]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
14 years agoRevert r200892, 200893 and 200894. There's companion changes
Warner Losh [Wed, 6 Jan 2010 12:15:10 +0000 (12:15 +0000)]
Revert r200892, 200893 and 200894.  There's companion changes
elsewhere that aren't quite ready, and these break the world in the
mean time.

14 years agoRemove all CFE-specific code from locore.S. The CFE entrypoint initialization
Neel Natu [Wed, 6 Jan 2010 06:42:08 +0000 (06:42 +0000)]
Remove all CFE-specific code from locore.S. The CFE entrypoint initialization
is now done in platform-specific code.

Approved by: imp (mentor)

14 years agoMerge from head at r201628.
Warner Losh [Wed, 6 Jan 2010 05:58:07 +0000 (05:58 +0000)]
Merge from head at r201628.

# This hasn't been tested, and there are at least three bad commits
# that need to be backed out before the branch will be stable again.

14 years agoFix build: getopt() returns int so use an integer to get the value.
Xin LI [Wed, 6 Jan 2010 00:20:37 +0000 (00:20 +0000)]
Fix build: getopt() returns int so use an integer to get the value.

14 years ago- Move potentially offensive quotes I committed recently to fortunes-o.
Stanislav Sedov [Tue, 5 Jan 2010 23:35:43 +0000 (23:35 +0000)]
- Move potentially offensive quotes I committed recently to fortunes-o.
- Reference date.
- Be more clear on context.

Suggested by: dougb

14 years agoAdd net80211 media status reporting.
Rui Paulo [Tue, 5 Jan 2010 22:59:59 +0000 (22:59 +0000)]
Add net80211 media status reporting.

PR: 142197
Submitted by: Paul <onemda at gmail.com>

14 years agoBe able to specify a certain date and/or time for which to calculate
Edwin Groothuis [Tue, 5 Jan 2010 21:14:48 +0000 (21:14 +0000)]
Be able to specify a certain date and/or time for which to calculate
the phase of the moon.
While not worlds best improvements, it will help calendar(1) later on.

14 years ago1) Mark usage as dead2.
David Malone [Tue, 5 Jan 2010 21:09:59 +0000 (21:09 +0000)]
1) Mark usage as dead2.
2) Deregister.
3) New style function definitions.

Some WARNS still remain here - some printf format warning on some
arches and the compiler can't see that a variable should always be
initialised.

14 years agoShuffle parens to avoid type-punning warning.
David Malone [Tue, 5 Jan 2010 21:00:23 +0000 (21:00 +0000)]
Shuffle parens to avoid type-punning warning.

14 years ago1) Make usage() as dead2 - it helps the compiler know that some vars
David Malone [Tue, 5 Jan 2010 20:53:55 +0000 (20:53 +0000)]
1) Make usage() as dead2 - it helps the compiler know that some vars
are not used uninitialised.
2) Fix some constness problems.
3) Avoid a signedness problem by casting to size_t. If bn != stuff,
than stuff-1-bn should be > 0.

14 years agoMove the PCI-specific logic of removing a cardbus device into a
John Baldwin [Tue, 5 Jan 2010 20:42:25 +0000 (20:42 +0000)]
Move the PCI-specific logic of removing a cardbus device into a
pci_delete_child() function called by the cardbus driver.  The new function
uses resource_list_unreserve() to release the BARs decoded by the device
being removed.

Reviewed by: imp
Tested by: brooks

14 years agoNew style function definitions.
David Malone [Tue, 5 Jan 2010 20:40:40 +0000 (20:40 +0000)]
New style function definitions.
Fix constness problem - don't know that pstatus won't change a string with no
whitespace.

14 years agoChange a char that is used as an index into an array into an unisgned char.
David Malone [Tue, 5 Jan 2010 20:32:08 +0000 (20:32 +0000)]
Change a char that is used as an index into an array into an unisgned char.
Add a missing new style function definition.

14 years agoUpdate acpi_ibm syctl nodes on resume.
Mitsuru IWASAKI [Tue, 5 Jan 2010 20:29:30 +0000 (20:29 +0000)]
Update acpi_ibm syctl nodes on resume.

This should fix some Thinkpad specific problems such as
connecting to a headphone jack is not functional on X41.

Reviewed by: takawata
MFC after: 1 week

14 years agoDo not rely on behaviour undefined by ANSI C, use thunks to adapt
Konstantin Belousov [Tue, 5 Jan 2010 20:20:31 +0000 (20:20 +0000)]
Do not rely on behaviour undefined by ANSI C, use thunks to adapt
alphasort-like interface to the comparision function required by
qsort() and qsort_r().

For opendir() thunk and alphasort(), comment on why we deviated from
POSIX by using strcmp() instead of strcoll().

Requested and reviewed by: bde
MFC after: 2 weeks

14 years agoFix a double free().
Christian Brueffer [Tue, 5 Jan 2010 20:18:41 +0000 (20:18 +0000)]
Fix a double free().

PR: 142339
Submitted by: Henning Petersen <henning.petersen@t-online.de>
MFC after: 2 weeks

14 years agoMove scandir(3) and alphasort(3) into XSI namespace.
Konstantin Belousov [Tue, 5 Jan 2010 20:17:13 +0000 (20:17 +0000)]
Move scandir(3) and alphasort(3) into XSI namespace.

Noted and reviewed by: bde
MFC after: 2 weeks

14 years agoRemove more duplicates
Doug Barton [Tue, 5 Jan 2010 19:40:32 +0000 (19:40 +0000)]
Remove more duplicates

Minor reformatting on a few

14 years agoFor completeness, add -s argument, manually specifying array block size.
Alexander Motin [Tue, 5 Jan 2010 13:25:12 +0000 (13:25 +0000)]
For completeness, add -s argument, manually specifying array block size.

14 years agoAdd test code for POSIX semaphore implementation.
David Xu [Tue, 5 Jan 2010 12:34:13 +0000 (12:34 +0000)]
Add test code for POSIX semaphore implementation.

14 years agothis file does not require ip_dummynet.h
Luigi Rizzo [Tue, 5 Jan 2010 11:00:31 +0000 (11:00 +0000)]
this file does not require ip_dummynet.h

14 years agoMove wakeup() out of mutex to reduce contention.
Alexander Motin [Tue, 5 Jan 2010 10:52:21 +0000 (10:52 +0000)]
Move wakeup() out of mutex to reduce contention.

14 years agoMove wakeup() out of mutex to reduce contention.
Alexander Motin [Tue, 5 Jan 2010 10:30:56 +0000 (10:30 +0000)]
Move wakeup() out of mutex to reduce contention.

14 years agoremove file thr_sem_new.c.
David Xu [Tue, 5 Jan 2010 07:50:31 +0000 (07:50 +0000)]
remove file thr_sem_new.c.

14 years agoThis change increases the size of the kernel stack for thread0 from
Neel Natu [Tue, 5 Jan 2010 06:58:54 +0000 (06:58 +0000)]
This change increases the size of the kernel stack for thread0 from
PAGE_SIZE to (2 * PAGE_SIZE). It depends on the memory allocated by
pmap_steal_memory() being aligned to a PAGE_SIZE boundary.

Approved by: imp (mentor)

14 years agoMore cleanup, remove _libc prefix because libthr no longer has stubs
David Xu [Tue, 5 Jan 2010 06:40:27 +0000 (06:40 +0000)]
More cleanup, remove _libc prefix because libthr no longer has stubs
referencing them.

14 years agoRemove extra new semaphore stubs, because libc already has them, and
David Xu [Tue, 5 Jan 2010 06:21:29 +0000 (06:21 +0000)]
Remove extra new semaphore stubs, because libc already has them, and
ld can find the newest version which is default.

Poked by: kan@

14 years agoforgot to remove SYM_FBP10. ;-)
David Xu [Tue, 5 Jan 2010 05:47:18 +0000 (05:47 +0000)]
forgot to remove  SYM_FBP10. ;-)

14 years agoRemove unused macros.
David Xu [Tue, 5 Jan 2010 05:44:52 +0000 (05:44 +0000)]
Remove unused macros.

14 years agoDon't check has_waiters twice, inline some small functions.
David Xu [Tue, 5 Jan 2010 03:39:31 +0000 (03:39 +0000)]
Don't check has_waiters twice, inline some small functions.
performance result on my machine:
mutex     Elapsed: 902115 us; per iteration: 90 ns.
semaphore Elapsed: 958780 us; per iteration: 95 ns.

14 years agoUse umtx to implement process sharable semaphore, to make this work,
David Xu [Tue, 5 Jan 2010 02:37:59 +0000 (02:37 +0000)]
Use umtx to implement process sharable semaphore, to make this work,
now type sema_t is a structure which can be put in a shared memory area,
and multiple processes can operate it concurrently.
User can either use mmap(MAP_SHARED) + sem_init(pshared=1) or use sem_open()
to initialize a shared semaphore.
Named semaphore uses file system and is located in /tmp directory, and its
file name is prefixed with 'SEMD', so now it is chroot or jail friendly.
In simplist cases, both for named and un-named semaphore, userland code
does not have to enter kernel to reduce/increase semaphore's count.
The semaphore is designed to be crash-safe, it means even if an application
is crashed in the middle of operating semaphore, the semaphore state is
still safely recovered by later use, there is no waiter counter maintained
by userland code.
The main semaphore code is in libc and libthr only has some necessary stubs,
this makes it possible that a non-threaded application can use semaphore
without linking to thread library.
Old semaphore implementation is kept libc to maintain binary compatibility.
The kernel ksem API is no longer used in the new implemenation.

Discussed on: threads@

14 years agoSlightly optimize XOR calculation.
Alexander Motin [Tue, 5 Jan 2010 02:06:05 +0000 (02:06 +0000)]
Slightly optimize XOR calculation.

14 years agoAn existing incomplete ARP entry would expire a subsequent
Qing Li [Tue, 5 Jan 2010 00:35:46 +0000 (00:35 +0000)]
An existing incomplete ARP entry would expire a subsequent
statically configured entry of the same host. This bug was
due to the expiration timer was not cancelled when installing
the static entry. Since there exist a potential race condition
with respect to timer cancellation, simply check for the
LLE_STATIC bit inside the expiration function instead of
cancelling the active timer.

MFC after: 5 days

14 years agoThe IFA_RTSELF address flag marks a loopback route has been installed
Qing Li [Mon, 4 Jan 2010 23:39:53 +0000 (23:39 +0000)]
The IFA_RTSELF address flag marks a loopback route has been installed
for the interface address. This marker is necessary to properly support
PPP types of links where multiple links can have the same local end
IP address. The IFA_RTSELF flag bit maps to the RTF_HOST value, which
was combined into the route flag bits during prefix installation in
IPv6. This inclusion causing the prefix route to be unusable. This
patch fixes this bug by excluding the IFA_RTSELF flag during route
installation.

MFC after: 5 days

14 years agoAdd a missing word.
Christian Brueffer [Mon, 4 Jan 2010 22:23:09 +0000 (22:23 +0000)]
Add a missing word.

PR: 140989
Submitted by: Lachlan Kang
MFC after: 1 week

14 years agoCatch up with the VFS_VPTOFH(9) -> VOP_VPTOFH(9) repocopy that happened
Christian Brueffer [Mon, 4 Jan 2010 22:22:00 +0000 (22:22 +0000)]
Catch up with the VFS_VPTOFH(9) -> VOP_VPTOFH(9) repocopy that happened
almost three years ago in r166794.

PR: 140989
Submitted by: Lachlan Kang
MFC after: 1 week

14 years agoRevert 200594. This file isn't intended for these sorts of things.
Warner Losh [Mon, 4 Jan 2010 21:30:04 +0000 (21:30 +0000)]
Revert 200594.  This file isn't intended for these sorts of things.

14 years agoMake TIOCSTI work again.
Ed Schouten [Mon, 4 Jan 2010 20:59:52 +0000 (20:59 +0000)]
Make TIOCSTI work again.

It looks like I didn't implement this when I imported MPSAFE TTY.
Applications like mail(1) still use this. I think it's conceptually bad.

Tested by: Pete French <petefrench ticketswitch com>
MFC after: 2 weeks

14 years agoStyle(9) pass.
Warner Losh [Mon, 4 Jan 2010 20:34:15 +0000 (20:34 +0000)]
Style(9) pass.

14 years agoVarious cleanup done in ipfw3-head branch including:
Luigi Rizzo [Mon, 4 Jan 2010 19:01:22 +0000 (19:01 +0000)]
Various cleanup done in ipfw3-head branch including:
- use a uniform mtag format for all packets that exit and re-enter
  the firewall in the middle of a rulechain. On reentry, all tags
  containing reinject info are renamed to MTAG_IPFW_RULE so the
  processing is simpler.

- make ipfw and dummynet use ip_len and ip_off in network format
  everywhere. Conversion is done only once instead of tracking
  the format in every place.

- use a macro FREE_PKT to dispose of mbufs. This eases portability.

On passing i also removed a few typos, staticise or localise variables,
remove useless declarations and other minor things.

Overall the code shrinks a bit and is hopefully more readable.

I have tested functionality for all but ng_ipfw and if_bridge/if_ethersubr.
For ng_ipfw i am actually waiting for feedback from glebius@ because
we might have some small changes to make.
For if_bridge and if_ethersubr feedback would be welcome
(there are still some redundant parts in these two modules that
I would like to remove, but first i need to check functionality).

14 years agoAdd ability to search up the directory hierarchy for the system directory.
David E. O'Brien [Mon, 4 Jan 2010 18:57:22 +0000 (18:57 +0000)]
Add ability to search up the directory hierarchy for the system directory.
Do by specifying ".../" with '-m' or MAKESYSPATH (new) environment variable.

Reviewed by: <sjg@NetBSD.org>
Obtained from: NetBSD (+ embellishment by me, sent back to NetBSD)

14 years agoThis should read #if __BSD_VISIBLE instead of #ifdef __BSD_VISIBLE.
Ed Schouten [Mon, 4 Jan 2010 18:46:54 +0000 (18:46 +0000)]
This should read #if __BSD_VISIBLE instead of #ifdef __BSD_VISIBLE.

14 years agoCorrect usage of parenthesis.
Michael Tuexen [Mon, 4 Jan 2010 18:25:38 +0000 (18:25 +0000)]
Correct usage of parenthesis.

PR: kern/142066
Approved by: rrs (mentor)
Obtained from: Henning Petersen, Bruce Cran.
MFC after: 2 weeks

14 years agoExpose the upper 256 ptys in the default devfs rules. I should have updated
John Baldwin [Mon, 4 Jan 2010 18:21:27 +0000 (18:21 +0000)]
Expose the upper 256 ptys in the default devfs rules.  I should have updated
this when expanding the old pty(4) driver to use 512 ptys by default.  This
is more important for 7.x.

MFC after: 1 week

14 years agoUnbreak bootparamd when setting WITHOUT_NIS.
Ed Schouten [Mon, 4 Jan 2010 18:08:16 +0000 (18:08 +0000)]
Unbreak bootparamd when setting WITHOUT_NIS.

Submitted by: bf1783 googlemail com

14 years agoTest index value is within the range before using it to reference
Xin LI [Mon, 4 Jan 2010 18:04:36 +0000 (18:04 +0000)]
Test index value is within the range before using it to reference
array member.

PR: bin/141838
Submitted by: Henning Petersen <henning.petersen@t-online.de>
MFC after: 2 weeks

14 years agoNow that I'm free from mentorship, it's only fair to document who
Benedict Reuschling [Mon, 4 Jan 2010 15:50:41 +0000 (15:50 +0000)]
Now that I'm free from mentorship, it's only fair to document who
my mentor was.

14 years agoBump __FreeBSD_version for scandir(3) and alphasort(3) prototype changes.
Konstantin Belousov [Mon, 4 Jan 2010 15:41:16 +0000 (15:41 +0000)]
Bump __FreeBSD_version for scandir(3) and alphasort(3) prototype changes.

14 years agoModernize scandir(3) and alphasort(3) interfaces according to the IEEE
Konstantin Belousov [Mon, 4 Jan 2010 15:40:17 +0000 (15:40 +0000)]
Modernize scandir(3) and alphasort(3) interfaces according to the IEEE
Std 1003.1-2008. Both Linux and Solaris conforms to the new definitions,
so we better follow too (older glibc used old BSDish alphasort prototype
and corresponding type of the comparision function for scandir). While
there, change the definitions of the functions to ANSI C and fix several
style issues nearby.

Remove requirement for "sys/types.h" include for functions from manpage.

POSIX also requires that alphasort(3) sorts as if strcoll(3) was used,
but leave the strcmp(3) call in the function for now.

Adapt in-tree callers of scandir(3) to new declaration. The fact that
select_sections() from catman(1) could modify supplied struct dirent is
a bug.

PR: standards/142255
MFC after: 2 weeks

14 years agoPG_NOSYNC is called VPO_NOSYNC for long time.
Konstantin Belousov [Mon, 4 Jan 2010 14:58:41 +0000 (14:58 +0000)]
PG_NOSYNC is called VPO_NOSYNC for long time.

MFC after: 3 days

14 years agoFix comments.
Edward Tomasz Napierala [Mon, 4 Jan 2010 12:39:42 +0000 (12:39 +0000)]
Fix comments.

14 years agoFix formatting.
David E. O'Brien [Mon, 4 Jan 2010 11:11:37 +0000 (11:11 +0000)]
Fix formatting.

14 years agoThere are actually 129 tests here.
David E. O'Brien [Mon, 4 Jan 2010 11:00:12 +0000 (11:00 +0000)]
There are actually 129 tests here.

14 years agoCatch up with UNIX98-style PTY's.
David E. O'Brien [Mon, 4 Jan 2010 10:59:14 +0000 (10:59 +0000)]
Catch up with UNIX98-style PTY's.

14 years ago'prove -r' isn't as smart as grog thought.
David E. O'Brien [Mon, 4 Jan 2010 10:58:08 +0000 (10:58 +0000)]
'prove -r' isn't as smart as grog thought.

14 years agoCommitted the wrong version in r201484. This time really fix the "-t"
David E. O'Brien [Mon, 4 Jan 2010 10:50:17 +0000 (10:50 +0000)]
Committed the wrong version in r201484.  This time really fix the "-t"
functionality.  Per the regression tests (pgrep-t.t & pkill-t.t), "-t"
should accept "v1", which means a plain number should be accepted for
UNIX98-style PTY's.

14 years agoFix the "-t" functionality. Per the regression tests (pgrep-t.t & pkill-t.t),
David E. O'Brien [Mon, 4 Jan 2010 10:37:07 +0000 (10:37 +0000)]
Fix the "-t" functionality.  Per the regression tests (pgrep-t.t & pkill-t.t),
"-t" should accept "v1", which means a plain number should be accepted for
UNIX98-style PTY's.

14 years agoYou need to be root to run this test.
David E. O'Brien [Mon, 4 Jan 2010 09:59:18 +0000 (09:59 +0000)]
You need to be root to run this test.

14 years agoAlso cleanup after the test run.
David E. O'Brien [Mon, 4 Jan 2010 09:57:00 +0000 (09:57 +0000)]
Also cleanup after the test run.

14 years agoDon't process 'lastcomm' if "AUTOMATED" is defined.
David E. O'Brien [Mon, 4 Jan 2010 09:54:24 +0000 (09:54 +0000)]
Don't process 'lastcomm' if "AUTOMATED" is defined.
The tests for it require manual setup.

14 years agoIt is expected these are chmod +x.
David E. O'Brien [Mon, 4 Jan 2010 09:50:30 +0000 (09:50 +0000)]
It is expected these are chmod +x.

14 years agoRegression test for r201456.
David E. O'Brien [Mon, 4 Jan 2010 09:49:23 +0000 (09:49 +0000)]
Regression test for r201456.

14 years ago- Add two particulary interesting quotes from the past decade.
Stanislav Sedov [Mon, 4 Jan 2010 09:16:35 +0000 (09:16 +0000)]
- Add two particulary interesting quotes from the past decade.

14 years agoAdd user-level semaphore synchronous type, this change allows multiple
David Xu [Mon, 4 Jan 2010 05:27:49 +0000 (05:27 +0000)]
Add user-level semaphore synchronous type, this change allows multiple
processes to share semaphore by using shared memory area, in simplest case,
only one atomic operation is needed in userland, waiter flag is maintained by
kernel and userland only checks the flag, if the flag is set, user code enters
kernel and does a wakeup() call.
Move type definitions into file _umtx.h to minimize compiling time.
Also type names need to be prefixed with underline character, this would reduce
name conflict (still in progress).

14 years agoRemove CNS11XXNAS.hints.
Rui Paulo [Mon, 4 Jan 2010 03:40:46 +0000 (03:40 +0000)]
Remove CNS11XXNAS.hints.

14 years agoAdd support for Cavium Econa CNS11XX ARM boards. These boards were
Rui Paulo [Mon, 4 Jan 2010 03:35:45 +0000 (03:35 +0000)]
Add support for Cavium Econa CNS11XX ARM boards. These boards were
previously know by StarSemi STR9104.

Tested by the submitter on an Emprex NSD-100 board.

Submitted by: Yohanes Nugroho <yohanes at gmail.com>
Reviewed by: freebsd-arm, stas
Obtained from: //depot/projects/str91xx/...

14 years agoReport lines that ought to contain a ':' operator but start with a '.'
David E. O'Brien [Mon, 4 Jan 2010 00:49:04 +0000 (00:49 +0000)]
Report lines that ought to contain a ':' operator but start with a '.'
as "Unknown directive" sinze they are more likely to be .elseif (etc).
(NetBSD PR 37222).

Obtained from: NetBSD

14 years agoAdd check for subversion "original" marker.
David E. O'Brien [Mon, 4 Jan 2010 00:32:22 +0000 (00:32 +0000)]
Add check for subversion "original" marker.

14 years agocardbus -> CardBus
Warner Losh [Sun, 3 Jan 2010 23:31:58 +0000 (23:31 +0000)]
cardbus -> CardBus

14 years agocardbus -> CardBus
Warner Losh [Sun, 3 Jan 2010 23:29:49 +0000 (23:29 +0000)]
cardbus -> CardBus

14 years agocardbus -> CardBus
Warner Losh [Sun, 3 Jan 2010 23:28:32 +0000 (23:28 +0000)]
cardbus -> CardBus

14 years agoCardBus instead of cardbus.
Warner Losh [Sun, 3 Jan 2010 23:26:29 +0000 (23:26 +0000)]
CardBus instead of cardbus.

14 years agoFix CardBus spelling.
Warner Losh [Sun, 3 Jan 2010 23:15:14 +0000 (23:15 +0000)]
Fix CardBus spelling.

MFC after: 1 month

14 years agoComplete ntptrace(8) removal:
Antoine Brodin [Sun, 3 Jan 2010 22:29:06 +0000 (22:29 +0000)]
Complete ntptrace(8) removal:
- do not install man page
- update ObsoleteFiles.inc

Approved by: roberto@

14 years agoFix regression introduced in r198318. BCM5754/BCM5754M uses the
Pyun YongHyeon [Sun, 3 Jan 2010 21:49:24 +0000 (21:49 +0000)]
Fix regression introduced in r198318. BCM5754/BCM5754M uses the
same ASIC ID of BCM5758 such that r198318 incorecctly enabled TSO
on BCM5754.BCM5754M controllers. BCM5754/BCM5754M needs a special
firmware to enable TSO and bge(4) does not support firmware based
TSO.

Reported by: ed
Tested by: ed

14 years agoForced commit, to provide correct commit message for r201440:
Gavin Atkinson [Sun, 3 Jan 2010 21:27:10 +0000 (21:27 +0000)]
Forced commit, to provide correct commit message for r201440:

Don't complain when we encounter the "cache" source, it's valid.  Also fix
the error message to include a line feed and not include a stray comma.

PR: bin/121671
Submitted by: Artis Caune  artis.caune gmail.com
Approved by: ed (mentor)
MFC after: 2 weeks

While here, change "> /dev/stderr" for more usual ">&2"

Submitted by: jilles

14 years agoThe size of credential messages is limited by CMGROUP_MAX rather than
Brooks Davis [Sun, 3 Jan 2010 20:45:49 +0000 (20:45 +0000)]
The size of credential messages is limited by CMGROUP_MAX rather than
NGROUPS.

MFC after: 1 week

14 years agoAdd vlan(4) to all GENERIC kernels.
Brooks Davis [Sun, 3 Jan 2010 20:40:54 +0000 (20:40 +0000)]
Add vlan(4) to all GENERIC kernels.

MFC after: 1 week

14 years agoThe test for "same client" for the experimental nfs server over NFSv4
Rick Macklem [Sun, 3 Jan 2010 20:08:10 +0000 (20:08 +0000)]
The test for "same client" for the experimental nfs server over NFSv4
was broken w.r.t. byte range lock conflicts when it was the same client
and the request used the open_to_lock_owner4 case, since lckstp->ls_clp
was not set. This patch fixes it by using "clp" instead of "lckstp->ls_clp".

MFC after: 2 weeks

14 years agoThe default hash table size is 257 not 255. Reword the rest of the line
Gavin Atkinson [Sun, 3 Jan 2010 19:28:51 +0000 (19:28 +0000)]
The default hash table size is 257 not 255.  Reword the rest of the line
slightly while here.

PR: bin/121671
Submitted by: Artis Caune  artis.caune gmail.com
Approved by: ed (mentor)

14 years agoThe default hash table size is 257 not 255. Reword the rest of the line
Gavin Atkinson [Sun, 3 Jan 2010 19:14:22 +0000 (19:14 +0000)]
The default hash table size is 257 not 255.  Reword the rest of the line
slightly while here.

PR: bin/121671
Submitted by: Artis Caune  artis.caune gmail.com
Approved by: ed (mentor)

14 years agoFix three related problems in the experimental nfs client when
Rick Macklem [Sun, 3 Jan 2010 18:27:10 +0000 (18:27 +0000)]
Fix three related problems in the experimental nfs client when
checking for conflicts w.r.t. byte range locks for NFSv4.
1 - Return 0 instead of EACCES when a conflict is found, for F_GETLK.
2 - Check for "same file" when checking for a conflict.
3 - Don't check for a conflict for the F_UNLCK case.

14 years agoMake mac_lomac(4) able to interpret NFSv4 access bits.
Edward Tomasz Napierala [Sun, 3 Jan 2010 17:19:14 +0000 (17:19 +0000)]
Make mac_lomac(4) able to interpret NFSv4 access bits.

Reviewed by: rwatson

14 years agoLower WARNS for sendmail.
Ed Schouten [Sun, 3 Jan 2010 15:32:25 +0000 (15:32 +0000)]
Lower WARNS for sendmail.

Even though it builds with WARNS=2, some users link sendmail from the
base system against SASL. This doesn't build in this case.

Reported by: Andrzej Tobola <ato iem pw edu pl>

14 years agoAdd manual page for gcache(8).
Edward Tomasz Napierala [Sun, 3 Jan 2010 15:20:52 +0000 (15:20 +0000)]
Add manual page for gcache(8).

14 years agosh: Send the "not found" message for builtin <cmd> to redirected fd 2.
Jilles Tjoelker [Sun, 3 Jan 2010 15:01:38 +0000 (15:01 +0000)]
sh: Send the "not found" message for builtin <cmd> to redirected fd 2.

14 years agoFix card/device names, no functional change
Martin Blapp [Sun, 3 Jan 2010 13:59:59 +0000 (13:59 +0000)]
Fix card/device names, no functional change

The ADMtek AN985 is the cardbus variant of ADMtek AN983
The Netgear FA511 is just a relabled ADMtek AN985

PR: kern/50574
MFC after: 1 month

14 years agoFix return code in the case of successful file transfer, broken in
Gavin Atkinson [Sun, 3 Jan 2010 13:48:58 +0000 (13:48 +0000)]
Fix return code in the case of successful file transfer, broken in
tftp.c 1.13

PR: bin/117452
Submitted by: Spencer Minear  minear securecomputing.com
Approved by: ed (mentor)
MFC after: 2 weeks

14 years agosh: Add a regression test that tries out all arithmetic ops.
Jilles Tjoelker [Sun, 3 Jan 2010 12:50:27 +0000 (12:50 +0000)]
sh: Add a regression test that tries out all arithmetic ops.

MFC after: 1 week

14 years agoUse t1 & t2 as more symbolic than 'a' and 'b' for the two times.
David E. O'Brien [Sun, 3 Jan 2010 12:17:51 +0000 (12:17 +0000)]
Use t1 & t2 as more symbolic than 'a' and 'b' for the two times.

14 years agoProvide clean output with 'prove -r'.
David E. O'Brien [Sun, 3 Jan 2010 11:53:48 +0000 (11:53 +0000)]
Provide clean output with 'prove -r'.

14 years agoHide expected error output so the 'prove -r' results are easy to read.
David E. O'Brien [Sun, 3 Jan 2010 09:31:33 +0000 (09:31 +0000)]
Hide expected error output so the 'prove -r' results are easy to read.

14 years agoAllow to be driven by 'prove -r'.
David E. O'Brien [Sun, 3 Jan 2010 09:30:50 +0000 (09:30 +0000)]
Allow to be driven by 'prove -r'.

14 years agoMake sure 'pack' is usable for 'prove -r'.
David E. O'Brien [Sun, 3 Jan 2010 08:44:18 +0000 (08:44 +0000)]
Make sure 'pack' is usable for 'prove -r'.

14 years agoTransform from kernel file to userland source.
David E. O'Brien [Sun, 3 Jan 2010 08:33:24 +0000 (08:33 +0000)]
Transform from kernel file to userland source.

14 years agoAvoid NULL dereference in arpresolve.
Navdeep Parhar [Sun, 3 Jan 2010 06:43:13 +0000 (06:43 +0000)]
Avoid NULL dereference in arpresolve.

14 years agoRe-enable more options and devices. Now kernel size problem is gone.
Yoshihiro Takahashi [Sun, 3 Jan 2010 04:38:38 +0000 (04:38 +0000)]
Re-enable more options and devices.  Now kernel size problem is gone.

MFC after: 2 week

14 years agoExclude options COMPAT_FREEBSD4 now that the MD freebsd4_sigreturn()
Marius Strobl [Sun, 3 Jan 2010 02:58:43 +0000 (02:58 +0000)]
Exclude options COMPAT_FREEBSD4 now that the MD freebsd4_sigreturn()
is gone since r201396 and which is also in line with the fact that
FreeBSD 4 didn't supported sparc64.

14 years agoMake sure that the WWNN is also created for 2100..2300 cards.
Matt Jacob [Sun, 3 Jan 2010 02:43:46 +0000 (02:43 +0000)]
Make sure that the WWNN is also created for 2100..2300 cards.
MFC after: 1 day

14 years agoReduce diff against OpenSolaris - move Giant acquire/release to
Xin LI [Sat, 2 Jan 2010 23:38:03 +0000 (23:38 +0000)]
Reduce diff against OpenSolaris - move Giant acquire/release to
zfs_znode.c.  As a side effect this also eliminates two potential
Giant leaks.

Approved by: pjd
MFC after: 1 month