]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years ago - Remove an unneeded comma
delphij [Mon, 19 Mar 2007 07:49:11 +0000 (07:49 +0000)]
 - Remove an unneeded comma
 - Remove a redundant .Xc
 - Spell "Largely" as it should.

17 years ago- errno -> becomes error in sctp_output.c and sctputil.c
rrs [Mon, 19 Mar 2007 06:53:02 +0000 (06:53 +0000)]
- errno -> becomes error in sctp_output.c and sctputil.c
- SB_CLEAR macro defined and used for sb clearing.
- Fix for CMT express_sack_handling did not do proper
  pseudo-cumack updates.
- Get rid of extraneous function that was never used ip_2_ip6_hdr()
- Fixed source address selection bug (initialization problem).
- Source address selection debug added.

17 years agodisplay two new Intel feature bits
sam [Mon, 19 Mar 2007 05:23:42 +0000 (05:23 +0000)]
display two new Intel feature bits

Submitted by: "Rui Paulo" <rpaulo@gmail.com>
MFC after: 2 weeks

17 years agoCorrect a typo.
jkoshy [Mon, 19 Mar 2007 03:52:20 +0000 (03:52 +0000)]
Correct a typo.

Submitted by:  Kai Wang <kaiw27 at gmail dot com>

17 years agoPoll only while interrupt is disabled.
simokawa [Mon, 19 Mar 2007 03:41:27 +0000 (03:41 +0000)]
Poll only while interrupt is disabled.

MFC: 3 days after

17 years agoWait SCLK to be stable after LPS enabled.
simokawa [Mon, 19 Mar 2007 03:35:45 +0000 (03:35 +0000)]
Wait SCLK to be stable after LPS enabled.
This should fix NMI problem in fwphy_rddata().

PR: kern/94146 kern/100356
MFC: after 3 days

17 years agoBump __FreeBSD_version after changes to how insmntque(), getnewvnode(), and
kensmith [Mon, 19 Mar 2007 00:19:35 +0000 (00:19 +0000)]
Bump __FreeBSD_version after changes to how insmntque(), getnewvnode(), and
vfs_hash_insert() work.

17 years agoGive a chance for packet to appear with a correct input interfaces
rik [Sun, 18 Mar 2007 23:28:53 +0000 (23:28 +0000)]
Give a chance for packet to appear with a correct input interfaces
in case of multiple interfaces with the same MAC in the same bridge.
This commit do not solve the entire problem. Only case where packet
arrived from such interface.

PR:     kern/109815
MFC after:      7 days
Submitted by:   Eygene Ryabinkin and rik@
Discussed with: bms@, thompsa@, yar@

17 years agoIn IPv4 fast forwarding path, send ICMP unreachable messages for
bms [Sun, 18 Mar 2007 23:05:20 +0000 (23:05 +0000)]
In IPv4 fast forwarding path, send ICMP unreachable messages for
routes which have RTF_REJECT set *and* a zero expiry timer.

PR: kern/109246
MFC after: 10 days
Submitted by: Ingo Flaschberger

17 years agoMore corrections, this time from Wojciech Koszek
julian [Sun, 18 Mar 2007 19:28:44 +0000 (19:28 +0000)]
More corrections, this time from Wojciech Koszek

17 years agoSpelling and whitespace fixes from Dunne Whitty
julian [Sun, 18 Mar 2007 18:57:38 +0000 (18:57 +0000)]
Spelling and whitespace fixes from Dunne Whitty

17 years agoEnforce sample size alignment on first direct-copy channel.
ariff [Sun, 18 Mar 2007 18:26:40 +0000 (18:26 +0000)]
Enforce sample size alignment on first direct-copy channel.

17 years agoPrefer more traditional spellings of some words in comments.
rwatson [Sun, 18 Mar 2007 16:49:50 +0000 (16:49 +0000)]
Prefer more traditional spellings of some words in comments.

17 years agoPoint out that PIM is the way to go.
bms [Sun, 18 Mar 2007 15:34:57 +0000 (15:34 +0000)]
Point out that PIM is the way to go.

MFC after: 3 days

17 years agoDocument a major IGMP related annoyance.
bms [Sun, 18 Mar 2007 15:31:06 +0000 (15:31 +0000)]
Document a major IGMP related annoyance.
Add comments about IGMP bits which will go when we take SSM.
Xref multicast(4).

MFC after: 3 days

17 years agoMFi386: revision 1.650
nyan [Sun, 18 Mar 2007 03:25:40 +0000 (03:25 +0000)]
MFi386: revision 1.650

  Eliminate an unused parameter.

17 years agoDon't deny unmounting file systems for jailed processes immediately, allow
pjd [Sun, 18 Mar 2007 02:39:19 +0000 (02:39 +0000)]
Don't deny unmounting file systems for jailed processes immediately, allow
prison_priv_check() to decide what to do.

This change is suppose not to change current (security) behaviour
in any way.

This change is simlar to the change of PRIV_VFS_MOUNT in previous revision.

17 years agoDisable burst mode by default. Testing has shown that while it works on
njl [Sun, 18 Mar 2007 01:03:03 +0000 (01:03 +0000)]
Disable burst mode by default.  Testing has shown that while it works on
most systems, it causes the EC not to respond for some Acer and Compaq/HP
laptops.  This is the default value for Linux also.  For systems that need
it, burst mode can be enabled via the tunable/sysctl:
debug.acpi.ec.burst="1"

17 years agoExplain more about the "C" qualifier for Celsius.
njl [Sun, 18 Mar 2007 00:57:07 +0000 (00:57 +0000)]
Explain more about the "C" qualifier for Celsius.

MFC after: 3 days

17 years ago - Handle the case where slptime == runtime.
jeff [Sat, 17 Mar 2007 23:32:48 +0000 (23:32 +0000)]
 - Handle the case where slptime == runtime.

Submitted by: Atoine Brodin

17 years agoEliminate an unused parameter.
alc [Sat, 17 Mar 2007 19:42:06 +0000 (19:42 +0000)]
Eliminate an unused parameter.

17 years agoFix (another, more to come) ivar memory leak during driver detach.
ariff [Sat, 17 Mar 2007 19:37:09 +0000 (19:37 +0000)]
Fix (another, more to come) ivar memory leak during driver detach.

17 years agoRemove pathlen argument from write_entry function. It has never been used.
cperciva [Sat, 17 Mar 2007 19:18:29 +0000 (19:18 +0000)]
Remove pathlen argument from write_entry function.  It has never been used.

Approved by: kientzle
MFC after: 3 days

17 years ago - Turn all explicit giant acquires into conditional VFS_LOCK_GIANTs.
jeff [Sat, 17 Mar 2007 18:18:08 +0000 (18:18 +0000)]
 - Turn all explicit giant acquires into conditional VFS_LOCK_GIANTs.
   Only ops which used namei still remained.
 - Implement a scheme for reducing the overhead of tracking which vops
   require giant by constantly reducing the number of recursive giant
   acquires to one, leaving us with only one vfslocked variable.
 - Remove all NFSD lock acquisition and release from the individual nfs
   ops.  Careful examination has shown that they are not required.  This
   greatly simplifies the code.

Sponsored by: Isilon Systems, Inc.
Discussed with: rwatson
Tested by: kkenn
Approved by: re

17 years ago - Cast the intermediate value in priority computtion back down to
jeff [Sat, 17 Mar 2007 18:13:32 +0000 (18:13 +0000)]
 - Cast the intermediate value in priority computtion back down to
   unsigned char.  Weirdly, casting the 1 constant to u_char still produces
   a signed integer result that is then used in the % computation.  This
   avoids that mess all together and causes a 0 pri to turn into 255 % 64
   as we expect.

Reported by: kkenn (about 4 times, thanks)

17 years agoFix long delay closing/syncing issues on mmaped buffer.
ariff [Sat, 17 Mar 2007 17:07:21 +0000 (17:07 +0000)]
Fix long delay closing/syncing issues on mmaped buffer.

17 years agoUpdate postfix MTA selection code to point at postfix 2.3 which is the
simon [Sat, 17 Mar 2007 16:43:28 +0000 (16:43 +0000)]
Update postfix MTA selection code to point at postfix 2.3 which is the
current version.

PR: misc/110447
Submitted by: Chris Fletcher <fletch.AT.brightsparks.net.au>
MFC after: 1 week

17 years agoFix logic bug; we want to do_chdir if arg doesn't start with / _and_ it
cperciva [Sat, 17 Mar 2007 16:17:14 +0000 (16:17 +0000)]
Fix logic bug; we want to do_chdir if arg doesn't start with / _and_ it
doesn't start with @/ either.

This unbreaks "tar -c -C /no/such/directory @/path/to/archive".

MFC after: 3 days

17 years agoDon't forget to increment the raw_position (bytes written) counter, even
cperciva [Sat, 17 Mar 2007 14:59:04 +0000 (14:59 +0000)]
Don't forget to increment the raw_position (bytes written) counter, even
when operating in non-buffered mode.

Pointy hat to: cperciva
MFC after: 3 days

17 years agoUnbreak IPv6 after consolidation of TCP options insertion.
andre [Sat, 17 Mar 2007 11:52:54 +0000 (11:52 +0000)]
Unbreak IPv6 after consolidation of TCP options insertion.

Submitted by: tegge

17 years agoFix the most obvious of the bugs introduced by recent syncache changes
kmacy [Sat, 17 Mar 2007 06:40:09 +0000 (06:40 +0000)]
Fix the most obvious of the bugs introduced by recent syncache changes

- *ip is not initialized in the case of inet6 connection, but ip->ip_len is
  being changed anyway

Now the question is, why does it think an ipv4 connection is an ipv6 connection?
xemacs still doesn't work over X11 forwarding, but the kernel no longer panics.

17 years ago- Moved the uninitialized variables from the data to the bss section.
kato [Sat, 17 Mar 2007 05:30:03 +0000 (05:30 +0000)]
- Moved the uninitialized variables from the data to the bss section.
- Fixed typos in comment.

17 years agomove inline function above use so that -O works
kmacy [Sat, 17 Mar 2007 05:23:45 +0000 (05:23 +0000)]
move inline function above use so that -O works

17 years agoo Add ENVIRONMENT section and mention there that TMPDIR is ignored
maxim [Fri, 16 Mar 2007 21:46:24 +0000 (21:46 +0000)]
o Add ENVIRONMENT section and mention there that TMPDIR is ignored
when issetugid(3) is true.

PR: docs/108346
Obtained from: OpenBSD
MFC after: 1 week

17 years agoRevert/re-make previous commit in a manner that maintains hyphenation of
rwatson [Fri, 16 Mar 2007 19:18:49 +0000 (19:18 +0000)]
Revert/re-make previous commit in a manner that maintains hyphenation of
extended attributes.  I'm not sure I like it, but it is grammatically more
correct.

Requested by: mckusick

17 years ago[stage: 9/9]
ariff [Fri, 16 Mar 2007 17:19:03 +0000 (17:19 +0000)]
[stage: 9/9]

- SWAPLR quirk for (unknown, luckily it is mine) broken uaudio stick.
  Fixing by rewiring is impossible without damaging it. Luckily,
  we can fix it using "other" methods :) .
- Add uaudio_get_vendor(), _product() and _release() in uaudio.c
  (currently used by uaudio_pcm quirk).
- Implement CHANNEL_SETFRAGMENTS().
- Drop channel locking in few places where it is about to sleep
  somewhere. This should help eliminating illegal locking acquisition
  where the current thread is about to sleep, and also few deadlock
  cases. Dropping it right here is quite safe since it is already
  protected by CHN_F_BUSY flag and other threads won't bother to touch it.
  Solving other illegal locking issues are quite tricky without converting
  most usbd_do_request() calls to its equivalent _async() calls,
  which I intend to do it later after getting full test report from
  other people with different uaudio hardwares.
- Fix memory leak issues during detach. This seems common to any drivers
  (notably emu10kx, csapcm?) with bridge functions.

17 years ago[stage: 8/9]
ariff [Fri, 16 Mar 2007 17:18:17 +0000 (17:18 +0000)]
[stage: 8/9]

Implement CHANNEL_SETFRAGMENTS() for snd_atiixp, snd_es137x, snd_hda
and snd_via8233. CHANNEL_SETBLOCKSIZE() will basically call
CHANNEL_SETFRAGMENTS() internally using conservative blocksize /
blockcount hints. Other drivers will be converted later.

17 years ago[stage: 7/9]
ariff [Fri, 16 Mar 2007 17:17:25 +0000 (17:17 +0000)]
[stage: 7/9]

EWOULDBLOCK -> EAGAIN.

17 years ago[stage: 6/9]
ariff [Fri, 16 Mar 2007 17:16:56 +0000 (17:16 +0000)]
[stage: 6/9]

- Disable stray buffer management, since sample size aligned buffering
  are pretty much guaranteed through out the entire feeder_* chain
  processes.
- Few style(9) cleanups.

17 years ago[stage: 5/9]
ariff [Fri, 16 Mar 2007 17:16:24 +0000 (17:16 +0000)]
[stage: 5/9]

channel.c/channel_if.m:
 - Macros cleanups, prefer inlined min() over MIN().
 - Rework chn_read()/chn_write() for better dead interrupt detection
   policy. Reduce scheduling overhead by doing pure 5 seconds sleep
   before giving up, instead of several cycle of brute micro sleeping.
 - Avoid calling wakeup_one() for non-sleeping channel (for example,
   vchan parent channel).
 - EWOULDBLOCK -> EAGAIN.
 - Fix possible divide-by-zero panic on chn_sync().
 - Re-enforce ^2 blocksize policy, since there are too many broken
   userland apps that blindly assume it without even trying to do
   serious calculations.
 - New channel method - CHANNEL_SETFRAGMENTS(), a refined version of
   CHANNEL_SETBLOCKSIZE(). It accept _both_ blocksize and blockcount
   arguments, so the driver internals will have better hints for
   buffering and timing calculations.
 - Hook FEEDER_SWAPLR into feederchain building process.

feeder_fmt.c:
- Unified version of various filters, avoiding duplications.
- malloc()less feeder_fmt. Informations can be retrieved dynamically
  by doing table lookup on static data. For cases such as converting
  from stereo to mono or reducing bit depth where input data is larger
  than output, cycle remaining available free space until it has been
  exhausted and start kicking 8 bytes reservoir space from there to
  complete the remaining requested count.
- Introduce FEEDER_SWAPLR. Few super broken hardwares (found on several
  extremely cheap uaudio stick, possibly others) mistakenly wired left
  and right channels wrongly, screwing output or input.

17 years ago[stage: 4/9]
ariff [Fri, 16 Mar 2007 17:15:33 +0000 (17:15 +0000)]
[stage: 4/9]

- Rearrange FEEDER_* constants starting from 0 to 31, so the future
  additions will be much easier and consistent.
- Introduce FEEDER_SWAPLR. Few super broken hardwares (found on several
  extremely cheap uaudio stick, possibly others) mistakenly wired left
  and right channels wrongly, screwing output or input.

17 years ago[stage: 3.2/9]
ariff [Fri, 16 Mar 2007 17:14:41 +0000 (17:14 +0000)]
[stage: 3.2/9]

malloc()less feeder_vchan. Informations can be retrieved dynamically
by doing table lookup on static data. Reduce mixing overhead by
doing direct copy on first channel. Mixing process will begin starting
from second channel onwards.

17 years ago[stage: 3.1/9]
ariff [Fri, 16 Mar 2007 17:14:19 +0000 (17:14 +0000)]
[stage: 3.1/9]

malloc()less feeder_volume. Informations can be retrieved dynamically
by doing table lookup on static data. Increase resolution from 6bit
to PCM_FXSHIFT (8bit) for better resolution and finer volume changes.

17 years ago[stage: 2/9]
ariff [Fri, 16 Mar 2007 17:13:44 +0000 (17:13 +0000)]
[stage: 2/9]

Use inlined min() rather than MIN() macross.

17 years ago[stage: 1/9]
ariff [Fri, 16 Mar 2007 17:13:12 +0000 (17:13 +0000)]
[stage: 1/9]

- Convert sx lock to plain mutex. Since the access of /dev/sndstat
  is pretty much exclusive and protected by toggling sndstat_isopen,
  plain mutex is more than enough.
- Enable SBUF_AUTOEXTEND to avoid buffer truncation.

17 years agoRemove an extra "The".
bmah [Fri, 16 Mar 2007 16:07:49 +0000 (16:07 +0000)]
Remove an extra "The".

MFC after: 3 days

17 years agoRemove unused and #if 0'd net.inet.tcp.tcp_rttdflt sysctl.
rwatson [Fri, 16 Mar 2007 13:42:26 +0000 (13:42 +0000)]
Remove unused and #if 0'd net.inet.tcp.tcp_rttdflt sysctl.

17 years agoMinor white space tweaks in comments.
rwatson [Fri, 16 Mar 2007 13:39:04 +0000 (13:39 +0000)]
Minor white space tweaks in comments.

17 years ago- Add cxgb(4) entry, remove stale wx(4) entry
brueffer [Fri, 16 Mar 2007 12:56:20 +0000 (12:56 +0000)]
- Add cxgb(4) entry, remove stale wx(4) entry
- (gigabit|fast) ethernet -> (Gigabit|Fast) Ethernet

Approved by: rwatson (mentor)
MFC after: 3 days

17 years agoImagine a situation where:
pjd [Fri, 16 Mar 2007 12:36:54 +0000 (12:36 +0000)]
Imagine a situation where:

# ls -ld /mnt/{foo,bar}
drwxr-xr-x  3 root  wheel  512 Mar 16 06:56 /mnt/bar
lrwxr-xr-x  1 root  wheel    3 Mar 16 12:10 /mnt/foo -> bar

# grep /mnt/foo /etc/fstab
/dev/da1 /mnt/foo ufs rw 0 0

Which means, we give symbolic link as a mount point to mount(8), but mount(8)
use realpath(3) before mounting the file systems, so we get:

# mount | grep /dev/da1
/dev/da1 on /mnt/bar (ufs, local)

Before the commit:

# snapinfo /mnt/foo
usage: snapinfo [-v] -a
       snapinfo [-v] mountpoint
# snapinfo /mnt/bar
/mnt/bar/snap

This commit makes snapinfo(8) to first realpath(3) the given mount point and
now we have:

# snapinfo /mnt/foo
/mnt/bar/snap
# snapinfo /mnt/bar
/mnt/bar/snap

17 years agoReplace xfer->act.hand with xfer->hand.
simokawa [Fri, 16 Mar 2007 05:39:33 +0000 (05:39 +0000)]
Replace xfer->act.hand with xfer->hand.

17 years agoRemove retry_count.
simokawa [Fri, 16 Mar 2007 05:17:23 +0000 (05:17 +0000)]
Remove retry_count.

17 years ago* Remove xfer->retry_req.
simokawa [Fri, 16 Mar 2007 05:11:42 +0000 (05:11 +0000)]
* Remove xfer->retry_req.
   It is unnecessary because retry is done by OHCI.
   Further retry should be done by applications.

17 years agoLess verbose debug messages.
simokawa [Fri, 16 Mar 2007 04:26:58 +0000 (04:26 +0000)]
Less verbose debug messages.

17 years agoDetect cycle lost.
simokawa [Fri, 16 Mar 2007 04:25:02 +0000 (04:25 +0000)]
Detect cycle lost.

17 years agoMention a limitation that was inherted from RFC1952, making
delphij [Fri, 16 Mar 2007 03:50:53 +0000 (03:50 +0000)]
Mention a limitation that was inherted from RFC1952, making
it impossible to obtain correct file size from a file that
is larger than 4GB before compression.

PR: bin/110329
MFC after: 1 week

17 years agoPass special device to the ufs_disk_fillout() function, instead of mount
pjd [Fri, 16 Mar 2007 03:23:32 +0000 (03:23 +0000)]
Pass special device to the ufs_disk_fillout() function, instead of mount
point path. This way we properly handle the case when file system listed
in /etc/fstab was unmounted and another file system was mounted on the
same mount point.

17 years agoThe ufs_disk_fillout(3) can take special device name (with or without /dev/
pjd [Fri, 16 Mar 2007 03:13:28 +0000 (03:13 +0000)]
The ufs_disk_fillout(3) can take special device name (with or without /dev/
prefix) as an argument and mount point path. At the end it has to find
device name file system is stored on, which means when mount point path is
given, it tries to look into /etc/fstab and find special device
corresponding to the given mount point. This is not perfect, because it
doesn't handle the case when file system is mounted by hand and mount point
is given as an argument.

I found this problem while trying to use snapinfo(8), which passes mount
points to the ufs_disk_fillout(3) function, but I had file system mounted
manually, so snapinfo(8) was exiting with the error below:

ufs_disk_fillout: No such file or directory

I modified libufs(3) to handle those arguments (the order is important):

1. special device with /dev/ prefix
2. special device without /dev/ prefix
3. mount point listed in /etc/fstab, directory exists
4. mount point listed in /etc/fstab, directory doesn't exist
5. mount point of a file system mounted by hand

17 years agoPrint warning for large DFLTPHYS.
simokawa [Fri, 16 Mar 2007 02:29:36 +0000 (02:29 +0000)]
Print warning for large DFLTPHYS.

17 years agoFix support for ASUS A7T ALC882 laptop (gpio0 quirk).
ariff [Fri, 16 Mar 2007 01:44:21 +0000 (01:44 +0000)]
Fix support for ASUS A7T ALC882 laptop (gpio0 quirk).

Reported/Tested by: cognet

17 years agoSupport MAXPHYS up to 512KB
simokawa [Fri, 16 Mar 2007 01:23:36 +0000 (01:23 +0000)]
Support MAXPHYS up to 512KB

- We need at least two OCBs with indirect pointers allocated in a 4KB page.
- SBP_MAXPHYS can increase to 1MB once we separate management OCB/ORB
  which usually does not need indirect pointers.
- We have to increase SBP_DMA_SIZE for MAXPHYS larger than 1MB.

MFC after: 3 days

17 years agoJust use 'fdrop()' instead of 'FILE_LOCK(); fdrop_locked()' in
jhb [Thu, 15 Mar 2007 21:19:21 +0000 (21:19 +0000)]
Just use 'fdrop()' instead of 'FILE_LOCK(); fdrop_locked()' in
dupfdopen().  While I'm at it, move the second fdrop() out from under the
filedesc lock.

17 years ago- Bring upgrade produce up-to-date for OpenSSL 0.9.8e.
simon [Thu, 15 Mar 2007 21:06:48 +0000 (21:06 +0000)]
- Bring upgrade produce up-to-date for OpenSSL 0.9.8e.
- Add reminder to bump version numer in Makefile.inc.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r167617,
simon [Thu, 15 Mar 2007 20:26:26 +0000 (20:26 +0000)]
This commit was generated by cvs2svn to compensate for changes in r167617,
which included commits to RCS files with non-trunk default branches.

17 years agoImport fix from upstream OpenSSL_0_9_8-stable branch:
simon [Thu, 15 Mar 2007 20:26:26 +0000 (20:26 +0000)]
Import fix from upstream OpenSSL_0_9_8-stable branch:

EVP_CIPHER_CTX_key_length() should return the set key length
in the EVP_CIPHER_CTX structure which may not be the same as
the underlying cipher key length for variable length ciphers.

This fixes problems in OpenSSH using some ciphers, and possibly other
applications.

See also: http://bugzilla.mindrot.org/show_bug.cgi?id=1291

17 years agoUpgrade to OpenSSL 0.9.8e.
simon [Thu, 15 Mar 2007 20:15:15 +0000 (20:15 +0000)]
Upgrade to OpenSSL 0.9.8e.

17 years agoResolve conflicts after import of OpenSSL 0.9.8e.
simon [Thu, 15 Mar 2007 20:07:27 +0000 (20:07 +0000)]
Resolve conflicts after import of OpenSSL 0.9.8e.

17 years agoVendor import of OpenSSL 0.9.8e.
simon [Thu, 15 Mar 2007 20:03:30 +0000 (20:03 +0000)]
Vendor import of OpenSSL 0.9.8e.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r167612,
simon [Thu, 15 Mar 2007 20:03:30 +0000 (20:03 +0000)]
This commit was generated by cvs2svn to compensate for changes in r167612,
which included commits to RCS files with non-trunk default branches.

17 years agoSpring cleanup on irrelevant NULL checking over M_WAITOK allocations.
ariff [Thu, 15 Mar 2007 18:19:01 +0000 (18:19 +0000)]
Spring cleanup on irrelevant NULL checking over M_WAITOK allocations.

17 years ago- Put some sanity break statement in few missing places.
ariff [Thu, 15 Mar 2007 17:35:04 +0000 (17:35 +0000)]
- Put some sanity break statement in few missing places.
- Remove NULL checking on snd_mtxcreate() (M_WAITOK) .

17 years agoNOOP (for now) for hdac_dma_nocache(). It is a wrong way to enforce
ariff [Thu, 15 Mar 2007 17:23:38 +0000 (17:23 +0000)]
NOOP (for now) for hdac_dma_nocache(). It is a wrong way to enforce
cache coherency, besides of causing train wreck in other places
(especially on amd64, possibly on i386).

Discussed with: kib@, rafan@
Tested by: rafan@

17 years agoFix severe out-of-bound mtx "type" pointer, causing WITNESS refcount
ariff [Thu, 15 Mar 2007 16:41:27 +0000 (16:41 +0000)]
Fix severe out-of-bound mtx "type" pointer, causing WITNESS refcount
confusions and panic provided that the following conditions are met:

  1) WITNESS is enabled (watch/trace).
  2) Using modules, instead of statically linked (Not a strict
     requirement, but easier to reproduce this way).
  3) 2 or more modules share the same mtx type ("sound softc").
     - They might share the same name (strcmp() == 0), but it always
       point to different address.
  4) Repetitive kldunload/load on any module that shares the same mtx
     type (Not a strict requirement, but easier to reproduce this way).

     Consider module A and module B:
     - From enroll() - subr_witness.c:
       * Load module A. Everything seems fine right now.
         wA-w_refcount == 1 ; wA-w_name = "sound softc"
       * Load module B.
       * w->w_name == description will always fail.
         ("sound softc" from A and B point to different address).
       * wA->w_refcount > 0 && strcmp(description, wA->w_name) == 0
       * enroll() will return wA instead of returning (possibly unique)
         wB.
         wA->w_refcount++ , == 2.
       * Unload module A, mtx_destroy(), wA->w_name become invalid,
         but wA->w_refcount-- become 1 instead of 0. wA will not be
         removed from witness list.
       * Some other places call mtx_init(), iterating witness list,
         found wA, failed on wA->w_name == description
       * wA->w_refcount > 0 && strcmp(description, wA->w_name)
       * Panic on strcmp() since wA->w_name no longer point to valid
         address.

Note that this could happened in other places as well, not just sound
(eg. consider lots of drivers that share simmilar MTX_NETWORK_LOCK).

Solutions (for sound case):
  1) Provide unique mtx type string for each mutex creation (chosen)
  or
  2) Put "sound softc" global variable somewhere and use it.

17 years agoAdd missing @
brueffer [Thu, 15 Mar 2007 16:12:08 +0000 (16:12 +0000)]
Add missing @

17 years agoConsolidate insertion of TCP options into a segment from within tcp_output()
andre [Thu, 15 Mar 2007 15:59:28 +0000 (15:59 +0000)]
Consolidate insertion of TCP options into a segment from within tcp_output()
and syncache_respond() into its own generic function tcp_addoptions().

tcp_addoptions() is alignment agnostic and does optimal packing in all cases.

In struct tcpopt rename to_requested_s_scale to just to_wscale.

Add a comment with quote from RFC1323: "The Window field in a SYN (i.e.,
a <SYN> or <SYN,ACK>) segment itself is never scaled."

Reviewed by: silby, mohans, julian
Sponsored by: TCP/IP Optimization Fundraise 2005

17 years agoRemove NULL allocation checking since malloc() is allow to wait.
ariff [Thu, 15 Mar 2007 15:06:54 +0000 (15:06 +0000)]
Remove NULL allocation checking since malloc() is allow to wait.
(I'll fix other places later..)

17 years agoNULL instead of 0 in mtx_init() .
ariff [Thu, 15 Mar 2007 14:57:54 +0000 (14:57 +0000)]
NULL instead of 0 in mtx_init() .

17 years agoAdd a description about hw.firewire.hold_count.
simokawa [Thu, 15 Mar 2007 14:44:03 +0000 (14:44 +0000)]
Add a description about hw.firewire.hold_count.

MFC: after 3 days
PR: kern/93083

17 years agoDon't mess with PCIM_CMD_SERRESPEN and PCIM_CMD_PERRESPEN.
simokawa [Thu, 15 Mar 2007 14:11:46 +0000 (14:11 +0000)]
Don't mess with PCIM_CMD_SERRESPEN and PCIM_CMD_PERRESPEN.
This will fix 'NMI RAM parity error' while booting on some machines.

PR: kern/95077
MFC after: 3 days

17 years agoRemove a spurious blank line at the start of vlan_growhash().
yar [Thu, 15 Mar 2007 14:10:51 +0000 (14:10 +0000)]
Remove a spurious blank line at the start of vlan_growhash().
Add a diagnostic message to the function about resizing vlan
hash table.

17 years agoMention the first RELENG_6 release to include this driver.
brueffer [Thu, 15 Mar 2007 13:08:50 +0000 (13:08 +0000)]
Mention the first RELENG_6 release to include this driver.

MFC after: 3 days

17 years ago- Sysctl's move to seperate file
rrs [Thu, 15 Mar 2007 11:27:14 +0000 (11:27 +0000)]
- Sysctl's move to seperate file
- moved away from ifn/ifa access to sctp_ifa/sctp_ifn
  built and managed by the add-ip code.
- cleaned up add-ip code to use the iterator
- made iterator be a thread, which enables auto-asconf now.
- rewrote and cleaned up source address selection (also
  made it use new structures).
- Fixed a couple of memory leaks.
- DACK now settable as to how many packets to delay as
  well as time.
- connectx() to latest socket API, new associd arg.
- Fixed issue with revoking and loosing potential to
  send when we inflate the flight size. We now inflate
  the cwnd too and deflate it later when the revoked
  chunk is sent or acked.
- Got rid of some temp debug code
- src addr selection moved to a common file (sctp_output.c)
- Support for simple VRF's (we have support for multi-vfr
  via compile switch that is scrubbed from BSD but we won't
  need multi-vrf until we first get VRF :-D)
- Rest of mib work for address information now done
- Limit number of addresses in INIT/INIT-ACK to
  a #def (30).

Reviewed by: gnn

17 years agoFix a comment in memstat.h: errors are associated with memory type lists,
rwatson [Thu, 15 Mar 2007 10:44:18 +0000 (10:44 +0000)]
Fix a comment in memstat.h: errors are associated with memory type lists,
not individual types.

Submitted by: Bryan Venteicher <bryanv at daemoninthecloset dot org>
MFC after: 3 days

17 years agoDon't consider an lstat(2) failure to be an error (in the sense of
cperciva [Thu, 15 Mar 2007 10:11:38 +0000 (10:11 +0000)]
Don't consider an lstat(2) failure to be an error (in the sense of
affecting the return value from bsdtar), since (a) it usually occurs
due to a perfectly innocent (and unavoidable) race condition where a
user deletes a file in the window between bsdtar reading a directory
and attempting to read the file; and (b) aside from printing a warning
message, bsdtar behaves exactly as if the file had been deleted prior
to bsdtar reading its parent directory.

Reviewed by: kientzle
MFC after: 6 days

17 years agoIn auth_script(), change the size of an array to match the number of
kevlo [Thu, 15 Mar 2007 09:16:54 +0000 (09:16 +0000)]
In auth_script(), change the size of an array to match the number of
arguments.

17 years agoDiff reduction with NetBSD; use IN_LOCAL_GROUP() to check if an address
bms [Thu, 15 Mar 2007 08:44:22 +0000 (08:44 +0000)]
Diff reduction with NetBSD; use IN_LOCAL_GROUP() to check if an address
is within the locally scoped multicast range 224.0.0.0/24.

17 years agoA silly buglet found by Coverity- check the return value from
mjacob [Thu, 15 Mar 2007 06:42:54 +0000 (06:42 +0000)]
A silly buglet found by Coverity- check the return value from
cam_periph_acquire.

17 years agofix include names
kmacy [Thu, 15 Mar 2007 04:09:40 +0000 (04:09 +0000)]
fix include names

17 years agoRemove vestiges of very specific fpga support for my company's board.
imp [Thu, 15 Mar 2007 03:31:49 +0000 (03:31 +0000)]
Remove vestiges of very specific fpga support for my company's board.
It isn't relevant to FreeBSD as a whole, breaks the build, and isn't
even needed for my company's boards anymore...

MFC After: 2 weeks

17 years agoAdd "force" to ext2_ops, to match what was in the old mount_ext2fs binary.
rodrigc [Thu, 15 Mar 2007 00:09:50 +0000 (00:09 +0000)]
Add "force" to ext2_ops, to match what was in the old mount_ext2fs binary.

Reported by: Ivan Voras <ivoras fer hr>

17 years agoDocument exact command preferred for sending ASL as part of bug reports.
njl [Wed, 14 Mar 2007 22:55:30 +0000 (22:55 +0000)]
Document exact command preferred for sending ASL as part of bug reports.

17 years agoCreate an identity mapping (V=P) super page for the low memory region on
njl [Wed, 14 Mar 2007 22:30:02 +0000 (22:30 +0000)]
Create an identity mapping (V=P) super page for the low memory region on
boot.  Then, just switch to the kernel pmap when suspending instead of
allocating/freeing our own mapping every time.  This should solve a panic
of pmap_remove() being called with interrupts disabled.  Thanks to Alan
Cox for developing this patch.

Note: this means that ACPI requires super page (PG_PS) support in the CPU.
This has been present since the Pentium and first documented in the
Pentium Pro.  However, it may need to be revisited later.

Submitted by: alc
MFC after: 1 month

17 years agoNew release notes: vge(4) altq support (+mfc), snd_hda(4) added
brueffer [Wed, 14 Mar 2007 22:28:41 +0000 (22:28 +0000)]
New release notes: vge(4) altq support (+mfc), snd_hda(4) added

Updated release notes: KDE updated to 3.5.6, GNOME updated to 2.16.3

Also moved cxgb(4) entry to the right place and fixed another entity.

17 years agoNew release notes: cxgb(4) added.
brueffer [Wed, 14 Mar 2007 21:31:53 +0000 (21:31 +0000)]
New release notes: cxgb(4) added.

While here, fix an entity.

17 years agoProperly move the setting of bstp_linkstate_p to the bridgestp module.
thompsa [Wed, 14 Mar 2007 20:55:51 +0000 (20:55 +0000)]
Properly move the setting of bstp_linkstate_p to the bridgestp module.

17 years agoFix markup and reword a sentence to actually make sense.
brueffer [Wed, 14 Mar 2007 20:09:23 +0000 (20:09 +0000)]
Fix markup and reword a sentence to actually make sense.

17 years agoOnly enter the debugger on a Fatal op if this is a debug build of the
njl [Wed, 14 Mar 2007 19:56:10 +0000 (19:56 +0000)]
Only enter the debugger on a Fatal op if this is a debug build of the
acpi module.  Also clean up print of args a little.

This was accidentally committed as 1.9.2.3 in the stable branch.  Since it
is harmless, I will let the "insta-MFC" stand unless there is a problem.

17 years agoUpdate a comment: Rather than suggesting suser(), suggest priv(9) for
rwatson [Wed, 14 Mar 2007 19:52:19 +0000 (19:52 +0000)]
Update a comment: Rather than suggesting suser(), suggest priv(9) for
checking privilege.

17 years agoMap the second CS of the compact flash too. This allow us access to
jhay [Wed, 14 Mar 2007 19:03:07 +0000 (19:03 +0000)]
Map the second CS of the compact flash too. This allow us access to
the alternate status and the control registers. Remove the local
version of ata_reset.

Add support for the ADI Pronghorn Metro boards. They use CS3 and CS4
instead of Avila's CS1 and CS2.

17 years agoRename sed.test to multitest.t to comply with the naming scheme of
dds [Wed, 14 Mar 2007 18:58:59 +0000 (18:58 +0000)]
Rename sed.test to multitest.t to comply with the naming scheme of
the regression suite.