]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoAdd RTC support. This may be of dubious value since the RTC is reset
imp [Fri, 24 Mar 2006 07:35:30 +0000 (07:35 +0000)]
Add RTC support.  This may be of dubious value since the RTC is reset
to 1998 every reboot.

18 years agoAdd explicit dependency on cam. This is necessary when both aha.ko and
imp [Fri, 24 Mar 2006 06:33:25 +0000 (06:33 +0000)]
Add explicit dependency on cam.  This is necessary when both aha.ko and
cam.ko are modules so that aha.ko's undefined symbols can be satisfied by
cam.ko.

Sumitted by: nork
Reviewed by: scottl

18 years agoThe year field is the 4 digit year (eg, 2006), not 'year - 1900' (eg
imp [Fri, 24 Mar 2006 06:27:34 +0000 (06:27 +0000)]
The year field is the 4 digit year (eg, 2006), not 'year - 1900' (eg
106).  Fix the comment to reflect this.

18 years agoOption MK_LIB32 only exists on amd64.
ru [Fri, 24 Mar 2006 06:14:31 +0000 (06:14 +0000)]
Option MK_LIB32 only exists on amd64.

18 years agoWhen res_send() fails, we should reserve an error code.
ume [Fri, 24 Mar 2006 04:49:30 +0000 (04:49 +0000)]
When res_send() fails, we should reserve an error code.

18 years agoset an upper limit for thread structures which can be allocated.
davidxu [Fri, 24 Mar 2006 04:34:06 +0000 (04:34 +0000)]
set an upper limit for thread structures which can be allocated.

18 years agoImplement SIGEV_THREAD notification for aio_fsync.
davidxu [Fri, 24 Mar 2006 03:24:27 +0000 (03:24 +0000)]
Implement SIGEV_THREAD notification for aio_fsync.

18 years ago1. Move code for scanning pending I/O from aio_fsync to aio_aqueue,
davidxu [Fri, 24 Mar 2006 00:50:06 +0000 (00:50 +0000)]
1. Move code for scanning pending I/O from aio_fsync to aio_aqueue,
   it has less overhead.
2. Avoid scheduling task if maximum number of I/O threads is reached.

18 years agoAdd USE_BRK-specific code in malloc_init_hard() to allow the first
jasone [Fri, 24 Mar 2006 00:28:08 +0000 (00:28 +0000)]
Add USE_BRK-specific code in malloc_init_hard() to allow the first
internally used chunk to start at the beginning of the heap, rather
than at a chunk-aligned address.  This reduces mapped memory somewhat
for 32-bit architectures.

Add the arena_run_link_t type and use it wherever a run object is only
used as a ring 'header'.  This saves approximately 40 kB of memory per
arena.

Remove an obsolete (no longer used) code path from base_alloc(), which
supported the internal allocation of objects larger than the chunk
size.

Enhance chunk_dealloc() to cache chunk addresses for all deallocated
chunks.  This has no impact for most programs, but has the potential
to reduce VM map fragmentation for programs that use huge
allocations.

18 years agoProtect spx_iss using its own mutex, spx_mtx, rather than piggy-backing
rwatson [Fri, 24 Mar 2006 00:26:25 +0000 (00:26 +0000)]
Protect spx_iss using its own mutex, spx_mtx, rather than piggy-backing
on the global IPX mutex, which is not held at all necessary strategic
points.

MFC after: 1 month

18 years agoMove definition of spxrexmtthresh to top of file with other global
rwatson [Fri, 24 Mar 2006 00:22:25 +0000 (00:22 +0000)]
Move definition of spxrexmtthresh to top of file with other global
variables.

MFC after: 1 month

18 years agoCanonicalize, update copyright.
rwatson [Fri, 24 Mar 2006 00:15:58 +0000 (00:15 +0000)]
Canonicalize, update copyright.
Remove 'register'.
Use ANSI prototypes, not K&R.

MFC after: 1 month

18 years agoUpdate copyright to 2006, comment on my contribution to this code in the
rwatson [Fri, 24 Mar 2006 00:02:15 +0000 (00:02 +0000)]
Update copyright to 2006, comment on my contribution to this code in the
style of previous contributors.

MFC after: 1 month

18 years agoComment that raw output filter code for IPX should run in a netisr so as
rwatson [Fri, 24 Mar 2006 00:00:23 +0000 (00:00 +0000)]
Comment that raw output filter code for IPX should run in a netisr so as
to avoid recursing the socket code, as this input path can run in the
call stack of an output path.

MFC after: 1 month

18 years agoDocument that the documented 2GB mmap limit is actually a
peter [Thu, 23 Mar 2006 23:37:06 +0000 (23:37 +0000)]
Document that the documented 2GB mmap limit is actually a
documentation bug.  We switched to page indexes some time around
FreeBSD 2.2.  The actual 'len' limit is the maximum file size or what
will fit in your address space, whichever comes first.  It should be
possible to make 1TB files on 32 bit systems, but of course address space
runs out long before then.

18 years agoAllow to use fast_ipsec(4) on debug.mpsafenet=0 and INVARIANTS-enabled
pjd [Thu, 23 Mar 2006 23:26:34 +0000 (23:26 +0000)]
Allow to use fast_ipsec(4) on debug.mpsafenet=0 and INVARIANTS-enabled
systems. Without the change it will panic on assertions.

MFC after: 2 weeks

18 years agoWhen the kernel is compiled with options IPXIP, run the network stack
rwatson [Thu, 23 Mar 2006 23:07:56 +0000 (23:07 +0000)]
When the kernel is compiled with options IPXIP, run the network stack
with Giant, as there is current unsafety in the IPX tunneled over IP
code.  There have been no reports of trouble, but there probably would
be if anyone were running this code at high speed on SMP systems.

MFC after: 3 days

18 years agoadd support for copying console messages to a remote gdb
sam [Thu, 23 Mar 2006 23:06:14 +0000 (23:06 +0000)]
add support for copying console messages to a remote gdb

Reviewed by: kan

18 years agoFix a bug in the NFS/TCP retransmission path.
kris [Thu, 23 Mar 2006 22:58:42 +0000 (22:58 +0000)]
Fix a bug in the NFS/TCP retransmission path.

The bug was that earlier, if a request was retransmitted,
we would do subsequent retransmits every 10 msecs.

This can cause data corruption under moderate loads by reordering
operations as seen by the client NFS attribute cache, and on the
server side when the retransmission occurs after the original request
has left the duplicate cache, since the operation will be committed
for a second time.

Further work on retransmission handling is needed (e.g. they are still
being done sent too often since they are scaled by HZ, and the size of
the dup cache is too small and easily overwhelmed on busy servers).

Submitted by: mohans

18 years agom_dup () packet not m_copypacket () since we will modify it. For more
rik [Thu, 23 Mar 2006 22:57:10 +0000 (22:57 +0000)]
m_dup () packet not m_copypacket () since we will modify it. For more
details see PR kern/94448.

PR:     kern/94448

Original patch: Eygene A. Ryabinkin <rea-fbsd at rea dot mbslab dot kiae dot ru>Final patch:    thompsa@
Tested by:      thompsa@, Eygene A. Ryabinkin

MFC after:      7 days

18 years agoFix utmp. There is some clever logic in configure.ac which attempts to
des [Thu, 23 Mar 2006 21:31:42 +0000 (21:31 +0000)]
Fix utmp.  There is some clever logic in configure.ac which attempts to
determine whether struct utmp contains the ut_host and ut_time fields.
Unfortunately, it reports a false negative for both on FreeBSD, and I
didn't check the resulting config.h closely enough to catch the error.

Noticed by: ache

18 years agoAdd proper dependencies for the version map, and make sure 'make clean'
des [Thu, 23 Mar 2006 20:37:04 +0000 (20:37 +0000)]
Add proper dependencies for the version map, and make sure 'make clean'
removes it if it was generated.

Reviewed by: ru

18 years agoFix whitespace.
le [Thu, 23 Mar 2006 20:01:13 +0000 (20:01 +0000)]
Fix whitespace.

18 years agoImplement the 'resetconfig' command.
le [Thu, 23 Mar 2006 19:58:43 +0000 (19:58 +0000)]
Implement the 'resetconfig' command.

PR:            kern/94835
Submitted by:  Ulf Lilleengen <lulf@stud.ntnu.no>

18 years agoMove spx_savesi from being a global variable to an automatically allocated
rwatson [Thu, 23 Mar 2006 19:58:12 +0000 (19:58 +0000)]
Move spx_savesi from being a global variable to an automatically allocated
variable on the spx_input() stack.  It's not very large, and this will
avoid parallelism issues when spx_input() runs in more than one thread at
a time.

MFC after: 1 month

18 years agoAdmit to ourselves that we don't actually implement pr_ctlinput() for
rwatson [Thu, 23 Mar 2006 19:50:00 +0000 (19:50 +0000)]
Admit to ourselves that we don't actually implement pr_ctlinput() for
IPX or SPX, as the code in the implementing functions is essentially
a no-op.  Replace with a comment indicating we don't implement these
currently.

18 years agoIt's not enough to provide symbol map files; you also have to add them
des [Thu, 23 Mar 2006 16:55:13 +0000 (16:55 +0000)]
It's not enough to provide symbol map files; you also have to add them
to ${SYM_MAPS}.  This unbreaks world with SYMVER_ENABLED.

Pointy hat to: ume

18 years agoHarmless typo (s/overide/override/).
flz [Thu, 23 Mar 2006 16:48:32 +0000 (16:48 +0000)]
Harmless typo (s/overide/override/).

Approved by: cperciva
MFC after: 3 days

18 years agoavoid heap overrun
suz [Thu, 23 Mar 2006 14:39:38 +0000 (14:39 +0000)]
avoid heap overrun

Obtained from: KAME
MFC after: 1 day

18 years agoResolve conflicts.
des [Thu, 23 Mar 2006 14:09:21 +0000 (14:09 +0000)]
Resolve conflicts.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r157043,
des [Thu, 23 Mar 2006 14:02:36 +0000 (14:02 +0000)]
This commit was generated by cvs2svn to compensate for changes in r157043,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor import of ZLib 1.2.3.
des [Thu, 23 Mar 2006 14:02:36 +0000 (14:02 +0000)]
Vendor import of ZLib 1.2.3.

18 years agoAdd basic support for BCM5780 PHY.
oleg [Thu, 23 Mar 2006 09:53:39 +0000 (09:53 +0000)]
Add basic support for BCM5780 PHY.

Submitted by: grehan
Approved by: glebius (mentor)
MFC after: 1 week

18 years agoAdd an example of how to set up a 3-wire serial terminal.
ru [Thu, 23 Mar 2006 08:48:38 +0000 (08:48 +0000)]
Add an example of how to set up a 3-wire serial terminal.
Usually, getty(8) does the trick, but if you just need a
serial console without getty(8), here's how.

18 years agoRegenerate.
davidxu [Thu, 23 Mar 2006 08:48:37 +0000 (08:48 +0000)]
Regenerate.

18 years agoAdd aio_fsync() prototype.
davidxu [Thu, 23 Mar 2006 08:47:28 +0000 (08:47 +0000)]
Add aio_fsync() prototype.

18 years agoImplement aio_fsync() syscall.
davidxu [Thu, 23 Mar 2006 08:46:42 +0000 (08:46 +0000)]
Implement aio_fsync() syscall.

18 years agoFix version number when the feature was added.
hrs [Thu, 23 Mar 2006 07:55:33 +0000 (07:55 +0000)]
Fix version number when the feature was added.

Spotted by: ceri
MFC after: 1 day

18 years agoAssert that /dev/null is a character device in a chroot early. Otherwise,
ru [Thu, 23 Mar 2006 07:48:58 +0000 (07:48 +0000)]
Assert that /dev/null is a character device in a chroot early.  Otherwise,
we might get surprises later, like /dev/null having error in the 4th line
reported by make(1).  :-)

Tested by: Dmitriy Kirhlarov (who attempted to make release in a jail)

18 years agoMFp4: Don't force single-user now we can go multi-user.
cognet [Wed, 22 Mar 2006 22:31:31 +0000 (22:31 +0000)]
MFp4: Don't force single-user now we can go multi-user.
Call cninit() only after the pagetable has been set, as locore.S won't
map the system device for us anymore.

18 years agoDon't set primary resume interrupt flag during channel initialization
ariff [Wed, 22 Mar 2006 22:24:23 +0000 (22:24 +0000)]
Don't set primary resume interrupt flag during channel initialization
since it can cause high interrupt rate (storm) and slowdown the entire
system.

Note: Please report back to me if this commit cause any abnormal
      behaviour, especially during suspend / resume.

Reported/Submitted by: [1] Daan Vreeken [PA4DAN] <Danovitsch_at_vitsch dot net>
Reported/Confirmed by: [2] Angka H. K. <harikurniawan at gmail dot com>

MFC after: 5 days

[1] http://lists.freebsd.org/mailman/htdig/freebsd-hackers/2004-December/009335.html
[2] http://lists.freebsd.org/pipermail/freebsd-multimedia/2006-March/003830.html

18 years agoMFp4: More special casing of when vector_page == 0x00000000 :
cognet [Wed, 22 Mar 2006 22:11:10 +0000 (22:11 +0000)]
MFp4: More special casing of when vector_page == 0x00000000 :
catch attempts to write to vector_page earlier in pmap_fault_fixup(),
and deny it.

18 years agoMisc. fixups:
ariff [Wed, 22 Mar 2006 21:29:47 +0000 (21:29 +0000)]
Misc. fixups:

- [1] Make the driver friendly towards kernel without PREEMPTION.
  Use msleep(9) instead of simple unlock-check_variable-lock mechanisme
  since the later not really effective in non-preemptible kernel
  (especially during codec detection routine).
- Free most driver resources in a sane manner to avoid possible
  double free and panics especially during device detach and codec
  detection failure.

MFC after: 3 days

[1] http://lists.freebsd.org/pipermail/freebsd-questions/2006-March/116515.html

18 years agoMFp4: If we're mapping the vector page (this will happen if we didn't
cognet [Wed, 22 Mar 2006 21:23:27 +0000 (21:23 +0000)]
MFp4: If we're mapping the vector page (this will happen if we didn't
relocate it), do not attempt to call pmap_vac_me_harder() on the page.
At this point m will be NULL, and we know we won't have any cache
issues with this page.

18 years agoMFp4: teach the KB920x bits how to know where the ELF trampoline puts the
cognet [Wed, 22 Mar 2006 21:16:51 +0000 (21:16 +0000)]
MFp4: teach the KB920x bits how to know where the ELF trampoline puts the
strtab and the symtab.

18 years agoMFp4: Handle break interrupts (it seems to only work for USART, not DBGU).
cognet [Wed, 22 Mar 2006 21:16:09 +0000 (21:16 +0000)]
MFp4: Handle break interrupts (it seems to only work for USART, not DBGU).

18 years agoWe shouldn't really care about the return value of mixer_uninit(),
ariff [Wed, 22 Mar 2006 20:50:04 +0000 (20:50 +0000)]
We shouldn't really care about the return value of mixer_uninit(),
except EBUSY.

18 years agoAdd port-tun.c.
des [Wed, 22 Mar 2006 20:42:05 +0000 (20:42 +0000)]
Add port-tun.c.

18 years agoRegenerate.
des [Wed, 22 Mar 2006 20:41:53 +0000 (20:41 +0000)]
Regenerate.

18 years agoMerge conflicts.
des [Wed, 22 Mar 2006 20:41:37 +0000 (20:41 +0000)]
Merge conflicts.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r157016,
des [Wed, 22 Mar 2006 19:46:12 +0000 (19:46 +0000)]
This commit was generated by cvs2svn to compensate for changes in r157016,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor import of OpenSSH 4.3p1.
des [Wed, 22 Mar 2006 19:46:12 +0000 (19:46 +0000)]
Vendor import of OpenSSH 4.3p1.

18 years agoMake scsi_target actually compile after a structure member
mjacob [Wed, 22 Mar 2006 17:00:14 +0000 (17:00 +0000)]
Make scsi_target actually compile after a structure member
change.

Obtained from: Either Scott, Nate, Ken or A. Nagy

18 years agoAdd new libsm file from sendmail 8.13.6
gshapiro [Wed, 22 Mar 2006 16:58:15 +0000 (16:58 +0000)]
Add new libsm file from sendmail 8.13.6

MFC after: 4 days

18 years agoRecord sendmail 8.13.6 upgrade
gshapiro [Wed, 22 Mar 2006 16:48:21 +0000 (16:48 +0000)]
Record sendmail 8.13.6 upgrade

18 years agoResolve conflicts from sendmail 8.13.6 import
gshapiro [Wed, 22 Mar 2006 16:45:56 +0000 (16:45 +0000)]
Resolve conflicts from sendmail 8.13.6 import

18 years agoThis commit was generated by cvs2svn to compensate for changes in r157001,
gshapiro [Wed, 22 Mar 2006 16:40:03 +0000 (16:40 +0000)]
This commit was generated by cvs2svn to compensate for changes in r157001,
which included commits to RCS files with non-trunk default branches.

18 years agoImport sendmail 8.13.6
gshapiro [Wed, 22 Mar 2006 16:40:03 +0000 (16:40 +0000)]
Import sendmail 8.13.6

Security:       FreeBSD-SA-06:13.sendmail

18 years agoAdd missing code needed for the detection of IPSec packet replays. [1]
cperciva [Wed, 22 Mar 2006 16:00:42 +0000 (16:00 +0000)]
Add missing code needed for the detection of IPSec packet replays. [1]

Correctly identify the user running opiepasswd(1) when the login name
differs from the account name. [2]

Security: FreeBSD-SA-06:11.ipsec [1]
Security: FreeBSD-SA-06:12.opie [2]

18 years agoAdd support for RTL8111B chip, that can be found on some mainboards,
glebius [Wed, 22 Mar 2006 07:33:03 +0000 (07:33 +0000)]
Add support for RTL8111B chip, that can be found on some mainboards,
for example ASUS P5PL2.

Tested by: Vadim Frolov <vadim uch.net>

18 years agoEven if a system does not support restoring extended attributes,
kientzle [Wed, 22 Mar 2006 02:42:17 +0000 (02:42 +0000)]
Even if a system does not support restoring extended attributes,
it's only a failure if there were actually attributes to be restored.
In particular, this fixes the problem where tar -xp always returned
a failure code on FreeBSD (which doesn't yet have all of the extended
attribute support).

Thanks to: Diego "Flameeyes" Petteno

18 years agoIf a tail queue is empty the return value is not
mckusick [Wed, 22 Mar 2006 02:40:38 +0000 (02:40 +0000)]
If a tail queue is empty the return value is not
undefined, it is NULL.

18 years agoCorrect the calculation of the report size and only look at reports
iedowse [Wed, 22 Mar 2006 02:04:12 +0000 (02:04 +0000)]
Correct the calculation of the report size and only look at reports
that have the specified kind, instead of assuming that there is
only one report of the right kind in the report descriptor.

Submitted by: Morten Johansen
Obtained from: NetBSD (indirectly)
PR: usb/77604

18 years agoAt least respond to REPORT LUNS with an ILLEGAL COMMAND response.
mjacob [Wed, 22 Mar 2006 01:30:07 +0000 (01:30 +0000)]
At least respond to REPORT LUNS with an ILLEGAL COMMAND response.

This keeps us from dumping core when modern OS' like Windows and Linux
see us.

18 years agoDestroy "bip" bio in error case.
pjd [Wed, 22 Mar 2006 00:42:41 +0000 (00:42 +0000)]
Destroy "bip" bio in error case.

Found by: Coverity Prevent analysis tool
Coverity ID: 795
MFC after: 3 days

18 years agoPointer align should be generic enough to handle awkward byte size
ariff [Wed, 22 Mar 2006 00:34:17 +0000 (00:34 +0000)]
Pointer align should be generic enough to handle awkward byte size
especially for true 24bit format.

18 years ago - Remove explicit giant acquires and replace it with VFS_LOCK_GIANT.
jeff [Wed, 22 Mar 2006 00:00:05 +0000 (00:00 +0000)]
 - Remove explicit giant acquires and replace it with VFS_LOCK_GIANT.

Sponsored by: Isilon Systems, Inc.

18 years ago - Remove explicit calls to lock and unlock Giant and replace them with
jeff [Tue, 21 Mar 2006 23:58:37 +0000 (23:58 +0000)]
 - Remove explicit calls to lock and unlock Giant and replace them with
   VFS_LOCK_GIANT/VFS_UNLOCK_GIANT calls.  This completely removes Giant
   acquisition in the syscall path for ffs.

Bug fix to kern_fhstatfs from: Todd Miller <Todd.Miller@sparta.com>
Sponsored by: Isilon Systems, Inc.

18 years agoFix the LINT build on alpha:
netchild [Tue, 21 Mar 2006 21:56:04 +0000 (21:56 +0000)]
Fix the LINT build on alpha:
- rename some file local structure definitions, the names clash with
  autogenerated names
- on !alpha add some compatibility defines for those renamed structures
- make some functions globally visible on alpha

18 years agoo fixes the locking of if_init().
damien [Tue, 21 Mar 2006 21:15:43 +0000 (21:15 +0000)]
o fixes the locking of if_init().
o don't send management frames if the IFF_DRV_RUNNING flag is not set.
  this prevents the timeout watchdog from being potentially re-armed
  when the interface is brought down.

fixes a crash that occurs with RT2661 based adapters.
reported by Arnaud Lacombe.

18 years agoRemove some dummy functions, we have corresponding real ones now.
netchild [Tue, 21 Mar 2006 21:08:02 +0000 (21:08 +0000)]
Remove some dummy functions, we have corresponding real ones now.

18 years agoProperly map mouse buttons
emax [Tue, 21 Mar 2006 18:42:52 +0000 (18:42 +0000)]
Properly map mouse buttons

PR: bin/94577
Submitted by: Krzysztof Jedruczyk < beaker at hot dot pl >
MFC after: 3 days

18 years agoWe have gmtime_r(3).
ume [Tue, 21 Mar 2006 18:31:24 +0000 (18:31 +0000)]
We have gmtime_r(3).

18 years agoEliminate unnecessary invalidations of the entire TLB by pmap_remove().
alc [Tue, 21 Mar 2006 18:07:42 +0000 (18:07 +0000)]
Eliminate unnecessary invalidations of the entire TLB by pmap_remove().
Specifically, on mappings with PG_G set pmap_remove() not only performs
the necessary per-page invlpg invalidations but also performs an
unnecessary invalidation of the entire set of non-PG_G entries.

Reviewed by: tegge

18 years agoExtended attribute support on write for Linux; FreeBSD hooks are
kientzle [Tue, 21 Mar 2006 17:03:51 +0000 (17:03 +0000)]
Extended attribute support on write for Linux; FreeBSD hooks are
forthcoming.  This commit also has a number of style(9) fixes and
minor corrections so the code works better with the build system being
used for non-FreeBSD builds.

Many thanks to: Jaakko Heinonen, who proposed a mechanism for extended
attribute support and implemented both the machine-independent portion
and the Linux-specific portion.

18 years agoPOSIX.1e-style Extended Attribute support
kientzle [Tue, 21 Mar 2006 16:55:46 +0000 (16:55 +0000)]
POSIX.1e-style Extended Attribute support

This commit implements storing/reading POSIX.1e-style extended
attribute information in "pax" format archives.  An outline of the
storage format is in the tar.5 manpage.  The archive_read_extract()
function has code to restore those archives to disk for Linux; FreeBSD
implementation is forthcoming.

Many thanks to Jaakko Heinonen for finding flaws in earlier
proposals and doing the bulk of the coding in this work.

18 years agoUpdate the resolver in libc to BIND9's one.
ume [Tue, 21 Mar 2006 16:11:11 +0000 (16:11 +0000)]
Update the resolver in libc to BIND9's one.

Since, res_sendsigned(3) and the friends use MD5 functions, it is
hard to include them without having MD5 functions in libc.  So,
res_sendsigned(3) is not merged into libc.

Since, res_update(3) in BIND9 is not binary compatible with our
res_update(3), res_update(3) is leaved as is, except some
necessary modifications.
The res_update(3) and the friends are not essential part of the
resolver.  They are not defined in resolv.h but defined in
res_update.h separately in BIND9.  Further, they are not called from
our tree.  So, I hide them from our resolv.h, but leave them only
for binary backward compatibility (perhaps, no one calls them).

Since, struct __res_state_ext is not exposed in BIND9, I hide it
from our resolv.h.  And, global variable _res_ext is removed.  It
breaks binary backward compatibility.  But, since it is not used from
outside of our libc, I think it is safe.

Reviewed by: arch@ (no objection)

18 years agomake sure install scripts are executable
sam [Tue, 21 Mar 2006 15:50:08 +0000 (15:50 +0000)]
make sure install scripts are executable

Submitted by: Ceri Davies

18 years agocorrect install location
sam [Tue, 21 Mar 2006 15:49:03 +0000 (15:49 +0000)]
correct install location

Submitted by: Ceri Davies (slightly modified)

18 years ago- Merge our local changes.
ume [Tue, 21 Mar 2006 15:37:16 +0000 (15:37 +0000)]
- Merge our local changes.
- Exclude unnecessary functions for us.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r156954,
ume [Tue, 21 Mar 2006 15:07:04 +0000 (15:07 +0000)]
This commit was generated by cvs2svn to compensate for changes in r156954,
which included commits to RCS files with non-trunk default branches.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r156952,
ume [Tue, 21 Mar 2006 15:03:19 +0000 (15:03 +0000)]
This commit was generated by cvs2svn to compensate for changes in r156952,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor import of BIND 9.3.2
ume [Tue, 21 Mar 2006 15:03:19 +0000 (15:03 +0000)]
Vendor import of BIND 9.3.2

18 years agoSubtracting two pointers produces a ptrdiff_t not a size_t so use
harti [Tue, 21 Mar 2006 15:00:10 +0000 (15:00 +0000)]
Subtracting two pointers produces a ptrdiff_t not a size_t so use
the %td to print this instead of %zu or %d.

18 years agoThis driver has been MPSAFE from the beginning so declare the interrupt
harti [Tue, 21 Mar 2006 14:56:42 +0000 (14:56 +0000)]
This driver has been MPSAFE from the beginning so declare the interrupt
as such.

Reminded by: rwatson@

18 years agoThis driver has been MPSAFE from the beginning, so declare the interrupt
harti [Tue, 21 Mar 2006 14:54:42 +0000 (14:54 +0000)]
This driver has been MPSAFE from the beginning, so declare the interrupt
as such.

Reminded by: rwatson@

18 years agoNo direct call to carp_ifdetach() anymore. It is called by
glebius [Tue, 21 Mar 2006 14:31:18 +0000 (14:31 +0000)]
No direct call to carp_ifdetach() anymore. It is called by
event handler.

PR: kern/82908
Submitted by: Dan Lukes <dan obluda.cz>

18 years agoo Introduce carp_multicast_cleanup(), which removes and frees
glebius [Tue, 21 Mar 2006 14:29:48 +0000 (14:29 +0000)]
o Introduce carp_multicast_cleanup(), which removes and frees
  multicast addresses from carp interface. [1]
o Rewrite carpdetach(), so that it does the following things: [1]
  - Stops callouts.
  - Decrements carp_suppress_preempt, if needed.
  - Downs interface and sets CARP state to INIT.
  - Calls carp_multicast_cleanup().
  - Detaches softc from carp_if and if we are the last frees
    the carp_if.
o Use new carpdetach() in carp_clone_destroy().
o In carp_ifdetach() acquire the carp_if lock and cleanup all
  interfaces hanging on carp_if. [1]
o Make carp_ifdetach() static and use EVENT(9) to call it
  from if_detach(). [2]
o In carp_setrun() exit if the softc doesn't have a valid pointer
  to parent. [1]

Obtained from: OpenBSD [1]
Submitted by: Dan Lukes <dan obluda.cz> [2]
PR: kern/82908 [2]

18 years agoWhen we are doing initialization against q, use its own size, not
delphij [Tue, 21 Mar 2006 12:39:29 +0000 (12:39 +0000)]
When we are doing initialization against q, use its own size, not
the size of q2.  This should be a no-op because q and q2 are of
the same type.

Submitted by: Alexey Dobriyan <adobriyan gmail com>

18 years agoBackout rev. 1.46. It caused Rx checksum offload breakage on little
yongari [Tue, 21 Mar 2006 12:21:51 +0000 (12:21 +0000)]
Backout rev. 1.46. It caused Rx checksum offload breakage on little
endian systems.

Reported by: joerg
Tested by: joerg

18 years agoUnbreak WITHOUT_LIBPHREAD/WITHOUT_LIBC_R option support, depending
ru [Tue, 21 Mar 2006 11:00:54 +0000 (11:00 +0000)]
Unbreak WITHOUT_LIBPHREAD/WITHOUT_LIBC_R option support, depending
on platform.

18 years agoDocument new world older for world/kernel build options.
ru [Tue, 21 Mar 2006 10:10:05 +0000 (10:10 +0000)]
Document new world older for world/kernel build options.

18 years agoRethink it a bit, if there is a STOP flag, don't bother to resume other
davidxu [Tue, 21 Mar 2006 10:05:15 +0000 (10:05 +0000)]
Rethink it a bit, if there is a STOP flag, don't bother to resume other
threads.

18 years agoRemove mentions of old compat options from here; equivalents are
ru [Tue, 21 Mar 2006 09:49:05 +0000 (09:49 +0000)]
Remove mentions of old compat options from here; equivalents are
now provided in src.conf(5).

18 years agoInstall the autogenerated src.conf(5) manpage.
ru [Tue, 21 Mar 2006 09:10:35 +0000 (09:10 +0000)]
Install the autogenerated src.conf(5) manpage.

18 years agoFix generation of '$FreeBSD$'.
ru [Tue, 21 Mar 2006 09:08:45 +0000 (09:08 +0000)]
Fix generation of '$FreeBSD$'.

18 years agoSlightly better description.
ru [Tue, 21 Mar 2006 09:06:24 +0000 (09:06 +0000)]
Slightly better description.

18 years agoA tool to generate the src.conf(5) manpage.
ru [Tue, 21 Mar 2006 09:05:42 +0000 (09:05 +0000)]
A tool to generate the src.conf(5) manpage.

18 years agoFolded WITH_HESIOD_SUPPORT into WITH_HESIOD.
ru [Tue, 21 Mar 2006 09:00:51 +0000 (09:00 +0000)]
Folded WITH_HESIOD_SUPPORT into WITH_HESIOD.

18 years agoDesupport the *_COLLECT2 no-op.
ru [Tue, 21 Mar 2006 08:59:36 +0000 (08:59 +0000)]
Desupport the *_COLLECT2 no-op.