]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
16 years agoUse a uint16_t type for the vlan tag rather an int.
thompsa [Thu, 18 Oct 2007 21:52:31 +0000 (21:52 +0000)]
Use a uint16_t type for the vlan tag rather an int.

16 years agoPullup IPFilter 4.1.28 from the vendor branch into HEAD.
darrenr [Thu, 18 Oct 2007 21:52:14 +0000 (21:52 +0000)]
Pullup IPFilter 4.1.28 from the vendor branch into HEAD.

MFC after: 7 days

16 years agoUnbreak arm build by removing duplicate symbols.
cognet [Thu, 18 Oct 2007 21:44:21 +0000 (21:44 +0000)]
Unbreak arm build by removing duplicate symbols.

16 years agoThis commit was generated by cvs2svn to compensate for changes in r172771,
darrenr [Thu, 18 Oct 2007 21:42:51 +0000 (21:42 +0000)]
This commit was generated by cvs2svn to compensate for changes in r172771,
which included commits to RCS files with non-trunk default branches.

16 years agoImport IPFilter 4.1.28
darrenr [Thu, 18 Oct 2007 21:42:51 +0000 (21:42 +0000)]
Import IPFilter 4.1.28

16 years agoImport IPFilter 4.1.28
darrenr [Thu, 18 Oct 2007 21:42:51 +0000 (21:42 +0000)]
Import IPFilter 4.1.28

16 years agoThe bridging output function puts the mbuf directly on the interfaces send
thompsa [Thu, 18 Oct 2007 21:22:15 +0000 (21:22 +0000)]
The bridging output function puts the mbuf directly on the interfaces send
queue so the output network card must support the same tagging mechanism as
how the frame was input (prepended Ethernet header tag or stripped HW mflag).

Now the vlan Ethernet header is _always_ stripped in ether_input and the mbuf
flagged, only only network cards with VLAN_HWTAGGING enabled would properly
re-tag any outgoing vlan frames.

If the outgoing interface does not support hardware tagging then readd the vlan
header to the front of the frame. Move the common vlan encapsulation in to
ether_vlanencap().

Reported by: Erik Osterholm, Jon Otterholm
MFC after: 1 week

16 years agoThis commit was generated by cvs2svn to compensate for changes in r172767,
simon [Thu, 18 Oct 2007 20:19:33 +0000 (20:19 +0000)]
This commit was generated by cvs2svn to compensate for changes in r172767,
which included commits to RCS files with non-trunk default branches.

16 years agoImport DTLS security fix from upstream OpenSSL_0_9_8-stable branch.
simon [Thu, 18 Oct 2007 20:19:33 +0000 (20:19 +0000)]
Import DTLS security fix from upstream OpenSSL_0_9_8-stable branch.

From the OpenSSL advisory:

Andy Polyakov discovered a flaw in OpenSSL's DTLS
implementation which could lead to the compromise of clients
and servers with DTLS enabled.

DTLS is a datagram variant of TLS specified in RFC 4347 first
supported in OpenSSL version 0.9.8. Note that the
vulnerabilities do not affect SSL and TLS so only clients and
servers explicitly using DTLS are affected.

We believe this flaw will permit remote code execution.

Security: CVE-2007-4995
Security: http://www.openssl.org/news/secadv_20071012.txt

16 years agoRemove _FREEFALL_CONFIG hacks
peter [Thu, 18 Oct 2007 19:42:50 +0000 (19:42 +0000)]
Remove _FREEFALL_CONFIG hacks

16 years agoRemove _FREEFALL_CONFIG hacks. su+pam_ksu works well enough to use on
peter [Thu, 18 Oct 2007 19:36:31 +0000 (19:36 +0000)]
Remove _FREEFALL_CONFIG hacks.  su+pam_ksu works well enough to use on
the freebsd.org cluster.

16 years agoForced commit to signify that previous commit was:
grehan [Thu, 18 Oct 2007 18:58:07 +0000 (18:58 +0000)]
Forced commit to signify that previous commit was:

Reviewed: yar@

16 years agoTeach /etc/rc.d/ppp how to start/stop individual instances
emax [Thu, 18 Oct 2007 17:10:40 +0000 (17:10 +0000)]
Teach /etc/rc.d/ppp how to start/stop individual instances
of ppp. This is an extension of previous commit.

Submitted by: Yuri Kurenkov < y dot kurenkov at init dot ru >
Reviewed by: mtm
MFC after: 3 days

16 years agoAdd a -z flag to nfsstat which zeros the NFS statistics after displaying
jhb [Thu, 18 Oct 2007 16:38:07 +0000 (16:38 +0000)]
Add a -z flag to nfsstat which zeros the NFS statistics after displaying
them.

MFC after: 1 week
Requested by: ps
Submitted by: ps (6 years ago)

16 years agoAdd noclusterr and noclusterw options to the options list. I forgot these
bde [Thu, 18 Oct 2007 16:25:47 +0000 (16:25 +0000)]
Add noclusterr and noclusterw options to the options list.  I forgot these
when I implemented clustering.

16 years agoFix some style bugs in the mount options list. Mainly, sort the list,
bde [Thu, 18 Oct 2007 15:48:10 +0000 (15:48 +0000)]
Fix some style bugs in the mount options list.  Mainly, sort the list,
leaving space for adding missing options.  Negative options are sorted
after removing their "no" prefix, and generic options are sorted before
msdosfs-specific ones.

16 years agoExplicitly verify if we have found the VERSION_GEN script
yar [Thu, 18 Oct 2007 15:21:35 +0000 (15:21 +0000)]
Explicitly verify if we have found the VERSION_GEN script
instead of passing the possibly null argument to awk(1) and
getting an obscure error from it.

16 years agoTeach makewhatis.c to treat .SS with the same severity as .SH.
edwin [Thu, 18 Oct 2007 12:53:11 +0000 (12:53 +0000)]
Teach makewhatis.c to treat .SS with the same severity as .SH.

At least one port (net-mgmt/net-snmp) creates man-pages which are
in the format:

.SH NAME
The Net-SNMP agent \- The snmp agent responds to SNMP queries from management stations.

.PP
.SS "Modules"

At this moment, makewhatis determines the end of the .SH NAME section
as where it finds .SH again, but there is none here, is it "terminated"
by the .SS.

PR: bin/116706
Submitted by: edwin@
Approved by: re (Ken Smith), grog (mentor)
MFC after: 1 week

16 years agoThe fork symbols aren't MD, they already live in sys/.
yar [Thu, 18 Oct 2007 11:28:38 +0000 (11:28 +0000)]
The fork symbols aren't MD, they already live in sys/.

Found by: version_gen.awk
Tested by: md5(1) (libc.so hasn't changed at all)

16 years agoThe upper while loop has already recycled child process, so the if
davidxu [Thu, 18 Oct 2007 11:05:30 +0000 (11:05 +0000)]
The upper while loop has already recycled child process, so the if
statement has never executed as expected, fix it.

MFC after: 3 days

16 years agoAdd Winchiphead (or Nanjin QinHeng Electronics) USB Serial converter driver.
takawata [Thu, 18 Oct 2007 10:51:06 +0000 (10:51 +0000)]
Add Winchiphead (or Nanjin QinHeng Electronics) USB Serial converter driver.
I tested it on cheap serial converter.(Only costs 980JPY.)

16 years agoDistinguish the cases when a dup symbol is in different versions
yar [Thu, 18 Oct 2007 10:32:28 +0000 (10:32 +0000)]
Distinguish the cases when a dup symbol is in different versions
(evil, needs a error) or in the same version (harmless, deserves
a mere warning).

Noticed by:     grehan@, tinderbox

16 years agoRemoving obsolete etc/cached.conf.
bushman [Thu, 18 Oct 2007 09:09:22 +0000 (09:09 +0000)]
Removing obsolete etc/cached.conf.

Approved by: brooks (mentor)

16 years agoDo not bail out for cleanworld target just because
delphij [Thu, 18 Oct 2007 08:41:52 +0000 (08:41 +0000)]
Do not bail out for cleanworld target just because
chflags is failed.

16 years agoForced commit to note cached.conf -> nscd.conf repocopy. etc/Makefile
bushman [Thu, 18 Oct 2007 08:26:20 +0000 (08:26 +0000)]
Forced commit to note cached.conf -> nscd.conf repocopy. etc/Makefile
changed accordingly.

Approved by: brooks (mentor)

16 years agoIn msdosfs_settattr(), don't do synchronous updates of the denode
bde [Thu, 18 Oct 2007 07:26:21 +0000 (07:26 +0000)]
In msdosfs_settattr(), don't do synchronous updates of the denode
(except indirectly for the size pseudo-attribute).  If anything deserves
a sync update, then it is ids and immutable flags, since these are
related to security, but ffs never synced these and msdosfs doesn't
support them.  (ufs_setattr() only does an update in one case where
it is least needed (for timestamps); it did pessimal sync updates for
timestamps until 1998/03/08 but was changed for unlogged reasons related
to soft updates.)

Now msdosfs calls deupdat() with waitfor == 0, which normally gives a
delayed update to disk but always gives a sync update of timestamps
in core, while for ffs everything is delayed until the syncer daemon
or other activity causes an update (except for timestamps).

This gives a large optimization mainly for things like cp -p, where
attribute adjustment could easily triple the number of physical I/O's
if it is done synchronously (but cp -p to msdosfs is not as bad as
that, since msdosfs doesn't support many attributes so null adjustments
are more common, and msdosfs doesn't support ctimes so even if cp
doesn't weed out null adjustments they don't become non-null after
clobbering the ctime).

16 years agoFix build from errors exposed with recent version_gen.awk commit.
grehan [Thu, 18 Oct 2007 07:23:31 +0000 (07:23 +0000)]
Fix build from errors exposed with recent version_gen.awk commit.
Not quite sure if this is 100% correct: awaiting review. But quieten
tinderbox in the meantime.

16 years agocorrect guard variable names.
imp [Thu, 18 Oct 2007 05:43:44 +0000 (05:43 +0000)]
correct guard variable names.

16 years agoMerge support from p4 (from NetBSD) for arm9e and arm10, arm11 cores. Not
imp [Thu, 18 Oct 2007 05:33:06 +0000 (05:33 +0000)]
Merge support from p4 (from NetBSD) for arm9e and arm10, arm11 cores.  Not
yet connected to the build, but reduces diffs to p4 repo.

Obtained from: NetBSD

16 years agoMerge definitions for ARM9E, ARM10 and ARM11 processors from p4 (which
imp [Thu, 18 Oct 2007 05:06:58 +0000 (05:06 +0000)]
Merge definitions for ARM9E, ARM10 and ARM11 processors from p4 (which
got them from NetBSD).

16 years ago- Correctly define CACHED_SOCKET_PATH as /var/run/nscd after cached to
tmclaugh [Wed, 17 Oct 2007 23:20:49 +0000 (23:20 +0000)]
- Correctly define CACHED_SOCKET_PATH as /var/run/nscd after cached to
  nscd renaming.

Approved by: mux

16 years agoMFp4:
yar [Wed, 17 Oct 2007 20:09:56 +0000 (20:09 +0000)]
MFp4:

- Check for duplicated symbols and suggest moving them to ObsoleteVersions.
- Improve and unify error handling.
- Make the regular expressions more uniform, robust, and less sensitive
  to harmless variations in the input such as those to whitespace amount.

Reviewed by: deischen
Tested with: md5 (Version.map files in /usr/obj stay the same)

16 years agoDelete an extra backslash at the end of the xterm-basic entry.
yar [Wed, 17 Oct 2007 19:58:50 +0000 (19:58 +0000)]
Delete an extra backslash at the end of the xterm-basic entry.
It can be missed easily that the following blank line formally
belongs to the xterm-basic entry due to the unneeded backslash.

PR: bin/80256 (audit trail)

16 years agoMention that less was upgraded to 409.
delphij [Wed, 17 Oct 2007 17:52:31 +0000 (17:52 +0000)]
Mention that less was upgraded to 409.

16 years agoAdd whitespace.
krion [Wed, 17 Oct 2007 13:47:10 +0000 (13:47 +0000)]
Add whitespace.

16 years agoUpdate for INDEX-8 support.
krion [Wed, 17 Oct 2007 13:45:43 +0000 (13:45 +0000)]
Update for INDEX-8 support.

16 years agoPrevent strange crashes in fmt with absurd goal lengths introduced by the
jmallett [Wed, 17 Oct 2007 11:41:40 +0000 (11:41 +0000)]
Prevent strange crashes in fmt with absurd goal lengths introduced by the
support for wide characters.

If the sizeof (wchar_t) times max_length would yield a value beyond
representation in a size_t, exit with a usage error up front, rather than
strange errors down the line from trying to malloc (well, realloc) with a size
of 0.

This is perhaps not the optimal behaviour - a clamp may be more appropriate as
we clamp the value of max_length now anyway, but this is at least better than
segfaulting or worse.  On systems which are friendly to malloc with a value of 0
the results could end up being strange corruption of the output.

16 years agoFix incorrect string formats for netstat/s
rrs [Wed, 17 Oct 2007 10:16:20 +0000 (10:16 +0000)]
Fix incorrect string formats for netstat/s

PR: 117175
Obtained from: Weongyo Jeong (weongyo.jeong@gmail.com)
MFC after: 1 week

16 years agoFix various memory leaks.
kib [Wed, 17 Oct 2007 09:52:08 +0000 (09:52 +0000)]
Fix various memory leaks.

Submitted by: rdivacky
Obtained from: OpenBSD
MFC after: 1 week

16 years agoThe previous version included too many file system types in the scan,
se [Wed, 17 Oct 2007 07:27:51 +0000 (07:27 +0000)]
The previous version included too many file system types in the scan,
since "local" includes also synthetic file systems (e.g. /dev, /proc)
and loopback mounts.

This version uses lsvfs to identify file system types that are local
and additionally not synthetik, loopback mounts, or read-only. This
has been suggested by Craig Rodrigues half a year ago. The patch that
has been committed is based on his suggestion, but slightly modified.

The comments in locate.rc have been updated to reflect the change and
o include zfs and xfs in the example file system parameter that can
be used to override the default outlined above.

PR: 114101
Submitted by: rodrigc at crodrigues dot org (Craig Rodrigues)
MFC: 2 weeks

16 years agoUse the direct mapping, if available, for pmap_zero_page_xscale() as well.
cognet [Tue, 16 Oct 2007 20:40:04 +0000 (20:40 +0000)]
Use the direct mapping, if available, for pmap_zero_page_xscale() as well.

16 years agoULE works fine on arm; allow it to be used
sam [Tue, 16 Oct 2007 19:25:26 +0000 (19:25 +0000)]
ULE works fine on arm; allow it to be used

Reviewed by: jeff, cognet, imp
MFC after: 1 week

16 years ago- Fix the handling of R_SPARC_OLO10, which is a bit of a special case
marius [Tue, 16 Oct 2007 19:17:48 +0000 (19:17 +0000)]
- Fix the handling of R_SPARC_OLO10, which is a bit of a special case
  in the way we implement handling of relocations.
  As for the kernel part this fixes the loading of lots of modules,
  which failed to load due to unresolvable symbols when built after
  the GCC 4.2.0 import. This wasn't due to a change in GCC itself
  though but one of several changes in configuration done along the
  import. Specfically, HAVE_AS_REGISTER_PSEUDO_OP, which causes GCC
  to denote global registers used for scratch purposes and in turn
  GAS uses R_SPARC_OLO10 relocations for, is now defined.
  While at it replace some more ELF_R_TYPE which should have been
  ELF64_R_TYPE_ID but didn't cause problems so far.
- Sync a sanity check between kernel and rtld(1) and change it to be
  maintenance free regarding the type used for the lookup table.
- Sprinkle const on lookup tables.
- Use __FBSDID.

Reported and tested by: yongari
MFC after: 5 days

16 years agoBackout rev 1.62, and revert to use -march=armv5te -D__XSCALE__ instead
cognet [Tue, 16 Oct 2007 18:32:37 +0000 (18:32 +0000)]
Backout rev 1.62, and revert to use -march=armv5te -D__XSCALE__ instead
of -mcpu=xscale for XScale.
gcc still has issues with -mcpu=xscale, and now crashes while building
systat.

Reported by: sam
MFC After: 3 days

16 years agoPrint the stack bounds of the thread.
marcel [Tue, 16 Oct 2007 17:52:59 +0000 (17:52 +0000)]
Print the stack bounds of the thread.

16 years ago- fix sctp_ifn initial refcount issue (prevents deletion)
rrs [Tue, 16 Oct 2007 14:05:51 +0000 (14:05 +0000)]
- fix sctp_ifn initial refcount issue (prevents deletion)
- fix a bug during cookie collision that prevented an
  association from coming up in a specific restart case.
- Fix it so the shutdown-pending flag gets removed (this is
  more for correctness then needed) when we enter shutdown-sent
  or shutdown-ack-sent states.
- Fix a bug that caused the receiver to sometimes NOT send
  a SACK when a duplicate TSN arrived. Without this fix
  it was possible for the association to fall down if the
- Deleted primary destination is also stored when SCTP_MOBILITY_BASE.
  (Previously, it is stored when only SCTP_MOBILITY_FASTHANDOFF)
- Fix a locking issue where we might call send_initiate_ack() and
  incorrectly state the lock held/not held. Also fix it so that
  when we release the lock the inp cannot be deleted on us.
- Add the debug option that can cause the stack to panic instead
  of aborting an assoc. This does not and should never show up
  in options but is useful for debugging unexpected aborts.
- Add cumack_log sent to track sending cumack information for
  the debug case where we are running a special log per assoc.
- Added extra () aroudn sctp_sbspace macro to avoid compile warnings.
MFC after: 1 week

16 years agoRescue hw.machine_arch in CTL_HW_NAMES.
ru [Tue, 16 Oct 2007 11:46:44 +0000 (11:46 +0000)]
Rescue hw.machine_arch in CTL_HW_NAMES.

16 years agoRescue whitespace nits.
ru [Tue, 16 Oct 2007 11:39:28 +0000 (11:39 +0000)]
Rescue whitespace nits.

16 years agoFix CTL_VM_NAMES.
ru [Tue, 16 Oct 2007 11:32:57 +0000 (11:32 +0000)]
Fix CTL_VM_NAMES.

16 years agoVM_METER is long deprecated.
ru [Tue, 16 Oct 2007 11:29:13 +0000 (11:29 +0000)]
VM_METER is long deprecated.

16 years agoRescue parts of the sensorsd commit that are still relevant:
ru [Tue, 16 Oct 2007 11:22:58 +0000 (11:22 +0000)]
Rescue parts of the sensorsd commit that are still relevant:

- HW_FLOATINGPOINT renamed to HW_FLOATINGPT.
- Documented HW_REALMEM.
- Sorted as per <sys/sysctl.h>.

16 years agoGet rid of qaddr_t.
alfred [Tue, 16 Oct 2007 10:54:55 +0000 (10:54 +0000)]
Get rid of qaddr_t.

Requested by: bde

16 years agoExport maxswzone, maxbcache, maxtsiz, dfldsiz, maxdsiz, dflssiz, maxssiz,
alfred [Tue, 16 Oct 2007 10:40:53 +0000 (10:40 +0000)]
Export maxswzone, maxbcache, maxtsiz, dfldsiz, maxdsiz, dflssiz, maxssiz,
and sgrowsiz via sysctl.

MFC after: 1 week

16 years agoUse macro THR_CLEANUP_PUSH/POP, they are cheaper than pthread_cleanup_push/pop.
davidxu [Tue, 16 Oct 2007 07:46:15 +0000 (07:46 +0000)]
Use macro THR_CLEANUP_PUSH/POP, they are cheaper than pthread_cleanup_push/pop.

16 years agoReverse the logic of UP and SMP.
davidxu [Tue, 16 Oct 2007 07:36:02 +0000 (07:36 +0000)]
Reverse the logic of UP and SMP.

Submitted by: jasone

16 years agoSome left over from the sensors framework.
mlaier [Tue, 16 Oct 2007 03:26:31 +0000 (03:26 +0000)]
Some left over from the sensors framework.

16 years agoSet PTE_ACCESSED in the PTE and before inserting it in the VHPT.
marcel [Tue, 16 Oct 2007 03:20:32 +0000 (03:20 +0000)]
Set PTE_ACCESSED in the PTE and before inserting it in the VHPT.
This avoids back-to-back faults for all TLB misses. This can be
improved further in the future by also setting PTE_DIRTY for TLB
misses for write accesses.

MFC after: 1 week

16 years agoThe flushrs instruction must be the first in an instruction
marcel [Tue, 16 Oct 2007 03:07:56 +0000 (03:07 +0000)]
The flushrs instruction must be the first in an instruction
group. GNU as(1) already made sure of that, but it's better
to actually have the code right.

MFC after: 1 week

16 years agoPrint instruction stops to improve analysis of dependency
marcel [Tue, 16 Oct 2007 02:59:03 +0000 (02:59 +0000)]
Print instruction stops to improve analysis of dependency
violations.

MFC after: 1 week

16 years agoFix disassembly of the invala, itc, itr and hint instructions
marcel [Tue, 16 Oct 2007 02:49:40 +0000 (02:49 +0000)]
Fix disassembly of the invala, itc, itr and hint instructions
by fixing the opcode ordering.

MFC after: 1 week

16 years agoBump version for libpcap and tcpdump imports
mlaier [Tue, 16 Oct 2007 02:39:16 +0000 (02:39 +0000)]
Bump version for libpcap and tcpdump imports

MFC after: 3 days

16 years agoUpdate for tcpdump 3.9.8
mlaier [Tue, 16 Oct 2007 02:32:44 +0000 (02:32 +0000)]
Update for tcpdump 3.9.8

16 years agoResolve merge conflicts
mlaier [Tue, 16 Oct 2007 02:31:48 +0000 (02:31 +0000)]
Resolve merge conflicts

Approved by: re (kensmith)
Obtained from: tcpdump.org

16 years agoThis commit was generated by cvs2svn to compensate for changes in r172683,
mlaier [Tue, 16 Oct 2007 02:20:42 +0000 (02:20 +0000)]
This commit was generated by cvs2svn to compensate for changes in r172683,
which included commits to RCS files with non-trunk default branches.

16 years agoImport of tcpdump v3.9.8
mlaier [Tue, 16 Oct 2007 02:20:42 +0000 (02:20 +0000)]
Import of tcpdump v3.9.8

16 years agoUpdate for libpcap 0.9.8
mlaier [Tue, 16 Oct 2007 02:12:06 +0000 (02:12 +0000)]
Update for libpcap 0.9.8

16 years agoUpdate for libpcap 0.9.8 import
mlaier [Tue, 16 Oct 2007 02:10:44 +0000 (02:10 +0000)]
Update for libpcap 0.9.8 import

16 years agoResolve merge conflicts
mlaier [Tue, 16 Oct 2007 02:07:55 +0000 (02:07 +0000)]
Resolve merge conflicts

Approved by: re (kensmith)
Obtained from: tcpdump.org

16 years agoThis commit was generated by cvs2svn to compensate for changes in r172677,
mlaier [Tue, 16 Oct 2007 02:02:02 +0000 (02:02 +0000)]
This commit was generated by cvs2svn to compensate for changes in r172677,
which included commits to RCS files with non-trunk default branches.

16 years agoImport of libpcap v0.9.8
mlaier [Tue, 16 Oct 2007 02:02:02 +0000 (02:02 +0000)]
Import of libpcap v0.9.8

16 years agoMake the updatedb script installed as /usr/libexec/locate.updatedb
se [Mon, 15 Oct 2007 22:09:19 +0000 (22:09 +0000)]
Make the updatedb script installed as /usr/libexec/locate.updatedb
inspect all local file systems, not only ufs and ext2fs. A number
of local file systems has been added over time, and at least zfs
has the potential to become a popular choice. Without this change
a ZFS root file system causes the script to ignore all file-systems
and leads to an empty locate db. (An alternative is to add all the
relevant file systems individually, which means that at least zfs,
xfs, ntfs, ntfs-3g, msdosfs should be added, probably more).

16 years agoMake ukbd_getc respect non-blocking behavior if one is requested. Change
kan [Mon, 15 Oct 2007 20:37:34 +0000 (20:37 +0000)]
Make ukbd_getc respect non-blocking behavior if one is requested. Change
ukbd_poll to mark this keyboard instance as polling before calling
usbd_set_polling at USB level. usbd_set_polling runs softintr before
returning, stealing our input and making consequent polling getchar
kind of pointless.

This allows USB keyboards to coexist peacefully with serial console in DDB
and other contexts where polling is used.

MFC after: 1 week

16 years agoBackout sensors framework.
netchild [Mon, 15 Oct 2007 20:00:24 +0000 (20:00 +0000)]
Backout sensors framework.

Requested by: phk
Discussed on: cvs-all

16 years agofix a bug of RX routines that can't process multi-frame transactions
sam [Mon, 15 Oct 2007 17:39:47 +0000 (17:39 +0000)]
fix a bug of RX routines that can't process multi-frame transactions
properly due to the shortage of the RX buffer size.  In a case of zyd
devices, up to 3 frames can be combined in an USB transaction.  So, RX
buffer should be at least ((MCLBYTES + extra structs) * 3)

Submitted by: Weongyo Jeong <weongyo.jeong@gmail.com>
MFC after: 3 days

16 years agoCorrect calculation of aac_sg_tablesize.
emaste [Mon, 15 Oct 2007 17:34:58 +0000 (17:34 +0000)]
Correct calculation of aac_sg_tablesize.

Obtained from: Adaptec, via driver b11669

16 years agoThis commit was generated by cvs2svn to compensate for changes in r172668,
mp [Mon, 15 Oct 2007 16:54:07 +0000 (16:54 +0000)]
This commit was generated by cvs2svn to compensate for changes in r172668,
which included commits to RCS files with non-trunk default branches.

16 years agoImport two vendor fixes from tcsh-6.15.01 for MFC to 7.0. The fixes are:
mp [Mon, 15 Oct 2007 16:54:07 +0000 (16:54 +0000)]
Import two vendor fixes from tcsh-6.15.01 for MFC to 7.0. The fixes are:
    - Fix pty detection for autologout setting
    - kill `foo` got stuck because sigchld was disabled too soon

Requested by: re

16 years agoOops, convert a tsleep() to a msleep() that was missed when adding locking
jhb [Mon, 15 Oct 2007 16:18:20 +0000 (16:18 +0000)]
Oops, convert a tsleep() to a msleep() that was missed when adding locking
to this driver.

Reported by: Michael Butler : imb of protected-networks net

16 years agoImport two vendor fixes from tcsh-6.15.01 for MFC to 7.0. The fixes are:
mp [Mon, 15 Oct 2007 15:23:07 +0000 (15:23 +0000)]
Import two vendor fixes from tcsh-6.15.01 for MFC to 7.0. The fixes are:
  - Fix pty detection for autologout setting
  - kill `foo` got stuck because sigchld was disabled too soon

Requested by: re

16 years agoThis commit was generated by cvs2svn to compensate for changes in r172665,
mp [Mon, 15 Oct 2007 15:23:07 +0000 (15:23 +0000)]
This commit was generated by cvs2svn to compensate for changes in r172665,
which included commits to RCS files with non-trunk default branches.

16 years agoOptimize for size on pc98. It enables to boot a kernel again.
nyan [Mon, 15 Oct 2007 14:20:24 +0000 (14:20 +0000)]
Optimize for size on pc98.  It enables to boot a kernel again.

I don't know what's wrong (loader, boot2 or others), but this change is
effective.

Tested by: NAKAJI Hiroyuki
MFC after: 3 days

16 years agoIf a required install tool is missing, provide a more friendly error
emaste [Mon, 15 Oct 2007 13:50:46 +0000 (13:50 +0000)]
If a required install tool is missing, provide a more friendly error
message (instead of cp's usage info).

16 years agoAdd comment explaining __mb_sb_limit trick here.
ache [Mon, 15 Oct 2007 09:51:30 +0000 (09:51 +0000)]
Add comment explaining __mb_sb_limit trick here.

16 years agoMarkup nit.
ru [Mon, 15 Oct 2007 08:29:53 +0000 (08:29 +0000)]
Markup nit.

16 years agoSort.
ru [Mon, 15 Oct 2007 08:26:12 +0000 (08:26 +0000)]
Sort.

16 years agoSort.
ru [Mon, 15 Oct 2007 08:17:12 +0000 (08:17 +0000)]
Sort.

16 years agoFix markup.
ru [Mon, 15 Oct 2007 08:07:15 +0000 (08:07 +0000)]
Fix markup.

16 years agoIntroduce FreeBSD 8.0 to mdoc(7).
ru [Mon, 15 Oct 2007 08:06:42 +0000 (08:06 +0000)]
Introduce FreeBSD 8.0 to mdoc(7).

16 years agoAdd myself to the FreeBSD calendar.
lme [Sun, 14 Oct 2007 19:55:32 +0000 (19:55 +0000)]
Add myself to the FreeBSD calendar.

Approved by:     miwi (mentor)

16 years agoAdd PCI IDs from Adaptec's driver version b11669:
emaste [Sun, 14 Oct 2007 19:40:15 +0000 (19:40 +0000)]
Add PCI IDs from Adaptec's driver version b11669:

Adaptec RAID 3085
Adaptec RAID 31205
Adaptec RAID 31605
Adaptec RAID 5085
Adaptec RAID 51205
Adaptec RAID 51605
Adaptec RAID 5445
Adaptec RAID 5805
IBM ServeRAID 8s
ICP RAID ICP5045BL
ICP RAID ICP5085BL
ICP RAID ICP5085SL
ICP RAID ICP5125BR
ICP RAID ICP5125SL
ICP RAID ICP5165BR
ICP RAID ICP5165SL
ICP RAID ICP5445SL
ICP RAID ICP5805BL
ICP RAID ICP5805SL
ICP9067MA SATA RAID

16 years agoFix makrup. There are still some OpenBSD-specific things
ru [Sun, 14 Oct 2007 19:38:11 +0000 (19:38 +0000)]
Fix makrup.  There are still some OpenBSD-specific things
in there that need tweaking.

16 years agostyle.Makefile(5) says: Do not use GCCisms (such as -g and -Wall)
ru [Sun, 14 Oct 2007 19:23:38 +0000 (19:23 +0000)]
style.Makefile(5) says: Do not use GCCisms (such as -g and -Wall)
in CFLAGS.

16 years agoSort.
ru [Sun, 14 Oct 2007 19:20:59 +0000 (19:20 +0000)]
Sort.

16 years agoBump document date for the last change.
ru [Sun, 14 Oct 2007 19:12:36 +0000 (19:12 +0000)]
Bump document date for the last change.

16 years agoFix mdoc.
ru [Sun, 14 Oct 2007 19:11:01 +0000 (19:11 +0000)]
Fix mdoc.

16 years agoRemove trailing whitespace.
ru [Sun, 14 Oct 2007 19:01:28 +0000 (19:01 +0000)]
Remove trailing whitespace.

16 years agoFix mdoc in last commit.
ru [Sun, 14 Oct 2007 18:57:32 +0000 (18:57 +0000)]
Fix mdoc in last commit.

16 years agoZFS_LOG adds a newline by itself.
thompsa [Sun, 14 Oct 2007 16:14:32 +0000 (16:14 +0000)]
ZFS_LOG adds a newline by itself.

Pointed out by: pjd

16 years agoThis changes give nullfs correctly work with latest unionfs.
daichi [Sun, 14 Oct 2007 13:57:11 +0000 (13:57 +0000)]
This changes give nullfs correctly work with latest unionfs.

Submitted by:   Masanori Ozawa <ozawa@ongs.co.jp> (unionfs developer)
Reviewed by:    jeff, kensmith
Approved by:    re (kensmith)
MFC after:      1 week

16 years agoAdded whiteout behavior option. ``-o whiteout=always'' is default mode
daichi [Sun, 14 Oct 2007 13:55:38 +0000 (13:55 +0000)]
Added whiteout behavior option. ``-o whiteout=always'' is default mode
(it is established practice) and ``-o whiteout=whenneeded'' is less
disk-space using mode especially for resource restricted environments
like embedded environments. (Contributed by Ed Schouten. Thanks)

Submitted by:   Masanori Ozawa <ozawa@ongs.co.jp> (unionfs developer)
Reviewed by:    jeff, kensmith
Approved by:    re (kensmith)
MFC after:      1 week