]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years agoFix typo.
Xin LI [Wed, 31 Jan 2007 07:13:25 +0000 (07:13 +0000)]
Fix typo.

Obtained from: DragonFly

17 years agoo DragonFly 1.8.0 added.
Maxim Konovalov [Wed, 31 Jan 2007 04:58:03 +0000 (04:58 +0000)]
o DragonFly 1.8.0 added.

17 years agoRemove GNU gzip left-offs, they are no longer necessary.
Xin LI [Wed, 31 Jan 2007 03:44:42 +0000 (03:44 +0000)]
Remove GNU gzip left-offs, they are no longer necessary.

17 years agoFix typo.
Gleb Smirnoff [Tue, 30 Jan 2007 20:22:07 +0000 (20:22 +0000)]
Fix typo.

Submitted by: pluknet <pluknet gmail.com>

17 years agoAdd missing library dependencies and sort libraries
Ruslan Ermilov [Tue, 30 Jan 2007 15:10:37 +0000 (15:10 +0000)]
Add missing library dependencies and sort libraries
using lorder(1) so that static linkage is possible.

17 years agoFix the logic bug that caused the custom versions of warn(3) and
Ruslan Ermilov [Tue, 30 Jan 2007 15:09:30 +0000 (15:09 +0000)]
Fix the logic bug that caused the custom versions of warn(3) and
warnx(3) to be compiled on systems that have it (e.g. FreeBSD),
while the intention was opposite, i.e., compile them on systems
that don't have them.  Also fixes static linkage of pkg_sign(1).

17 years agoMFsparc64: Add .cvsignore file here too.
Ruslan Ermilov [Tue, 30 Jan 2007 10:50:55 +0000 (10:50 +0000)]
MFsparc64: Add .cvsignore file here too.

17 years agoRemove the last vestige of opt_msdosfs.h.
Ruslan Ermilov [Tue, 30 Jan 2007 10:17:36 +0000 (10:17 +0000)]
Remove the last vestige of opt_msdosfs.h.

Submitted by: grep(1)

17 years agoAdd the 3Com 3c996-SX.
Christian Brueffer [Tue, 30 Jan 2007 09:47:31 +0000 (09:47 +0000)]
Add the 3Com 3c996-SX.

Submitted by: Johan Ström
MFC after: 3 days

17 years agoXref altq(4) for drivers that support it according to altq(4).
Christian Brueffer [Tue, 30 Jan 2007 08:40:04 +0000 (08:40 +0000)]
Xref altq(4) for drivers that support it according to altq(4).

17 years agoMinor updates:
Andrew Gallatin [Tue, 30 Jan 2007 08:39:44 +0000 (08:39 +0000)]
Minor updates:

- initialize ifq_drv_maxlen correctly
- mark the interface as jumbo capable
- keep stats on the number of times the hw transmit queue filled and
  was restarted.

17 years agoXref altq(4).
Christian Brueffer [Tue, 30 Jan 2007 08:17:45 +0000 (08:17 +0000)]
Xref altq(4).

17 years agoFixing compilation bustage by removing references to opt_msdosfs.h.
Tai-hwa Liang [Tue, 30 Jan 2007 08:05:04 +0000 (08:05 +0000)]
Fixing compilation bustage by removing references to opt_msdosfs.h.

This auto-generated header file no longer exists since the removal of
MSDOSFS_LARGE in sys/conf/options:1.574.

17 years agoRemove MSDOSFS_LARGE compile time option. It has been converted
Craig Rodrigues [Tue, 30 Jan 2007 05:01:06 +0000 (05:01 +0000)]
Remove MSDOSFS_LARGE compile time option.  It has been converted
to a run time "-o large" mount option.

PR: 105964
MFC after: 2 weeks

17 years agoFix spacing from my previous commit to this file:
Tom Rhodes [Tue, 30 Jan 2007 04:41:38 +0000 (04:41 +0000)]
Fix spacing from my previous commit to this file:

Noticed by: fjoe

17 years agoAdd a "-o large" mount option for msdosfs. Convert compile-time checks for
Craig Rodrigues [Tue, 30 Jan 2007 03:11:45 +0000 (03:11 +0000)]
Add a "-o large" mount option for msdosfs.  Convert compile-time checks for
#ifdef MSDOSFS_LARGE to run-time checks to see if "-o large" was specified.

Test case provided by Oliver Fromme:
  truncate -s 200G test.img
  mdconfig -a -t vnode -f test.img -u 9
  newfs_msdos -s 419430400 -n 1 /dev/md9 zip250
  mount -t msdosfs /dev/md9 /mnt    # should fail
  mount -t msdosfs -o large /dev/md9 /mnt   # should succeed

PR: 105964
Requested by: Oliver Fromme <olli lurza secnetix de>
Tested by: trhodes
MFC after: 2 weeks

17 years agoUse our own timer that piggybacks on npe_tick() callout instead of
Kevin Lo [Tue, 30 Jan 2007 01:18:29 +0000 (01:18 +0000)]
Use our own timer that piggybacks on npe_tick() callout instead of
if_watchdog/if_timer interface.

Approved by: sam, cognet

17 years agoInstead of always hard-coding the socket type for the nfs root mount as
Kris Kennaway [Tue, 30 Jan 2007 00:26:04 +0000 (00:26 +0000)]
Instead of always hard-coding the socket type for the nfs root mount as
SOCK_DGRAM (i.e. UDP), respect the value configured earlier.  This allows
TCP NFS root mounts using e.g. the boot.nfsroot.options="tcp" tunable.

In this case some of the connection parameters like the retry timer were
previously set appropriately for TCP but inappropriately for the UDP
socket that was actually used, leading to e.g. extremely long recovery
times (O(hours)) after a nfs server reboot.

Reviewed by:    mohans
MFC After:      2 weeks

17 years agoAdd gjournal(8) manual page. Documentation of kern.geom.journal.* sysctls
Pawel Jakub Dawidek [Mon, 29 Jan 2007 22:12:22 +0000 (22:12 +0000)]
Add gjournal(8) manual page. Documentation of kern.geom.journal.* sysctls
is still missing, ehh.

Reviewed by: trhodes

17 years agoUpdate generated files for BIND 9.3.4
Doug Barton [Mon, 29 Jan 2007 18:33:18 +0000 (18:33 +0000)]
Update generated files for BIND 9.3.4

17 years agoThis commit was generated by cvs2svn to compensate for changes in r166332,
Doug Barton [Mon, 29 Jan 2007 18:31:57 +0000 (18:31 +0000)]
This commit was generated by cvs2svn to compensate for changes in r166332,
which included commits to RCS files with non-trunk default branches.

17 years agoVendor import of BIND 9.3.4
Doug Barton [Mon, 29 Jan 2007 18:31:57 +0000 (18:31 +0000)]
Vendor import of BIND 9.3.4

17 years agoActually implement rev 1.12 for host names and NIS domain names. We
Brooks Davis [Mon, 29 Jan 2007 15:55:13 +0000 (15:55 +0000)]
Actually implement rev 1.12 for host names and NIS domain names.  We
were removing the invalid option, but still rejecting the lease.

Reported by: Yoshihiko Sarumaru <mistral at imasy dot or dot jp>

17 years agoUpdate comment for struct bpf_d: we now store buffered packets for BPF
Robert Watson [Mon, 29 Jan 2007 14:41:03 +0000 (14:41 +0000)]
Update comment for struct bpf_d: we now store buffered packets for BPF
in malloc'd storage, not in mbuf clusters.

17 years agoAdd MLINKS for sf_buf.9.
Robert Watson [Mon, 29 Jan 2007 11:27:44 +0000 (11:27 +0000)]
Add MLINKS for sf_buf.9.

17 years agoPass "errmsg" to nmount(), so that if nmount() fails, we can get
Craig Rodrigues [Mon, 29 Jan 2007 01:49:08 +0000 (01:49 +0000)]
Pass "errmsg" to nmount(), so that if nmount() fails, we can get
strings provided by vfs_mount_error().

17 years agoWe expect 'bio_data != NULL' for BIO_{READ,WRITE,GETATTR}, but for
Pawel Jakub Dawidek [Sun, 28 Jan 2007 23:36:07 +0000 (23:36 +0000)]
We expect 'bio_data != NULL' for BIO_{READ,WRITE,GETATTR}, but for
BIO_{DELETE,FLUSH} we expect 'bio_data == NULL'.

Reviewed by: phk

17 years agotypo: s/tranport layer/transport layer/
Wilko Bulte [Sun, 28 Jan 2007 21:34:07 +0000 (21:34 +0000)]
typo: s/tranport layer/transport layer/

MFC after: 1 week

17 years agoLine break before "All rights reserved" in the license example.
Joel Dahl [Sun, 28 Jan 2007 20:51:04 +0000 (20:51 +0000)]
Line break before "All rights reserved" in the license example.

Reveiwed by: bde, rwatson

17 years agoClean up the BSD license to match the preferred license in
Joel Dahl [Sun, 28 Jan 2007 20:38:07 +0000 (20:38 +0000)]
Clean up the BSD license to match the preferred license in
/usr/share/examples/etc/bsd-style-copyright.  I've fixed a
few minor wording and formatting differences.

Approved by: matk, Hannu Savolainen <hannu@opensound.com>
Reviewed by: imp

17 years agoIt is possible that GEOM taste provider before SMP is started.
Pawel Jakub Dawidek [Sun, 28 Jan 2007 20:29:12 +0000 (20:29 +0000)]
It is possible that GEOM taste provider before SMP is started.
We can't bind to a CPU which is not yet on-line, so add code that wait for
CPUs to go on-line before binding to them.

Reported by: Alin-Adrian Anton <aanton@spintech.ro>
MFC after: 2 weeks

17 years agoath and ath_rate_sample now have a compile-time dependency on the hal
Sam Leffler [Sun, 28 Jan 2007 18:35:46 +0000 (18:35 +0000)]
ath and ath_rate_sample now have a compile-time dependency on the hal
so we need to build them only on architectures where there's a released
hal; this hack can be eliminated when an ia64 hal build is present

17 years agoAs we now have an SFB_NOWAIT flag, change 'will' to 'may' where the
Robert Watson [Sun, 28 Jan 2007 17:39:03 +0000 (17:39 +0000)]
As we now have an SFB_NOWAIT flag, change 'will' to 'may' where the
comment for sf_buf_alloc(9) talks about sleeping.

17 years agoRemove slightly dubious comment; add descriptive strings for several
Robert Watson [Sun, 28 Jan 2007 16:38:44 +0000 (16:38 +0000)]
Remove slightly dubious comment; add descriptive strings for several
sysctls.

MFC after: 3 days

17 years agoUpdate pmap_extract() implementation notes: we appear no longer to serialize
Robert Watson [Sun, 28 Jan 2007 16:23:55 +0000 (16:23 +0000)]
Update pmap_extract() implementation notes: we appear no longer to serialize
this call with Giant on any platform.

17 years agoAdd a rudimentary man page for sf_bufs, based on my rudimentary
Robert Watson [Sun, 28 Jan 2007 16:07:50 +0000 (16:07 +0000)]
Add a rudimentary man page for sf_bufs, based on my rudimentary
understanding of sf_bufs.

17 years agoAdd vnconfig to this list
Poul-Henning Kamp [Sun, 28 Jan 2007 14:52:07 +0000 (14:52 +0000)]
Add vnconfig to this list

Reminded by: netchild

17 years agoFix minor type: "struct disk *" instead of "struct *disk".
Robert Watson [Sun, 28 Jan 2007 14:37:13 +0000 (14:37 +0000)]
Fix minor type: "struct disk *" instead of "struct *disk".

MFC after: 3 days

17 years agoAdd support for serial communication with Windows CE based Handheld Computer.
Takanori Watanabe [Sun, 28 Jan 2007 11:56:14 +0000 (11:56 +0000)]
Add support for serial communication with Windows CE based Handheld Computer.

Obtained from: NetBSD

17 years agoAdd some vendor IDs mainly from NetBSD.
Takanori Watanabe [Sun, 28 Jan 2007 10:46:32 +0000 (10:46 +0000)]
Add some vendor IDs mainly from NetBSD.

17 years agoLong overdue removal of vnconfig(8)
Poul-Henning Kamp [Sun, 28 Jan 2007 08:53:48 +0000 (08:53 +0000)]
Long overdue removal of vnconfig(8)

mdconfig(8) took over the job back when GEOM happened.

PR: 108360
Submitted by: Alex Kozlov<spam@rm-rf.kiev.ua>

17 years agoMFi386: revision 1.647.
Yoshihiro Takahashi [Sun, 28 Jan 2007 07:19:14 +0000 (07:19 +0000)]
MFi386: revision 1.647.

  exclude the icu and clock lock from LOCK_PROFILING

17 years agofor newer hal's we need opt_ah.h as it specifies how the hal has been
Sam Leffler [Sun, 28 Jan 2007 04:38:35 +0000 (04:38 +0000)]
for newer hal's we need opt_ah.h as it specifies how the hal has been
configured and that in turn controls the descriptor layout

17 years agofor newer hal's we need opt_ah.h as it specifies how the hal has been
Sam Leffler [Sun, 28 Jan 2007 04:36:05 +0000 (04:36 +0000)]
for newer hal's we need opt_ah.h as it specifies how the hal has been
configured and that in turn controls the descriptor layout; the rate
control module has no business peeking inside the descriptor but until
we can change the api so the driver records the tx rates and passes
them deal with it

17 years agoAdd speaker control for HP xw4300. This hardware doesn't respond to
Ariff Abdullah [Sun, 28 Jan 2007 03:16:54 +0000 (03:16 +0000)]
Add speaker control for HP xw4300. This hardware doesn't respond to
unsolicited pin sense event and need manual control to turn off speaker
volume while attaching headphone.

Tested by: Ingeborg Hellemo <Ingeborg.Hellemo@cc.uit.no>

Disable global Acer + ALC883 headphone automute settings since there are
few models that does not respect this and causing broken behaviour.

Reported/Tested by: Pavel Argentov <argentoff@rtelekom.ru>

17 years agoConvert mount_ext2fs to a simple program which passes "-o option" to nmount().
Craig Rodrigues [Sun, 28 Jan 2007 00:51:01 +0000 (00:51 +0000)]
Convert mount_ext2fs to a simple program which passes "-o option" to nmount().

17 years agoSimple README file which shows how to use fstest.
Pawel Jakub Dawidek [Sun, 28 Jan 2007 00:10:28 +0000 (00:10 +0000)]
Simple README file which shows how to use fstest.

17 years agoFix grammar [1] and formatting.
Ruslan Ermilov [Sat, 27 Jan 2007 23:06:00 +0000 (23:06 +0000)]
Fix grammar [1] and formatting.

Submitted by: pluknet AT gmail DOT com [1]

17 years agoAdd the SMART command to the ATA instruction set.
Remko Lodder [Sat, 27 Jan 2007 21:15:59 +0000 (21:15 +0000)]
Add the SMART command to the ATA instruction set.
When the disk has an error, it will now print SMART
instead of 'Unknown CMD'.

PR: kern/93368
Submitted by: Garry Belka <garry at NetworkPhysics dot COM>
Approved by: sos

17 years agoIn case we are supplied with an imagename that matches a module, but not a
Max Laier [Sat, 27 Jan 2007 19:52:08 +0000 (19:52 +0000)]
In case we are supplied with an imagename that matches a module, but not a
firmware in that module (eventhough this is a programming error) - drop the
reference to the module again.

Submitted by: Benjamin Close
MFC after: 3 days

17 years agoAdd a missing verb in describing MAP_PREFAULT_MADVISE.
Robert Watson [Sat, 27 Jan 2007 18:58:33 +0000 (18:58 +0000)]
Add a missing verb in describing MAP_PREFAULT_MADVISE.

17 years agoUse a known good stack at the time of servicing an NMI --- reuse
Joseph Koshy [Sat, 27 Jan 2007 18:13:24 +0000 (18:13 +0000)]
Use a known good stack at the time of servicing an NMI --- reuse
the space allocated for the double fault handler since this space
is otherwise unused till the time a double fault occurs.

This change should have been committed alongside r1.127 of
"exception.S", but I somehow missed doing so.

Problem reported by: jeff
Pointy hat to: jkoshy

17 years agoRemove BSD < 199103 compatibility entries in the bpf_d structure: they are
Robert Watson [Sat, 27 Jan 2007 18:12:50 +0000 (18:12 +0000)]
Remove BSD < 199103 compatibility entries in the bpf_d structure: they are
not used in any of our code.  Also remove explicit padding variable that
kept the bpf_d structure the same size before and after the change in
select implementation, since binary compatibility is not required for this
data structure on 7-CURRENT.

17 years agoRemove now unused bpf_compat.h. This compatibility file emulates malloc(9)
Robert Watson [Sat, 27 Jan 2007 17:32:12 +0000 (17:32 +0000)]
Remove now unused bpf_compat.h.  This compatibility file emulates malloc(9)
using the mbuf allocator.

17 years agoRearrange locking order to avoid LOR (cat /dev/midistat).
Ariff Abdullah [Sat, 27 Jan 2007 15:55:59 +0000 (15:55 +0000)]
Rearrange locking order to avoid LOR (cat /dev/midistat).

Reported by: rodrigc

17 years agoMassive inlining cleanups/removal to make it survive on WARNS=2.
Ariff Abdullah [Sat, 27 Jan 2007 13:30:19 +0000 (13:30 +0000)]
Massive inlining cleanups/removal to make it survive on WARNS=2.

17 years agoReduce maximum DMA segments from 128 to 64. We don't need more than that.
Ariff Abdullah [Sat, 27 Jan 2007 07:35:05 +0000 (07:35 +0000)]
Reduce maximum DMA segments from 128 to 64. We don't need more than that.

17 years agoTotal DMA segments should include total number of record channel(s).
Ariff Abdullah [Fri, 26 Jan 2007 23:53:56 +0000 (23:53 +0000)]
Total DMA segments should include total number of record channel(s).

17 years agoRevert nd6.c revs. 1.67, 1.68, 1.69, 1.70 in an attempt to unbreak
Bruce A. Mah [Fri, 26 Jan 2007 23:22:58 +0000 (23:22 +0000)]
Revert nd6.c revs. 1.67, 1.68, 1.69, 1.70 in an attempt to unbreak
IPv6 over point-to-point gif(4) tunnels.

These revisions caused a host route to the destination of a
point-to-point gif(4) interface to not get installed when the interface
and destination addresses were assigned.  This caused
"no route to host" errors when trying to send traffic over the
interface.  The first packet arriving inbound over the tunnel,
however, would cause the correct route to get installed, allowing
subsequent outbound traffic to be routed correctly.

gif(4) interfaces with prefix lengths of less than 128 bits
(i.e. no explicit destination address assigned) were not affected
by this bug.

This bug fix is a possible candidate for a 6.2-RELEASE errata note.

Approved by: jhay (original committer)
Discussed with: jhay, JINMEI Tatuya
MFC after: 3 days

17 years agoFix forever broken ua_chan_setblocksize() uninitialized return value
Ariff Abdullah [Fri, 26 Jan 2007 19:14:41 +0000 (19:14 +0000)]
Fix forever broken ua_chan_setblocksize() uninitialized return value
which causing divide by zero panic in other places (notably chn_sync()).

17 years agoSync uaudio_sndstat_prepare_pcm() output with sndstat_prepare_pcm() to get
Ariff Abdullah [Fri, 26 Jan 2007 19:06:17 +0000 (19:06 +0000)]
Sync uaudio_sndstat_prepare_pcm() output with sndstat_prepare_pcm() to get
simmilar (debugging) output.

17 years agoAdd missing MIIBUS_MEDIAINIT() call.
Doug White [Fri, 26 Jan 2007 17:06:02 +0000 (17:06 +0000)]
Add missing MIIBUS_MEDIAINIT() call.

17 years agoCollapse 5706C and 5708C PHYs into one entry. ID 0x15 is actually used for
Doug White [Fri, 26 Jan 2007 17:05:24 +0000 (17:05 +0000)]
Collapse 5706C and 5708C PHYs into one entry. ID 0x15 is actually used for
the SERDES PHY on these chips and we want gentbi to pick this up, not brgphy.

17 years agoAdd support for SERDES PHY configurations. These are commonly found in
Doug White [Fri, 26 Jan 2007 17:03:51 +0000 (17:03 +0000)]
Add support for SERDES PHY configurations. These are commonly found in
blade systems, such as the Dell 1955 and the Intel SBXD132.

Development hardware for this work was provided by Broadcom and iXsystems.
A SBXD132 blade for testing was provided by Iron Systems.

17 years agoWhile we do not expect any change before and after GNU gzip
Xin LI [Fri, 26 Jan 2007 14:57:17 +0000 (14:57 +0000)]
While we do not expect any change before and after GNU gzip
is replaced with BSD gzip, let's make it possible to
distinguish between the two with a __FreeBSDversion bump,
just in case some developers want it.

Suggested by: linimon

17 years agoIf nmount() fails to export a directory, in the syslog() error message,
Craig Rodrigues [Fri, 26 Jan 2007 13:26:43 +0000 (13:26 +0000)]
If nmount() fails to export a directory, in the syslog() error message,
add the errmsg string returned by nmount().

17 years agoRegen.
Xin LI [Fri, 26 Jan 2007 10:20:59 +0000 (10:20 +0000)]
Regen.

17 years agoReplace the GNU gzip with a slightly modified NetBSD gzip. The
Xin LI [Fri, 26 Jan 2007 10:19:08 +0000 (10:19 +0000)]
Replace the GNU gzip with a slightly modified NetBSD gzip.  The
NetBSD version is a feature-to-feature re-implementation of GNU
gzip using the freely-redistributable zlib and this version is
expected to be mostly bug-to-bug compatible with the GNU
implementation.

 - Because this is a piece of mature code and we want to make
   changes so it is added directly rather than importing to
   src/contrib.
 - Connect newly added code to src/usr.bin/ and rescue/rescue
   build.
 - Disconnect the GNU gzip code from build for now, they will
   be eventually removed completely.
 - Provide two new src.conf(5) knobs, WITHOUT_BZIP2_SUPPORT and
   WITHOUT_BZIP2.

Tested by: kris (full exp-7 pointyhat build)
Approved by: core (importing a 4-clause BSD licensed file)
Approved by: re (adding new utility during -HEAD code slush)

17 years agoo Remove screen, created by the unnecessary ofw_console(4)
Marcel Moolenaar [Fri, 26 Jan 2007 06:22:34 +0000 (06:22 +0000)]
o  Remove screen, created by the unnecessary ofw_console(4)
o  Remove ttyyX, created by the obsolete zs(4)
o  Replace ttydX by ttyuX, which is created by uart(4)
o  Enable ttyu0 as a console to better support the xserve.

MFC after: 1 week

17 years agoRemove stale header.
Marcel Moolenaar [Fri, 26 Jan 2007 04:58:31 +0000 (04:58 +0000)]
Remove stale header.

MFC after: 3 days

17 years agoFix comments.
Kevin Lo [Fri, 26 Jan 2007 01:37:32 +0000 (01:37 +0000)]
Fix comments.

Approved by: cognet

17 years ago - Implement much more intelligent ipi sending. This algorithm tries to
Jeff Roberson [Thu, 25 Jan 2007 23:51:59 +0000 (23:51 +0000)]
 - Implement much more intelligent ipi sending.  This algorithm tries to
   minimize IPIs and rescheduling when scheduling like tasks while keeping
   latency low for important threads.
   1) An idle thread is running.
   2) The current thread is worse than realtime and the new thread is
      better than realtime.  Realtime to realtime doesn't preempt.
   3) The new thread's priority is less than the threshold.

17 years agoList obsolete objformat files and man pages. This may have to be
Peter Wemm [Thu, 25 Jan 2007 23:13:55 +0000 (23:13 +0000)]
List obsolete objformat files and man pages.  This may have to be
revisited if too much stuff breaks.

17 years agoBah. Kris says the default-to-a.out knowledge has migrated into the
Peter Wemm [Thu, 25 Jan 2007 23:12:19 +0000 (23:12 +0000)]
Bah.  Kris says the default-to-a.out knowledge has migrated into the
official gnu configure scripts and a couple of other places.

Add an example noisy, loud and annoying placeholder for /usr/bin/objformat
if it turns out to too much trouble to be gone.  It is not connected to
the build yet.

17 years agoRemove unused reference to objformat.h
Peter Wemm [Thu, 25 Jan 2007 22:38:58 +0000 (22:38 +0000)]
Remove unused reference to objformat.h

17 years agoOops, remove an objformat.h reference.
Peter Wemm [Thu, 25 Jan 2007 22:38:04 +0000 (22:38 +0000)]
Oops, remove an objformat.h reference.

17 years agoRemove getobjformat() from libc's symbol map. It probably should have been
Peter Wemm [Thu, 25 Jan 2007 22:36:36 +0000 (22:36 +0000)]
Remove getobjformat() from libc's symbol map.  It probably should have been
in the private area anyway.  Nothing in FreeBSD uses it any more anyway.

17 years agoRemove more a.out strays, as threatened in 2002.
Peter Wemm [Thu, 25 Jan 2007 22:34:22 +0000 (22:34 +0000)]
Remove more a.out strays, as threatened in 2002.

17 years agoRetire objformat(1) as threatened in 2002.
Peter Wemm [Thu, 25 Jan 2007 22:26:41 +0000 (22:26 +0000)]
Retire objformat(1) as threatened in 2002.

Laughed-at-by: kris
17 years agoRetire more remnants of a.out support, as threatened in 2002.
Peter Wemm [Thu, 25 Jan 2007 22:25:59 +0000 (22:25 +0000)]
Retire more remnants of a.out support, as threatened in 2002.

Laughed-at-by: kris
17 years ago- Create ng_ppp_bypass() function, that prepares a packet
Gleb Smirnoff [Thu, 25 Jan 2007 21:16:50 +0000 (21:16 +0000)]
- Create ng_ppp_bypass() function, that prepares a packet
  with bypass header, to send it out to userland.
- Use ng_ppp_bypass() in ng_ppp_proto_recv().
- Use ng_ppp_bypass() in ng_ppp_comp_recv() and in
  ng_ppp_crypt_recv() if compression or encryption is
  disabled, respectively.
- Any LCP packet goes directly to ng_ppp_bypass(), instead
  of passing through PPP stack.
- Any non-LCP packet on disabled link is discarded. This
  is behavior defined in RFC.

Submitted by: Alexander Motin <mav alkar.net>

17 years agoAdd "server mode" to rfcomm_sppd(1).
Maksim Yevmenkin [Thu, 25 Jan 2007 20:54:59 +0000 (20:54 +0000)]
Add "server mode" to rfcomm_sppd(1).

Submitted by: Dave Eckhardt, bms
Tested by: Dave Eckhardt, Eric Anderson, bms

17 years agoMFp4: - When new object is created, it's group ID can be set to process'
Pawel Jakub Dawidek [Thu, 25 Jan 2007 20:50:02 +0000 (20:50 +0000)]
MFp4: - When new object is created, it's group ID can be set to process'
  effective group ID or to group ID of its parent directory.
- Add some comments from POSIX.
- Verify that after successful O_TRUNC open, size is equal to 0.

17 years agoMFp4: When user is not a member of the group which owns a file, even if
Pawel Jakub Dawidek [Thu, 25 Jan 2007 20:48:14 +0000 (20:48 +0000)]
MFp4: When user is not a member of the group which owns a file, even if
he is the file's owner, he can't set set-gid bit.
POSIX requires to return 0 and clear the bit, but FreeBSD returns
EPERM for UFS in such case. For now do the same in ZFS.

17 years agoBump .Dd.
Mike Pritchard [Thu, 25 Jan 2007 19:54:03 +0000 (19:54 +0000)]
Bump .Dd.

17 years ago - Get rid of the unused DIDRUN flag. This was really only present to
Jeff Roberson [Thu, 25 Jan 2007 19:14:11 +0000 (19:14 +0000)]
 - Get rid of the unused DIDRUN flag.  This was really only present to
   support sched_4bsd.
 - Rename the KTR level for non schedgraph parsed events.  They take event
   space from things we'd like to graph.
 - Reset our slice value after we sleep.  The slice is simply there to
   prevent starvation among equal priorities.  A thread which had almost
   exhausted it's slice and then slept doesn't need to be rescheduled a
   tick after it wakes up.
 - Set the maximum slice value to a more conservative 100ms now that it is
   more accurately enforced.

17 years agoMake it possible that carpdetach() unlocks on return. Then, in
Gleb Smirnoff [Thu, 25 Jan 2007 18:03:40 +0000 (18:03 +0000)]
Make it possible that carpdetach() unlocks on return. Then, in
carp_clone_destroy() we are on a safe side, we don't need to
unlock the cif, that can me already non-existent at this point.

Reported by: Anton Yuzhaninov <citrin rambler-co.ru>

17 years agoWhoops- #ifdef problem caused uninitialized transport. Not horribly
Matt Jacob [Thu, 25 Jan 2007 18:02:23 +0000 (18:02 +0000)]
Whoops- #ifdef problem caused uninitialized transport. Not horribly
a problem, but caused annoying messages.

17 years agoSpacing.
Gleb Smirnoff [Thu, 25 Jan 2007 17:58:16 +0000 (17:58 +0000)]
Spacing.

17 years agoThe TCP checksum offload handling in the 8111B/8168B and 8101E PCIe can
Bill Paul [Thu, 25 Jan 2007 17:30:30 +0000 (17:30 +0000)]
The TCP checksum offload handling in the 8111B/8168B and 8101E PCIe can
apparently be confused by short TCP segments that have been manually
padded to the minimum ethernet frame size. The driver does short frame
padding in software as a workaround for a bug in the 8169 PCI devices
that causes short IP fragments to be corrupted due to an apparent
conflict between the hardware autopadding and hardware IP checksumming.

To fix this, we avoid software padding for short TCP segments, since
the hardware seems to autopad and checksum these correctly (even the
older 8169 NICs get these right). Short UDP packets appear to be
handled correctly in all cases. This should work around the IP header
checksum bug in the 8169 while not tripping the TCP checksum bug in
the 8111B/8168B and 8101E.

17 years agoRename some functions and variables from nfs_* to nfs4_* to avoid
Bruce Evans [Thu, 25 Jan 2007 14:33:13 +0000 (14:33 +0000)]
Rename some functions and variables from nfs_* to nfs4_* to avoid
collisions with nfsclient's names.  Even static names should have a
unique prefix so that they can be debugged easily.

Hide the unused colliding variable nfsv3_commit_on_close in "#if 0"
together with other unused sysctl variables.  Duplicating the nfs sysctl
under nfs4 is probably just a bug.

Fix some nearby style bugs.

Remove duplicate $FreeBSD$.

17 years agoRename some functions and variables (mainly vfsops entry points) from
Bruce Evans [Thu, 25 Jan 2007 14:18:40 +0000 (14:18 +0000)]
Rename some functions and variables (mainly vfsops entry points) from
nfs_* to nfs4_* to avoid collisions with nfsclient's names.   Even
static names should have a unique prefix so that they can be debugged
easily.

Most of the renamed functions can probably be shared.  nfs4_cmount()
and nfs4_sync() are identical to the nfs_* versions, and all the others
except nfs4_vfsops() seem to be idendentical except for style bugs,
missing support for mountroot, and bugs.

Fix some nearby style bugs.

Remove duplicate $FreeBSD$.

17 years agoUnstaticize nfs_iosize() in nfsclient and use it in nfs4client instead
Bruce Evans [Thu, 25 Jan 2007 13:07:25 +0000 (13:07 +0000)]
Unstaticize nfs_iosize() in nfsclient and use it in nfs4client instead
of duplicating it except for larger style bugs in the copy.

Fix some nearby style bugs (including a harmless type mismatch)
in and near the remaining copy.

This is part of fixing collisions of the 2 nfs*client's names.  Even
static names should have a unique prefixes so that they can be debugged
easily.

17 years agoAdd a BUGS section that shows that ids that appear to be
Mike Pritchard [Thu, 25 Jan 2007 12:42:18 +0000 (12:42 +0000)]
Add a BUGS section that shows that ids that appear to be
negative are now ignored by the quota system and that extremely
large ids may make quotacheck run for a very long time.

Also mention that "options QUOTA" is required for the kernel
to provide quota support.

17 years agoWhen the following conditions are meet:
Pawel Jakub Dawidek [Thu, 25 Jan 2007 11:44:03 +0000 (11:44 +0000)]
When the following conditions are meet:
- First configured key is based only on keyfile (no passphrase).
- Device is attached.
- User changes first key (setkey) from keyfile to passphrase and doesn't
  specify number of iterations (with -i option).
...geli(8) won't store calculated number of iterations in metadata.
This result in device beeing unaccesable after detach.

One can recover from this situation by guessing number of iterations
generated, storing it in metadata and trying to attach device.
Recovery procedure isn't nice, but one's data is not lost.

Reported by: Thomas Nickl <T.Nickl@gmx.net>
MFC after: 1 week

17 years agoImplement gctl_change_param() function, which changes value of existing
Pawel Jakub Dawidek [Thu, 25 Jan 2007 11:35:27 +0000 (11:35 +0000)]
Implement gctl_change_param() function, which changes value of existing
parameter.

MFC after: 1 week

17 years agoTry to avoid a possible infinite loop when parsing an invalid kernel dump file.
Craig Rodrigues [Thu, 25 Jan 2007 06:39:25 +0000 (06:39 +0000)]
Try to avoid a possible infinite loop when parsing an invalid kernel dump file.

PR: 108229
Submitted by: Jessica Han <jessicah juniper net>
Reviewed by: marcel
MFC after: 1 week

17 years agoFix for problems that occur when all mbuf clusters migrate to the mbuf packet
Mohan Srinivasan [Thu, 25 Jan 2007 01:05:23 +0000 (01:05 +0000)]
Fix for problems that occur when all mbuf clusters migrate to the mbuf packet
zone. Cluster allocations fail when this happens. Also processes that may have
blocked on cluster allocations will never be woken up. Thanks to rwatson for
an overview of the issue and pointers to the mbuma paper and his tool to dump
out UMA zones.

Reviewed by: andre@

17 years agoDisplay the name of the quota data file in verbose mode.
Mike Pritchard [Wed, 24 Jan 2007 22:52:32 +0000 (22:52 +0000)]
Display the name of the quota data file in verbose mode.

17 years agoFix for a bug where only one process (of multiple) blocked on
Mohan Srinivasan [Wed, 24 Jan 2007 22:49:11 +0000 (22:49 +0000)]
Fix for a bug where only one process (of multiple) blocked on
maxpages on a zone is woken up, with the rest never being woken up as
a result of the ZFLAG_FULL flag being cleared. Wakeup all such blocked
procsses instead. This change introduces a thundering herd, but since
this should be relatively infrequent, optimizing this (by introducing
a count of blocked processes, for example) may be premature.

Reviewd by: ups@