 2002 21:52:34 +0000 (21:52 +0000)]
Add a shell script to do what genwakecode.pl does.

Obtained from: ade

22 years agoAdd some spurious veritcal whitespace to reduce diffs to OpenBSD.
Juli Mallett [Wed, 1 May 2002 21:45:22 +0000 (21:45 +0000)]
Add some spurious veritcal whitespace to reduce diffs to OpenBSD.

22 years agoMerge local changes again, against ANSIfied m4(1).
Juli Mallett [Wed, 1 May 2002 21:37:29 +0000 (21:37 +0000)]
Merge local changes again, against ANSIfied m4(1).

22 years agoRedo the sigio locking.
Alfred Perlstein [Wed, 1 May 2002 20:44:46 +0000 (20:44 +0000)]
Redo the sigio locking.

Turn the sigio sx into a mutex.

Sigio lock is really only needed to protect interrupts from dereferencing
the sigio pointer in an object when the sigio itself is being destroyed.

In order to do this in the most unintrusive manner change pgsigio's
sigio * argument into a **, that way we can lock internally to the
function.

22 years agoWe don't need no stinkin' echos here.
Warner Losh [Wed, 1 May 2002 19:24:26 +0000 (19:24 +0000)]
We don't need no stinkin' echos here.

Instead, don't run kldxref if you don't have one on your system.

22 years agoAdd an epic_stats_update() function (called once per second). Even though
Semen Ustimenko [Wed, 1 May 2002 19:23:04 +0000 (19:23 +0000)]
Add an epic_stats_update() function (called once per second). Even though
we don't collect any stats in it, we mii_tick() in it! This fix the bug
when autonegotiating fullduplex modes.

Also, pause activity before setting TXCON in epic_miibus_statchg(). Though
i've never seen problmes from not doing that, the documentation says we
need to do it.

MFC after: 1 week

22 years agoAdd support for using the profiled versions of the C++ (and related) libs.
David E. O'Brien [Wed, 1 May 2002 19:19:22 +0000 (19:19 +0000)]
Add support for using the profiled versions of the C++ (and related) libs.

22 years agoMII_TICK handlers need not restart aneg on these PHYs, they behave pretty
Semen Ustimenko [Wed, 1 May 2002 19:15:54 +0000 (19:15 +0000)]
MII_TICK handlers need not restart aneg on these PHYs, they behave pretty
well as is, so - just fetch current status upon MII_TICK.

Also do IFM_INST verification at the top of *_service() then doing it
separately for every case in switch.

acphy: do not read MII_ACPHY_DIAG twice, there is nothing latching.
qsphy: always fetch actual link status from MII_QSPHY_PCTL.

MFC after: 1 week

22 years agoAxe unused SESS_UNLOCK_NOSWITCH() and PGRP_UNLOCK_NOSWITCH() macros. The
John Baldwin [Wed, 1 May 2002 18:11:16 +0000 (18:11 +0000)]
Axe unused SESS_UNLOCK_NOSWITCH() and PGRP_UNLOCK_NOSWITCH() macros.  The
MTX_NOSWITCH flag was deprecated a while ago.

22 years agoBump the version to mark the fixed FSM TLD ordering
Brian Somers [Wed, 1 May 2002 10:08:10 +0000 (10:08 +0000)]
Bump the version to mark the fixed FSM TLD ordering

22 years agoIf we receive a config request while Open, call TLD *before* decoding
Brian Somers [Wed, 1 May 2002 10:04:49 +0000 (10:04 +0000)]
If we receive a config request while Open, call TLD *before* decoding
the request as TLD may initialise fields that are set when decoding the
request.

MFC after: 1 week

22 years agoFix some edge cases where bad string handling could occur.
Alfred Perlstein [Wed, 1 May 2002 08:29:41 +0000 (08:29 +0000)]
Fix some edge cases where bad string handling could occur.

Submitted by: ps

22 years agocleanup:
Alfred Perlstein [Wed, 1 May 2002 08:08:24 +0000 (08:08 +0000)]
cleanup:
fix line wraps, add some comments, fix macro definitions, fix for(;;) loops.