]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoDocument the relation between $command and $foo_program.
yar [Wed, 22 Feb 2006 10:57:19 +0000 (10:57 +0000)]
Document the relation between $command and $foo_program.

PR: docs/68453
MFC after: 3 days

18 years agoDo not use bio structure after g_io_deliver(), it may not longer by valid.
pjd [Wed, 22 Feb 2006 10:21:05 +0000 (10:21 +0000)]
Do not use bio structure after g_io_deliver(), it may not longer by valid.

Found and fixed by: Vsevolod Lobko <seva@ip.net.ua>
MFC after: 3 days

18 years ago - Revert r1.406 until a solution can be found that doesn't break nfs. The
jeff [Wed, 22 Feb 2006 09:52:25 +0000 (09:52 +0000)]
 - Revert r1.406 until a solution can be found that doesn't break nfs.  The
   statfs handler in nfs will lock vnodes which may lead to deadlock or
   recursion.

Found by: kris
Pointy hat to: me

18 years agoMention NO_NLS_CATALOGS.
ume [Wed, 22 Feb 2006 09:47:54 +0000 (09:47 +0000)]
Mention NO_NLS_CATALOGS.

Suggested by: Niclas Zeising <lothrandil__at__n00b.apagnu.se>

18 years ago - We must hold a reference to a vnode before calling vgone() otherwise
jeff [Wed, 22 Feb 2006 09:05:40 +0000 (09:05 +0000)]
 - We must hold a reference to a vnode before calling vgone() otherwise
   it may not be removed from the freelist.

MFC After: 1 week
Found by: kris

18 years ago - We can not hold a vnode lock while we do a lookup. Search for and load
jeff [Wed, 22 Feb 2006 06:29:55 +0000 (06:29 +0000)]
 - We can not hold a vnode lock while we do a lookup.  Search for and load
   modules prior to looking up the directory which we will cover to avoid
   this problem in mount.
 - We must hold the coveredvp locked before we can busy the mountpoint to
   prevent a lock order reversal with the vfs_busy() in lookup which holds
   the directory lock prior to doing a vfs_busy().  The directory lock is
   required to safely clear the v_mountedhere field on the directory.

MFC After: 1 week

18 years ago - Grab a mnt ref in vfs_busy() before dropping the interlock. This will
jeff [Wed, 22 Feb 2006 06:20:12 +0000 (06:20 +0000)]
 - Grab a mnt ref in vfs_busy() before dropping the interlock.  This will
   prevent the mount point from going away while we're waiting on the lock.
   The ref does not need to persist once we have the lock because the
   lock prevents the mount point from being unmounted.

MFC After: 1 week

18 years ago - Hold the vnode used in the statfs related functions until we're done with
jeff [Wed, 22 Feb 2006 06:19:08 +0000 (06:19 +0000)]
 - Hold the vnode used in the statfs related functions until we're done with
   the VFS_STATFS call to prevent the mount from disappearing while we're
   stating.
 - Convert these routines to use MPSAFE namei semantics.

MFC After: 1 week

18 years ago - spell VOP_LOCK(vp, LK_RELEASE... VOP_UNLOCK(vp,... so that asserts in
jeff [Wed, 22 Feb 2006 06:17:31 +0000 (06:17 +0000)]
 - spell VOP_LOCK(vp, LK_RELEASE... VOP_UNLOCK(vp,... so that asserts in
   vop_lock_post do not trigger.
 - Rearrange null_inactive to null_hashrem earlier so there is no chance
   of finding the null node on the hash list after the locks have been
   switched.
 - We should never have a NULL lowervp in null_reclaim() so there is
   no need to handle this situation.  panic instead.

MFC After: 1 week

18 years ago - Assert that the lowervp is locked in null_hashget().
jeff [Wed, 22 Feb 2006 06:15:12 +0000 (06:15 +0000)]
 - Assert that the lowervp is locked in null_hashget().
 - Simplify the logic dealing with recycled vnodes in null_hashget() and
   null_hashins().  Since we hold the lower node locked in both cases
   the null node can not be undergoing recycling unless reclaim somehow
   called null_nodeget().  The logic that was in place was not safe and
   was essentially dead code.

MFC After: 1 week

18 years ago - Using LK_NOWAIT in qsync() can get us into infinite loop situations that
jeff [Wed, 22 Feb 2006 06:12:53 +0000 (06:12 +0000)]
 - Using LK_NOWAIT in qsync() can get us into infinite loop situations that
   lead to deadlocks.  Remove it.

MFC After: 1 week

18 years ago - Deadfs should not use the std GETWRITEMOUNT routine. Add one that always
jeff [Wed, 22 Feb 2006 06:11:59 +0000 (06:11 +0000)]
 - Deadfs should not use the std GETWRITEMOUNT routine.  Add one that always
   returns NULL.

MFC After: 1 week

18 years agoRemove commented out qualifier to dumping a message.
mjacob [Wed, 22 Feb 2006 05:19:50 +0000 (05:19 +0000)]
Remove commented out qualifier to dumping a message.

18 years agoAdd option -w to specify graph width.
mdodd [Wed, 22 Feb 2006 04:10:20 +0000 (04:10 +0000)]
Add option -w to specify graph width.
Use COLUMNS, terminal width for default graph width.

Reviewed by:  rwatson

18 years agoMention NO_NLS_CATALOGS.
ume [Wed, 22 Feb 2006 03:42:56 +0000 (03:42 +0000)]
Mention NO_NLS_CATALOGS.

18 years agoAdd multi-process support to simple http micro-benchmark, now the default.
rwatson [Wed, 22 Feb 2006 02:54:43 +0000 (02:54 +0000)]
Add multi-process support to simple http micro-benchmark, now the default.
Use -t for thread support.  Also added are -n to specify number of threads
or process, and -s to set the length of the test.

18 years agoAdd multi-process support to simple httpd micro-benchmark, which is now
rwatson [Wed, 22 Feb 2006 02:54:04 +0000 (02:54 +0000)]
Add multi-process support to simple httpd micro-benchmark, which is now
the default.  With -t, threads are used.

18 years agoDisable POSIX SIGEV_THREAD notification support, soon we will have a common
davidxu [Wed, 22 Feb 2006 02:52:22 +0000 (02:52 +0000)]
Disable POSIX SIGEV_THREAD notification support, soon we will have a common
library which can be reused both for libthr and libpthread.

18 years agoAbstract function mqfs_create_node() to create a mqueue node.
davidxu [Wed, 22 Feb 2006 02:38:25 +0000 (02:38 +0000)]
Abstract function mqfs_create_node() to create a mqueue node.

18 years agoIf block size is zero, use normal file operations to do I/O,
davidxu [Wed, 22 Feb 2006 00:05:12 +0000 (00:05 +0000)]
If block size is zero, use normal file operations to do I/O,
this eliminates a divided-by-zero fault.

Recommended by: phk

18 years agouse the specified key index for non-group keys; this fixes static
sam [Tue, 21 Feb 2006 23:55:38 +0000 (23:55 +0000)]
use the specified key index for non-group keys; this fixes static
wep key configure at key indices > 0 and 802.1x/EAPOL operation
with ap's that want the station to install a key at indices > 0.

Hard work by: Joe Love
Reviewed by: avatar
MFC after: 1 week

18 years agohonor user-specified key index for global key slots
sam [Tue, 21 Feb 2006 23:51:19 +0000 (23:51 +0000)]
honor user-specified key index for global key slots

Submitted by: Joe Love
Reviewed by: avatar
MFC after: 1 week

18 years agoLock the vm_object while checking its type to see if it is a vnode-backed
jhb [Tue, 21 Feb 2006 22:09:54 +0000 (22:09 +0000)]
Lock the vm_object while checking its type to see if it is a vnode-backed
object that requires Giant in vm_object_deallocate().  This is somewhat
hairy in that if we can't obtain Giant directly, we have to drop the
object lock, then lock Giant, then relock the object lock and verify that
we still need Giant.  If we don't (because the object changed to OBJT_DEAD
for example), then we drop Giant before continuing.

Reviewed by: alc
Tested by: kris

18 years agoMove the ruadd() in kern_exit() to save our final stats in our child
jhb [Tue, 21 Feb 2006 21:48:42 +0000 (21:48 +0000)]
Move the ruadd() in kern_exit() to save our final stats in our child
stats even further down in exit1() so that it includes the runtime and
tick counts from the final time slice for the dying thread.

Reviewed by: phk

18 years agoSplit calcru() back into a calcru1() function shared with calccru() and
jhb [Tue, 21 Feb 2006 21:47:46 +0000 (21:47 +0000)]
Split calcru() back into a calcru1() function shared with calccru() and
a calcru() wrapper that passes a local rusage_ext on the stack that is
a snapshot to do the calculations on.  Now we can pass p->p_crux to
calcru1() in calccru() again which fixes the issues with runtime going
backwards messages when dead processes are harvested by init.

Reviewed by: phk
Tested by: Stefan Ehmann shoesoft at gmx dot net

18 years ago- In the interrupt handler clear the interrupt source flags before
marius [Tue, 21 Feb 2006 20:20:43 +0000 (20:20 +0000)]
- In the interrupt handler clear the interrupt source flags before
  processing the interrupt events. If we clear them afterwards we
  can completely miss some events as the NIC can change the source
  flags while we're in the handler. In order to not get another
  interrupt while we're in ifp->if_input() with the driver lock
  dropped we now turn off NIC interrupts while in the interrupt
  handler. Previously this was meant to be achieved by clearing the
  interrupt source flags after processing the interrupt events but
  didn't really work as clearing these flags doesn't actually
  acknowledge and re-enable the events.
  This fixes the device timeouts seen with the VMware LANCE.
- Relax the watchdog timer somewhat; don't enable it until the last
  packet is enqueued and if there is a TX interrupt but there are
  still outstanding ones reload the timer.

Reported and tested by: Morten Rodal <morten@rodal.no>
MFC after: 3 days

18 years agoNO_WPA_SUPPLICANT_EAPOL is a bool, not a str
sam [Tue, 21 Feb 2006 17:15:24 +0000 (17:15 +0000)]
NO_WPA_SUPPLICANT_EAPOL is a bool, not a str

18 years agoClear csum_flags after reading data from socket buffer. Otherwise,
ru [Tue, 21 Feb 2006 13:04:39 +0000 (13:04 +0000)]
Clear csum_flags after reading data from socket buffer.  Otherwise,
if ksocket is connected to an interface-type node somewhere later
in the graph (e.g., ng_eiface or ng_iface), the csum_data may be
applied to a wrong packet (if we encapsulate Ethernet or IP).

MFC after: 3 days

18 years agoFor pts, print the pts number, instead of the full name. As it was, we ended
cognet [Tue, 21 Feb 2006 13:02:18 +0000 (13:02 +0000)]
For pts, print the pts number, instead of the full name. As it was, we ended
up always printing "pts".

Submitted by: Michal Mertl <mime at traveller dot cz>

18 years agoIn wall and who, check that the utmp entry isn't stalled, as it is done in w.
cognet [Tue, 21 Feb 2006 13:01:00 +0000 (13:01 +0000)]
In wall and who, check that the utmp entry isn't stalled, as it is done in w.
Apparently with the new pts code stalled entries are printed, when they are
not with the BSD ptys.

Submitted by: Michal Mertl <mime at traveller dot cz>

18 years agoInitialize the port_bst and port_bsh variables.
nyan [Tue, 21 Feb 2006 12:01:39 +0000 (12:01 +0000)]
Initialize the port_bst and port_bsh variables.
The ed driver on pc98 was broken by if_edvar.h rev1.31.

Reported by: Kaho Toshikazu (kaho at elam kais kyoto-u ac jp)
Tested by: Eiji Kato (ekato at a1 mbn or jp)
MFC after: 3 days

18 years agoNew release notes: GNOME 2.12.3, KDE 3.5.1, Perl 5.8.8, X.org 6.9.0.
bmah [Tue, 21 Feb 2006 07:39:11 +0000 (07:39 +0000)]
New release notes:  GNOME 2.12.3, KDE 3.5.1, Perl 5.8.8, X.org 6.9.0.

MFCs noted:  OpenSSH 4.2p1, sendmail 8.13.4.

18 years agoRemove unused variable.
njl [Tue, 21 Feb 2006 03:19:24 +0000 (03:19 +0000)]
Remove unused variable.

Coverity ID: 558
Found by: Coverity Prevent

18 years agoRemove unused variable.
njl [Tue, 21 Feb 2006 03:16:58 +0000 (03:16 +0000)]
Remove unused variable.

Coverity ID: 548
Found by: Coverity Prevent

18 years agoRemove unused variable.
njl [Tue, 21 Feb 2006 03:15:26 +0000 (03:15 +0000)]
Remove unused variable.

Coverity ID: 546
Found by: Coverity Prevent

18 years agoAdd semicolons on the end of some "&merged;" entity usages.
bmah [Mon, 20 Feb 2006 23:06:57 +0000 (23:06 +0000)]
Add semicolons on the end of some "&merged;" entity usages.
Fortunately this was fairly harmless.

Rusty SGML: bmah

18 years agoAttempt to make running the new rc in a jail more automatic by
dougb [Mon, 20 Feb 2006 21:54:30 +0000 (21:54 +0000)]
Attempt to make running the new rc in a jail more automatic by
resetting of early_late_divider to a more appropriate value if
the admin has not modified the default.

Reviewed by: brooks

18 years agoEnable automatic hardware notes generation for myri10ge(4).
brueffer [Mon, 20 Feb 2006 21:46:54 +0000 (21:46 +0000)]
Enable automatic hardware notes generation for myri10ge(4).

18 years agocorrect check for whether wpa is enabled
sam [Mon, 20 Feb 2006 20:49:56 +0000 (20:49 +0000)]
correct check for whether wpa is enabled

MFC after: 3 days

18 years agoUpdate for the 9550 series of controllers.
brueffer [Mon, 20 Feb 2006 19:42:04 +0000 (19:42 +0000)]
Update for the 9550 series of controllers.

Missing hardware entries notices by: nemesis on irc

18 years agotreat setting the optional ie to a zero-length datum to mean "delete
sam [Mon, 20 Feb 2006 17:17:00 +0000 (17:17 +0000)]
treat setting the optional ie to a zero-length datum to mean "delete
the ie"; this helps drivers that (wrongly) check ic_opt_ie against NULL
when making decisions

Reviewed by: luigi, avatar
MFC after: 3 days

18 years agoObey opt_inet6.h in kernel build directory.
ume [Mon, 20 Feb 2006 12:30:32 +0000 (12:30 +0000)]
Obey opt_inet6.h in kernel build directory.

Reported by: Peter Losher <plosher-keyword-freebsd.a36e57__at__plosh.net>
MFC after: 3 days

18 years agoDisable the myri10ge driver on pc98.
nyan [Mon, 20 Feb 2006 12:05:52 +0000 (12:05 +0000)]
Disable the myri10ge driver on pc98.

18 years agoCopy MAC address to fix alignment problems.
ticso [Mon, 20 Feb 2006 10:29:40 +0000 (10:29 +0000)]
Copy MAC address to fix alignment problems.
Tested on alpha (AS600) with RTL8169S
Suggested by: jhb

18 years agoFix scancode translation table
markus [Mon, 20 Feb 2006 10:07:40 +0000 (10:07 +0000)]
Fix scancode translation table

Approved by: emax
MFC after: 3 days

18 years agoRead cycle count before sending signal, this fixes a userland thread
davidxu [Mon, 20 Feb 2006 09:02:40 +0000 (09:02 +0000)]
Read cycle count before sending signal, this fixes a userland thread
suspension bug.

MFC after: 3 days

18 years agoFix grammar.
glebius [Mon, 20 Feb 2006 07:54:16 +0000 (07:54 +0000)]
Fix grammar.

Submitted by: ru

18 years agoHook the myri10ge driver to the i386 and amd64 module builds.
gallatin [Mon, 20 Feb 2006 02:50:09 +0000 (02:50 +0000)]
Hook the myri10ge driver to the i386 and amd64 module builds.

Sponsored by: Myricom Inc.
Reviewed by: scottl

18 years ago${.CURDIR} != pwd and since we use the shell version of if now we don't even
mlaier [Mon, 20 Feb 2006 01:08:33 +0000 (01:08 +0000)]
${.CURDIR} != pwd and since we use the shell version of if now we don't even
need to prepend it.

Found-by: gallatin

18 years ago10GbE mode driver and binary firmware for Myricom's PCI-express NICs.
gallatin [Sun, 19 Feb 2006 22:39:19 +0000 (22:39 +0000)]
10GbE mode driver and binary firmware for Myricom's PCI-express NICs.
More info regarding these nics can be found at http://www.myri.com.

Please note that the files
sys/dev/myri10ge/{mcp_gen_header.h,myri10ge_mcp.h} are internally
shared between all our drivers (solaris, macosx, windows, linux, etc).
I'd like to keep these files unchanged, so I can just import newer
versions of them when the firmware API/ABI changes.  This means I'm
stuck with some of the crazy-long #define names, and possibly
non-style(9) characteristics of these files.

Many thanks to mlaier for doing firmware(9) just as I
needed it, and to scottl for his helpful review.

Reviewed by: scottl, glebius
Sponsored by: Myricom Inc.

18 years agoHer name was Anne Frank, not Ann Frank.
wilko [Sun, 19 Feb 2006 21:46:29 +0000 (21:46 +0000)]
Her name was Anne Frank, not Ann Frank.

MFC after: 1 month

18 years agoFix misinformation in "setconfig" message description.
glebius [Sun, 19 Feb 2006 19:11:54 +0000 (19:11 +0000)]
Fix misinformation in "setconfig" message description.

Noticed by: Brian A. Seklecki <lavalamp spiritual-machines.org>

18 years agoAdd a UQ_OPEN_CLEARSTALL quirk for devices that need a clear-stall
iedowse [Sun, 19 Feb 2006 14:48:02 +0000 (14:48 +0000)]
Add a UQ_OPEN_CLEARSTALL quirk for devices that need a clear-stall
operation when a pipe is opened, and add an entry for the ST Micro
biometric CPU.

Submitted by: Fredrik Lindberg
MFC after: 1 week

18 years ago- Don't bother traversing trap frames in stack_save(). This fixes panics
marius [Sun, 19 Feb 2006 11:54:46 +0000 (11:54 +0000)]
- Don't bother traversing trap frames in stack_save(). This fixes panics
  when option DEBUG_LOCKS is used. Trap frames are determined by checking
  whether the caller was one of the tl0_*() or tl1_*() asm functions via
  a newly added pair of dummy symbols in exception.S which mark the begin
  and end of these functions. The tl_trap_* pair marks those in the special
  .trap section and the tl_text_* in the regular .text section. Because
  of their performance penalty db_search_symbol()/db_symbol_values() and
  linker_ddb_search_symbol()/linker_ddb_symbol_values() aren't used here
  for determining the caller, with db_search_symbol()/db_symbol_values()
  additionally not being reentrant.
- For consistency, change db_backtrace() to also use the new markers for
  determining the tl0_*() and tl1_*() asm functions instead of bcmp()'ing
  the symbol name.
- Use FBSDID in db_trace.c.

PR: 93226
Based on a patch by: Antoine Brodin <antoine.brodin@laposte.net>
Ok'ed by: jhb

18 years agoMake sure abi (and therefore archdep) start before SERVERS as part of
dougb [Sun, 19 Feb 2006 08:18:48 +0000 (08:18 +0000)]
Make sure abi (and therefore archdep) start before SERVERS as part of
the ongoing re-alignment of ordering that is necessary as a result of
including local scripts in the base rcorder. [1]

Accomplish this by removing the BEFORE's, and using REQUIRE instead.
This makes the dependencies more obvious, and less susceptible to turning
circular and/or nonsensical when seemingly innocent changes are made
in one place and not another.

Requested by: delphij [1]

18 years agoEnable NLS catalog of csh(1).
ume [Sun, 19 Feb 2006 06:40:29 +0000 (06:40 +0000)]
Enable NLS catalog of csh(1).
The tcsh 6.14 uses libiconv to convert catalogs to appropriate charset
dynamically.  However, we don't have libiconv in our tree.  So, I made
some hack to load libiconv dynamically.  If libiconv is available, you
can use catalogs for more locales than the locales which catalog is
actually installed.
To use this feature, you need to symlink catalogs to appropriate
locales.  You can do this by installing ports/shells/tcsh_nls.

Reviewed by: arch (no objection)
MFC after: 1 week

18 years agoThis commit was generated by cvs2svn to compensate for changes in r155832,
ume [Sun, 19 Feb 2006 06:28:39 +0000 (06:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r155832,
which included commits to RCS files with non-trunk default branches.

18 years agoImport required files to enable use of libiconv from tcsh-6.14.00.
ume [Sun, 19 Feb 2006 06:28:39 +0000 (06:28 +0000)]
Import required files to enable use of libiconv from tcsh-6.14.00.

Reviewed by: arch (no objection)

18 years agoCorrectly locate the character preceeding the matched string in -w
tjr [Sun, 19 Feb 2006 04:27:39 +0000 (04:27 +0000)]
Correctly locate the character preceeding the matched string in -w
mode when in non-UTF-8 multibyte locales (e.g. EUC, GB2312, etc.).

PR: 91909

18 years agoFor the coda5 module, CODA_COMPAT_5 is mandatory, not optional.
yar [Sun, 19 Feb 2006 02:43:42 +0000 (02:43 +0000)]
For the coda5 module, CODA_COMPAT_5 is mandatory, not optional.
Without CODA_COMPAT_5, it would be equivalent to the plain coda
module.  Therefore just add -DCODA_COMPAT_5 to CFLAGS instead of
fiddling with opt_coda.h.  This is particularly important when
the module is built along with the kernel and CODA_COMPAT_5 isn't
in the kernel conf file (and so not in opt_coda.h either).

MFC after: 3 days

18 years agoAdd a quite late 20051014 entry for the changes in module Makefiles,
yar [Sun, 19 Feb 2006 01:05:57 +0000 (01:05 +0000)]
Add a quite late 20051014 entry for the changes in module Makefiles,
mainly to stick to the conventional MFC procedure.

Requested by: Thomas E. Zander <riggs at rrr dot de>
MFC after: 3 days

18 years agoBack out revs 1.18 and 1.19; they were based on incorrect assumptions.
des [Sun, 19 Feb 2006 00:40:28 +0000 (00:40 +0000)]
Back out revs 1.18 and 1.19; they were based on incorrect assumptions.

Requested by: nate

18 years agoStyle: Remove blank lines before EOF.
yar [Sat, 18 Feb 2006 23:49:28 +0000 (23:49 +0000)]
Style: Remove blank lines before EOF.

Tested by: cvs diff -B

18 years agoWork around the fact that nfsclient SRCS must include
yar [Sat, 18 Feb 2006 23:31:49 +0000 (23:31 +0000)]
Work around the fact that nfsclient SRCS must include
nfs_diskless.c if NFS_ROOT is in effect, e.g., present
in the kernel config file.  Otherwise the built module
won't load due to an undefined reference to nfs_setup_diskless.

MFC after: 3 days

18 years agoRemove empty lines at the end
ache [Sat, 18 Feb 2006 22:08:28 +0000 (22:08 +0000)]
Remove empty lines at the end

18 years agoThe sysctls kern.ipc.[max_linkhdr|max_protohdr|max_hdr|max_datalen]
andre [Sat, 18 Feb 2006 17:16:18 +0000 (17:16 +0000)]
The sysctls kern.ipc.[max_linkhdr|max_protohdr|max_hdr|max_datalen]
can't be changed from userland.  Make them read-only and provide
descriptions.

kern.ipc.max_datalen must never be less than one byte.  Enforce this
with a panic in net_init_domain().

Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoRemove unneeded includes and provide more accurate description
andre [Sat, 18 Feb 2006 17:05:00 +0000 (17:05 +0000)]
Remove unneeded includes and provide more accurate description
to others.

Submitted by: garys
PR: kern/86437

18 years agoA file can also be a link, so check not only for a file, but also for a link
netchild [Sat, 18 Feb 2006 16:58:21 +0000 (16:58 +0000)]
A file can also be a link, so check not only for a file, but also for a link
in the delete-old and check-old targets.

We don't install a lib (libXY.so.Z) as a link, but an user may have created
something like this. This is dangerous if this link points to a different
version of the lib. So check for a link also in the *-lib targets (an
annoyed user which absolutely wants this redirection of a lib should use
libmap.conf instead of a link).

A directory can also be a link, but in this case just echo a message to
remove it by hand.

18 years agoAdd missing TH_PUSH to the TH_FLAGS enumeration.
andre [Sat, 18 Feb 2006 16:50:08 +0000 (16:50 +0000)]
Add missing TH_PUSH to the TH_FLAGS enumeration.

Submitted by: Andre Albsmeier <Andre.Albsmeier-at-siemens.com>
PR: kern/85203

18 years agoXref chkgrp(8).
brueffer [Sat, 18 Feb 2006 16:48:56 +0000 (16:48 +0000)]
Xref chkgrp(8).

PR: 93529
Submitted by: Daniel Gerzo <danger@rulez.sk>
MFC after: 3 days

18 years agoBuild audit command line tools only if NO_AUDIT isn't defined.
rwatson [Sat, 18 Feb 2006 16:40:31 +0000 (16:40 +0000)]
Build audit command line tools only if NO_AUDIT isn't defined.

Submitted by: brueffer
Obtained from: TrustedBSD Project

18 years agoAdd system call auditing support for sparc64.
rwatson [Sat, 18 Feb 2006 16:36:56 +0000 (16:36 +0000)]
Add system call auditing support for sparc64.

Submitted by: brueffer
Obtained from: TrustedBSD Project

18 years ago - add libxpg4
netchild [Sat, 18 Feb 2006 16:35:57 +0000 (16:35 +0000)]
 - add libxpg4
 - correct typo in variable name
 - remove etc/security (file):
   While this doesn't affect the new directory, we should remove it to keep
   the list "clean".

18 years agoImproved description for packet zone statistics.
andre [Sat, 18 Feb 2006 16:09:38 +0000 (16:09 +0000)]
Improved description for packet zone statistics.

Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoPrint statistics on mbuf+clusters in packet zone.
andre [Sat, 18 Feb 2006 16:03:07 +0000 (16:03 +0000)]
Print statistics on mbuf+clusters in packet zone.

Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoLook for cpu/ichss instead of cpu/p4tcc, which is i386/amd64-only.
des [Sat, 18 Feb 2006 16:01:07 +0000 (16:01 +0000)]
Look for cpu/ichss instead of cpu/p4tcc, which is i386/amd64-only.

18 years agoPrint statistics on jumbo mbuf clusters.
andre [Sat, 18 Feb 2006 15:55:19 +0000 (15:55 +0000)]
Print statistics on jumbo mbuf clusters.

Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days

18 years agoUse kld(3) for loading geom_md.ko.
pjd [Sat, 18 Feb 2006 11:40:24 +0000 (11:40 +0000)]
Use kld(3) for loading geom_md.ko.

18 years agoLoad cpufreq at startup if it isn't there already.
des [Sat, 18 Feb 2006 11:35:12 +0000 (11:35 +0000)]
Load cpufreq at startup if it isn't there already.

18 years agoReplace the guts of usbmodule() with calls to the new kld(3) functions.
des [Sat, 18 Feb 2006 11:26:36 +0000 (11:26 +0000)]
Replace the guts of usbmodule() with calls to the new kld(3) functions.

18 years agoAdd utility functions for checking if a given kernel module is loaded,
des [Sat, 18 Feb 2006 11:25:28 +0000 (11:25 +0000)]
Add utility functions for checking if a given kernel module is loaded,
and loading it.

18 years agoInform when label disappears.
pjd [Sat, 18 Feb 2006 11:24:00 +0000 (11:24 +0000)]
Inform when label disappears.

MFC after: 3 days

18 years agoAllow to use g_slice_orphan() from outside.
pjd [Sat, 18 Feb 2006 11:21:17 +0000 (11:21 +0000)]
Allow to use g_slice_orphan() from outside.

MFC after: 3 days

18 years ago- Do not depend on fact that file system covers entire provider.
pjd [Sat, 18 Feb 2006 10:59:47 +0000 (10:59 +0000)]
- Do not depend on fact that file system covers entire provider.
  It won't work for file systems created with -s option.
  Use better file system verfication.
- Add myself to the copyright.

MFC after: 3 days

18 years ago- Fix severe crackling after long running and abusive module load / unload by
ariff [Sat, 18 Feb 2006 10:24:48 +0000 (10:24 +0000)]
- Fix severe crackling after long running and abusive module load / unload by
  forcing DMA alignment to default buffer size.
- Make sure DMA pointer properly aligned to avoid being truncated by caller
  which causing severe underruns and random popping (especially in 32bit
  playback / recording).
- Add AC97 inverted external amplifier quirk for Maxselect x710s
  - http://maxselect.ru/

MFC after: 1 week

18 years agoFix random popping while playing in 32bit by increasing default buffer size
ariff [Sat, 18 Feb 2006 10:23:37 +0000 (10:23 +0000)]
Fix random popping while playing in 32bit by increasing default buffer size
from 4096 to 8192.

MFC after: 1 week

18 years agoThis function returns nothing.
pjd [Sat, 18 Feb 2006 03:04:26 +0000 (03:04 +0000)]
This function returns nothing.

18 years agoIf provider's sector size prevents reading SBLOCKSIZE bytes return
pjd [Sat, 18 Feb 2006 03:00:49 +0000 (03:00 +0000)]
If provider's sector size prevents reading SBLOCKSIZE bytes return
immediatelly.

18 years agoUse shell's version of if to check if the firmware really exists in the
mlaier [Sat, 18 Feb 2006 01:38:44 +0000 (01:38 +0000)]
Use shell's version of if to check if the firmware really exists in the
current directory to allow user rules to create the firmware (e.g. from a
uuencoded blob).  make's version of if is evaluated too early to catch this.

Found-by: gallatin

18 years agouse s/w bmiss facility
sam [Sat, 18 Feb 2006 01:04:08 +0000 (01:04 +0000)]
use s/w bmiss facility

MFC after: 1 week

18 years agostop device so we don't panic on card eject
sam [Sat, 18 Feb 2006 01:03:09 +0000 (01:03 +0000)]
stop device so we don't panic on card eject

MFC after: 1 week

18 years agoThese files apply to all the atmel parts that freebsd is going to run on,
imp [Fri, 17 Feb 2006 22:33:13 +0000 (22:33 +0000)]
These files apply to all the atmel parts that freebsd is going to run on,
so name them more generically.  If we do support the MMU-less ARM7 parts,
then we'll need to, at that time, expand the files we have.

18 years agoThis file was obsolete when committed. Catchup and delete it.
imp [Fri, 17 Feb 2006 22:23:36 +0000 (22:23 +0000)]
This file was obsolete when committed.  Catchup and delete it.

18 years agoExpand scope of marker to reduce the number of page queue scan restarts.
tegge [Fri, 17 Feb 2006 21:02:39 +0000 (21:02 +0000)]
Expand scope of marker to reduce the number of page queue scan restarts.

18 years agoBelatedly bump the date for the fan control changes in rev. 1.9.
brueffer [Fri, 17 Feb 2006 20:29:32 +0000 (20:29 +0000)]
Belatedly bump the date for the fan control changes in rev. 1.9.

18 years agoRe-work the bus attachment somewhat. Make the rids unique for
ambrisko [Fri, 17 Feb 2006 18:46:18 +0000 (18:46 +0000)]
Re-work the bus attachment somewhat.  Make the rids unique for
the SMI/TCO address space.  Switch the bus space I/O to the
one specific for either the SMI or TCO space.  Re-calibrate
the tick.  Add some more device id's, 82801FBR submitted by des.
This makes it work on the platforms I've tested with.

Go ahead by: des

18 years agoCheck return value from nonblocking call to vn_start_write().
tegge [Fri, 17 Feb 2006 18:22:19 +0000 (18:22 +0000)]
Check return value from nonblocking call to vn_start_write().

18 years ago1) Ignore link events for MII/GMII cards if MI auto-polling disabled. This
oleg [Fri, 17 Feb 2006 14:33:35 +0000 (14:33 +0000)]
1) Ignore link events for MII/GMII cards if MI auto-polling disabled. This
   should fix strange link state behaviour reported for bcm5721 & bcm5704c
2) Clear bge_link flag in bge_stop()
3) Force link state check after bge_ifmedia_upd(). Otherwise we can miss link
   event if PHY changes it's state fast enough.

Tested by: phk (bcm5704c)
Approved by: glebius (mentor)
MFC after: 1 week

18 years agoJMicron RAID metadata support.
brueffer [Fri, 17 Feb 2006 14:26:49 +0000 (14:26 +0000)]
JMicron RAID metadata support.

18 years agoReplace the 4k fixed sized jumbo mbuf clusters with PAGE_SIZE sized
andre [Fri, 17 Feb 2006 14:14:15 +0000 (14:14 +0000)]
Replace the 4k fixed sized jumbo mbuf clusters with PAGE_SIZE sized
jumbo mbuf clusters.  To make the variable size clear they are named
MJUMPAGESIZE.

Having jumbo clusters with the native PAGE_SIZE is more useful than
a fixed 4k size according the device driver writers using this API.

The 9k and 16k jumbo mbuf clusters remain unchanged.

Requested by: glebius, gallatin
Sponsored by: TCP/IP Optimization Fundraise 2005
MFC after: 3 days