]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
17 years ago- Fix my name.
mnag [Fri, 30 Mar 2007 23:03:49 +0000 (23:03 +0000)]
- Fix my name.

17 years agoTeardown interrupt only when sc->ih is not NULL.
simokawa [Fri, 30 Mar 2007 22:25:26 +0000 (22:25 +0000)]
Teardown interrupt only when sc->ih is not NULL.

MFC after: 3 days

17 years agoAdd myself, fjoe, sergei, sem, sat, miwi, brooks and philip.
krion [Fri, 30 Mar 2007 22:25:17 +0000 (22:25 +0000)]
Add myself, fjoe, sergei, sem, sat, miwi, brooks and philip.

17 years agoAdd dates for joerg@ and obrien@.
le [Fri, 30 Mar 2007 22:20:43 +0000 (22:20 +0000)]
Add dates for joerg@ and obrien@.

17 years agoAdd my mentor and mentees.
ceri [Fri, 30 Mar 2007 22:06:56 +0000 (22:06 +0000)]
Add my mentor and mentees.
Add a comment clarifying the date format as it wasn't immediately
obvious from the current entries.

17 years agoAdd danger, garys, pav, jim, and trhodes to the list of committers, and
keramida [Fri, 30 Mar 2007 21:59:58 +0000 (21:59 +0000)]
Add danger, garys, pav, jim, and trhodes to the list of committers, and
list the first three as my mentees, and trhodes as co-mentor of danger.

17 years agoNew line for new sentense.
takawata [Fri, 30 Mar 2007 21:59:55 +0000 (21:59 +0000)]
New line for new sentense.

Pointed out by: brueffer

17 years agoNote that it was Nik Clayton who lured me into doc-committers.
keramida [Fri, 30 Mar 2007 21:45:08 +0000 (21:45 +0000)]
Note that it was Nik Clayton who lured me into doc-committers.

17 years agoHook acpi_aiboost man page and modify the man page a bit.
takawata [Fri, 30 Mar 2007 21:19:10 +0000 (21:19 +0000)]
Hook acpi_aiboost man page and modify the man page a bit.

17 years agoAdd man page for acpi_aiboost. Based on man page for NetBSD.
takawata [Fri, 30 Mar 2007 20:35:37 +0000 (20:35 +0000)]
Add man page for acpi_aiboost. Based on man page for NetBSD.

17 years agoFix more style(9) nits[1] and remove unnecessary use of '#if !defined(_KERNEL)'.
jkim [Fri, 30 Mar 2007 19:33:53 +0000 (19:33 +0000)]
Fix more style(9) nits[1] and remove unnecessary use of '#if !defined(_KERNEL)'.

Pointed out by: bde[1]

17 years agoRecord rc.d/nfslocking dependency on rc.d/rpcbind.
mtm [Fri, 30 Mar 2007 19:08:58 +0000 (19:08 +0000)]
Record rc.d/nfslocking dependency on rc.d/rpcbind.

PR: conf/105465
Submitted By: ru (with minor cosmetic change)
MFC-After: 1 month

17 years agoAdd my mentor
ehaupt [Fri, 30 Mar 2007 19:08:33 +0000 (19:08 +0000)]
Add my mentor

17 years agoAdd myself and mentor and grand-mentor. All the dates are from access log.
jkim [Fri, 30 Mar 2007 19:01:36 +0000 (19:01 +0000)]
Add myself and mentor and grand-mentor.  All the dates are from access log.

17 years agoFix \n on clement entry
garga [Fri, 30 Mar 2007 18:53:15 +0000 (18:53 +0000)]
Fix \n on clement entry

17 years agoAdd my two mentors, grog@ and joerg@, and myself.
le [Fri, 30 Mar 2007 18:49:36 +0000 (18:49 +0000)]
Add my two mentors, grog@ and joerg@, and myself.

Correct date for schweikh@.

17 years agoAdd myself + offspring, grand- and co-mentors.
mlaier [Fri, 30 Mar 2007 18:39:12 +0000 (18:39 +0000)]
Add myself + offspring, grand- and co-mentors.

17 years agoSome rc.d commands (such as stop|restart etc.) won't automagically work
mtm [Fri, 30 Mar 2007 18:36:45 +0000 (18:36 +0000)]
Some rc.d commands (such as stop|restart etc.) won't automagically work
if we don't explicitly set the name of the executable program.

PR: conf/104408

17 years agoUse the same wisdom of sys/i386/i386/support.s 1.97 to remove obfuscation.
jkim [Fri, 30 Mar 2007 18:27:57 +0000 (18:27 +0000)]
Use the same wisdom of sys/i386/i386/support.s 1.97 to remove obfuscation.

Pointed out by: bde

17 years ago- acardenas -> acm
flz [Fri, 30 Mar 2007 18:22:53 +0000 (18:22 +0000)]
- acardenas -> acm
- Sort alphabetically.

17 years ago- Add myself and mnag.
jmelo [Fri, 30 Mar 2007 18:16:22 +0000 (18:16 +0000)]
- Add myself and mnag.

17 years agodon't display ssid cloaking status as "ssid HIDE", use the cmd line
sam [Fri, 30 Mar 2007 18:14:04 +0000 (18:14 +0000)]
don't display ssid cloaking status as "ssid HIDE", use the cmd line
parameter like everything else

MFC after: 2 weeks

17 years ago- Use PARTIAL_PICKUP_GIANT() to implement PICKUP_GIANT().
jhb [Fri, 30 Mar 2007 18:10:08 +0000 (18:10 +0000)]
- Use PARTIAL_PICKUP_GIANT() to implement PICKUP_GIANT().
- Move UGAR() macro up to the comment that describes it.
- Fix a couple of typos.

17 years ago- Drop memory barriers in rw_try_upgrade(). We don't need an 'acq' memory
jhb [Fri, 30 Mar 2007 18:08:55 +0000 (18:08 +0000)]
- Drop memory barriers in rw_try_upgrade().  We don't need an 'acq' memory
  barrier here as the earlier rw_rlock() already contained one.
- Comment fix.

17 years agoAdd myself and db
ehaupt [Fri, 30 Mar 2007 18:08:51 +0000 (18:08 +0000)]
Add myself and db

17 years agoAdd some cross references to locking.9 from related pages.
julian [Fri, 30 Mar 2007 18:07:26 +0000 (18:07 +0000)]
Add some cross references to locking.9 from related pages.

17 years ago- Use lock_init/lock_destroy() to setup the lock_object inside of lockmgr.
jhb [Fri, 30 Mar 2007 18:07:24 +0000 (18:07 +0000)]
- Use lock_init/lock_destroy() to setup the lock_object inside of lockmgr.
  We can now use LOCK_CLASS() as a stronger check in lockmgr_chain() as a
  result.  This required putting back lk_flags as lockmgr's use of flags
  conflicted with other flags in lo_flags otherwise.
- Tweak 'show lock' output for lockmgr to match sx, rw, and mtx.

17 years agoAdd my 3 mentees (acm@, alepulver@ and vd@)
garga [Fri, 30 Mar 2007 18:05:28 +0000 (18:05 +0000)]
Add my 3 mentees (acm@, alepulver@ and vd@)

17 years agoAdd my mentor, edwin, and my mentees clement, lth, simon and lbr.
erwin [Fri, 30 Mar 2007 18:04:00 +0000 (18:04 +0000)]
Add my mentor, edwin, and my mentees clement, lth, simon and lbr.
gabor already added himself.

17 years agoUse underlying structures instead of kernel_sysctlbyname() for msginfo and
jkim [Fri, 30 Mar 2007 17:56:44 +0000 (17:56 +0000)]
Use underlying structures instead of kernel_sysctlbyname() for msginfo and
seminfo because kernel_sysctlbyname() is slow.  There is no dependency
problem since linux module depends on both sysvmsg and sysvsem and linprocfs
depends on it in turn.

Pointed out by: des
Reviewed by: des

17 years agoConnect the locking man page to the build.
julian [Fri, 30 Mar 2007 17:56:19 +0000 (17:56 +0000)]
Connect the locking man page to the build.
Next step is to add it as a "See Also" to related man pages.

17 years agoAdd an entry for schweikh. I couldn't find the commit bit date though.
flz [Fri, 30 Mar 2007 17:47:04 +0000 (17:47 +0000)]
Add an entry for schweikh. I couldn't find the commit bit date though.

17 years agoAdd some comments about entries formats.
flz [Fri, 30 Mar 2007 17:42:01 +0000 (17:42 +0000)]
Add some comments about entries formats.
Group by mentor, sort alphabetically, move adamw to alumni.

17 years agoMFP4: Fix style(9) nits and grammar in comments.
jkim [Fri, 30 Mar 2007 17:27:13 +0000 (17:27 +0000)]
MFP4: Fix style(9) nits and grammar in comments.

17 years ago- Add myself as keramida's mentee and sort
gabor [Fri, 30 Mar 2007 17:24:18 +0000 (17:24 +0000)]
- Add myself as keramida's mentee and sort

Reviewed by: flz

17 years agoAdd tmclaughs info
ahze [Fri, 30 Mar 2007 17:22:58 +0000 (17:22 +0000)]
Add tmclaughs info

17 years agoAdd dds
dds [Fri, 30 Mar 2007 17:22:08 +0000 (17:22 +0000)]
Add dds

17 years ago- Add myself as erwin's mentee
gabor [Fri, 30 Mar 2007 17:20:22 +0000 (17:20 +0000)]
- Add myself as erwin's mentee

Reviewed by: flz

17 years agoAdd my mentor tree neighbours.
marck [Fri, 30 Mar 2007 17:19:59 +0000 (17:19 +0000)]
Add my mentor tree neighbours.

17 years agoAdd tmclaugh to my list
ahze [Fri, 30 Mar 2007 17:17:46 +0000 (17:17 +0000)]
Add tmclaugh to my list

17 years agoMFP4: 114193, 114194
jkim [Fri, 30 Mar 2007 17:16:51 +0000 (17:16 +0000)]
MFP4: 114193, 114194

Dont "return" in linux_clone() after we forked the new process in a case
of problems.  Move the copyout of p2->p_pid outside the emul_lock coverage.

Submitted by: Roman Divacky

17 years agoAdd adamw's orphans
pav [Fri, 30 Mar 2007 17:12:58 +0000 (17:12 +0000)]
Add adamw's orphans

17 years agoAdd stub files to keep track of the mentor/mentee relationships.
flz [Fri, 30 Mar 2007 16:56:38 +0000 (16:56 +0000)]
Add stub files to keep track of the mentor/mentee relationships.
Have a look at committers-ports.dot for a start.

Simple command to generate the graph is:

$ dot -T png -o file.png file.dot

Dot can be installed via graphics/graphviz.

Discussed on: -developers

17 years agoUpdate with bzip2's URL.
delphij [Fri, 30 Mar 2007 16:33:05 +0000 (16:33 +0000)]
Update with bzip2's URL.

17 years ago - Be more verbose when saying "foo" not found.
delphij [Fri, 30 Mar 2007 16:32:08 +0000 (16:32 +0000)]
 - Be more verbose when saying "foo" not found.
 - In gctl_get_geom(), don't issue error when we were not
   provided with an parameter, like gctl_get_provider() did.

Reviewed by: pjd

17 years agoFree tlabel in fw_xfer_done().
simokawa [Fri, 30 Mar 2007 15:43:56 +0000 (15:43 +0000)]
Free tlabel in fw_xfer_done().

17 years ago- Don't call fw_busreset() in firewire_attach().
simokawa [Fri, 30 Mar 2007 14:41:24 +0000 (14:41 +0000)]
- Don't call fw_busreset() in firewire_attach().
This should fix the problem that the first bus reset is
sometimes ignored because of FWBUSRESET status.

MFC after: 3 days

17 years agoInstead of direct manipulation on queue and worklist mutexes, bring macros
wkoszek [Fri, 30 Mar 2007 14:34:34 +0000 (14:34 +0000)]
Instead of direct manipulation on queue and worklist mutexes, bring macros
for doing this job. This change will make it easy to migrate from using
spinning locks to adaptive ones.

Reviewed by: glebius, julian
Approved by: cognet (mentor)

17 years agoDo the right thing with symlinks in the skeleton directory.
le [Fri, 30 Mar 2007 13:18:52 +0000 (13:18 +0000)]
Do the right thing with symlinks in the skeleton directory.

PR:     bin/63659

17 years agoAccept passwords which contain whitespace.
le [Fri, 30 Mar 2007 12:57:25 +0000 (12:57 +0000)]
Accept passwords which contain whitespace.

PR:     bin/53434

17 years agoAdd home directory creation mode to pw.conf(5) and be a bit
le [Fri, 30 Mar 2007 11:23:10 +0000 (11:23 +0000)]
Add home directory creation mode to pw.conf(5) and be a bit
more specific about the effect of the current umask on -M.

17 years agoAdd -M argument to usage() output.
le [Fri, 30 Mar 2007 10:02:47 +0000 (10:02 +0000)]
Add -M argument to usage() output.

17 years agoThe last change is not needed for HEAD.
glebius [Fri, 30 Mar 2007 08:23:59 +0000 (08:23 +0000)]
The last change is not needed for HEAD.

Explained by: ru

17 years agoBack out my previous commit to this area, there are differences between
trhodes [Fri, 30 Mar 2007 06:12:45 +0000 (06:12 +0000)]
Back out my previous commit to this area, there are differences between
our implementation and OpenBSD's.

Requested by: des

17 years agoLine wrap previous entry
kris [Fri, 30 Mar 2007 04:31:53 +0000 (04:31 +0000)]
Line wrap previous entry

Submitted by: des

17 years agoDocument that EPERM is returned when the calling thread does not have
imp [Fri, 30 Mar 2007 04:10:04 +0000 (04:10 +0000)]
Document that EPERM is returned when the calling thread does not have
the mutex locked.  Also tweak the wording to make it more consistant
between pthread_cond_wait and pthread_cond_tiedwait.

Confirmed with the opengroup's web site that this is a valid return
value.  Wording used specifically not that of opengroup's online man
pages.

MFC After: 1 week

17 years agoMFP4: Linux futex support for amd64.
jkim [Fri, 30 Mar 2007 01:07:28 +0000 (01:07 +0000)]
MFP4: Linux futex support for amd64.

Initial patch was submitted by kib and additional work was done
by Divacky Roman.

Tested by: emulation

17 years agoRegen for set_thread_area.
jkim [Fri, 30 Mar 2007 00:08:21 +0000 (00:08 +0000)]
Regen for set_thread_area.

17 years agoMFP4: Linux set_thread_area syscall (aka TLS) support for amd64.
jkim [Fri, 30 Mar 2007 00:06:21 +0000 (00:06 +0000)]
MFP4: Linux set_thread_area syscall (aka TLS) support for amd64.

Initial version was submitted by Divacky Roman and mostly rewritten by me.

Tested by: emulation

17 years agoOops wrong line commented out in prev fix
ache [Thu, 29 Mar 2007 21:44:23 +0000 (21:44 +0000)]
Oops wrong line commented out in prev fix

17 years agoBack out network.subr :- fix and comment out dhc*_fxp0 examples instead
ache [Thu, 29 Mar 2007 21:42:19 +0000 (21:42 +0000)]
Back out network.subr :- fix and comment out dhc*_fxp0 examples instead

Submitted by:   jhb

17 years agoFix a bug in IPv4 address configuration exposed by refcounting.
bms [Thu, 29 Mar 2007 21:39:22 +0000 (21:39 +0000)]
Fix a bug in IPv4 address configuration exposed by refcounting.
 * Join the IPv4 all-hosts multicast group 224.0.0.1 once only;
   that is, when an IPv4 address is first configured on an interface.
 * Do not join it for subsequent IPv4 addresses as this violates IGMP.
 * Be sure to leave the group when all IPv4 addresses have been removed
   from the interface.
 * Add two DIAGNOSTIC printfs related to the issue.

Further care and attention is needed in this area; it is suggested that
netinet's attachment to the ifnet structure be compartmentalized and
non-implicit.

Bug found by: andre
MFC after: 1 month

17 years agoCorrect the name of the 'proto' command.
thompsa [Thu, 29 Mar 2007 21:30:26 +0000 (21:30 +0000)]
Correct the name of the 'proto' command.

MFC after: 3 days

17 years agosome minor error message cleanups
mjacob [Thu, 29 Mar 2007 21:29:26 +0000 (21:29 +0000)]
some minor error message cleanups

17 years agoUse size_t instead of unsigned for pagesize-related values, in order to
jasone [Thu, 29 Mar 2007 21:07:17 +0000 (21:07 +0000)]
Use size_t instead of unsigned for pagesize-related values, in order to
avoid downcasting issues.  In particular, this change fixes
posix_memalign(3) for alignments greater than 2^31 on LP64 systems.

Make sure that NDEBUG is always set to be compatible with MALLOC_DEBUG. [1]

Reported by: [1] Lee Hyo geol <hyogeollee@gmail.com>

17 years agovm_map_delete should be used only internally, by the VM subsystem. Replace
wkoszek [Thu, 29 Mar 2007 13:26:13 +0000 (13:26 +0000)]
vm_map_delete should be used only internally, by the VM subsystem. Replace
it with vm_map_remove, which not only embeds additional check, but also
takes care of locking.

Reviewed by: alc
Approved by: alc, cognet (mentor)

17 years agoFix get_if_var() with 3 args (i.e. with default)
ache [Thu, 29 Mar 2007 10:55:33 +0000 (10:55 +0000)]
Fix get_if_var() with 3 args (i.e. with default)

All xxx_<ifname> flags are set to empty strings automatically earlier so
eval echo \${${prefix}${_if}${suffix}-${_default}}
not substitute the default but return just the empty string.
Fix it using
eval echo \${${prefix}${_if}${suffix}:-${_default}}
(i.e. treat empty strings as unset)

The bug manifistates itself with the following warning from checkyesno():
/etc/rc.d/dhclient: WARNING: $background_dhclient is not set properly -
see rc.conf(5)

17 years agoAdd missing El.
des [Thu, 29 Mar 2007 10:11:22 +0000 (10:11 +0000)]
Add missing El.

MFC after: 1 week

17 years agoRevert rev. 1.205. Replace unconditional acquision of Giant when QUOTAS are
kib [Thu, 29 Mar 2007 08:26:04 +0000 (08:26 +0000)]
Revert rev. 1.205. Replace unconditional acquision of Giant when QUOTAS are
defined with VFS_LOCK_GIANT(NULL) call.
This shall fix softdep operation when mpsafe_vfs = 0.

Reported and tested by: kris
Submitted by: tegge
MFC after: 1 week

17 years agoExtend rev. 1.210 to avoid dereference NULL mp in VFS_NEEDSGIANT and
kib [Thu, 29 Mar 2007 08:21:09 +0000 (08:21 +0000)]
Extend rev. 1.210 to avoid dereference NULL mp in VFS_NEEDSGIANT and
VFS_ASSERT_GIANT. Stop using reserved namespace.

Reported and tested by: kris
Reviewed and enhanced by: tegge
MFC after: 1 week

17 years agoAdd some more random offensive quotes.
kris [Thu, 29 Mar 2007 08:08:34 +0000 (08:08 +0000)]
Add some more random offensive quotes.

17 years agoA note from Peter Wemm about how to tell the difference between amd64 and
kris [Thu, 29 Mar 2007 07:49:59 +0000 (07:49 +0000)]
A note from Peter Wemm about how to tell the difference between amd64 and
ia64.

17 years agoDialogue gem from Deadwood
kris [Thu, 29 Mar 2007 07:40:49 +0000 (07:40 +0000)]
Dialogue gem from Deadwood

17 years agoDeterministically set the default refresh interval to one second.
des [Thu, 29 Mar 2007 07:23:49 +0000 (07:23 +0000)]
Deterministically set the default refresh interval to one second.

Approved by: phk
MFC after: 1 week

17 years agoo Revert last. The chip is already supported by puc(4).
maxim [Thu, 29 Mar 2007 04:26:52 +0000 (04:26 +0000)]
o Revert last.  The chip is already supported by puc(4).

Pointed out by: marcel, Joseph Terner

17 years agoImplement the openat() linux syscall
julian [Thu, 29 Mar 2007 02:11:46 +0000 (02:11 +0000)]
Implement the openat() linux syscall
Submitted by: Roman Divacky (rdivacky@)
MFC after: 2 weeks

17 years agoRTC_TIMR's RTC_SEC field is BCD. That makes it unsuitable for
imp [Wed, 28 Mar 2007 22:40:37 +0000 (22:40 +0000)]
RTC_TIMR's RTC_SEC field is BCD.  That makes it unsuitable for
GetSeconds().  Instead, use CRTR register shifted right 15.  This
gives us a range of 32 seconds we can do for timeout.

Shift to using == rather than < or > for calculating the timeout,
since if we can't read the ST_CTRT register twice in a second we have
even bigger problems to worry about, and == deals with the 'wrap'
issue.

This lets me type at the boot2 prompt again!  Woo Hoo!

Bogusness noticed by: tisco
Pointy Hat to: That silly imp guy

17 years agoSince we're about to set ST_RTMR to 1 to increase the accuracy of the
imp [Wed, 28 Mar 2007 22:38:01 +0000 (22:38 +0000)]
Since we're about to set ST_RTMR to 1 to increase the accuracy of the
CRTR register to be 1/32768th of a second in init, we don't need to do
it here.

17 years agoAll SD cards have a block size of 512. The READ_BL_LEN field in the
imp [Wed, 28 Mar 2007 22:31:32 +0000 (22:31 +0000)]
All SD cards have a block size of 512.  The READ_BL_LEN field in the
CSD is usually 512 (well, 9), but for 2GB (and the rogue 4GB SD cards)
it is 1024 (or 2048 for 4GB).  This value doesn't work for the block
read commands (which really want 512).  Hardcode 512 for those.  This
may break really old MMC cards that don't have a 512 block size (I've
never seen one: make my day and send me one :-), but since the MMC
side of the house is currently broken, it should only have the effect
that 2GB (and non-conforming 4GB) SD cards will work.

My 'non-conforming' 4GB SD card also works now too.  The
non-conforming 4GB SD cards were sold for a while before the SD
association was worried they would be (a) incompatible (different FAT
flavor on them) and (b) confusing for the new SDHC standard and
cracked down on suppliers' bogus use of the SD trademark...

17 years agoRe-enable the HPET timer after a resume.
njl [Wed, 28 Mar 2007 22:28:48 +0000 (22:28 +0000)]
Re-enable the HPET timer after a resume.

Submitted by: Andrea Bittau <a.bittau@cs.ucl.ac.uk>
MFC after: 3 days

17 years agoTry to silence Coverity by adding (void) in front of function call.
emax [Wed, 28 Mar 2007 21:25:56 +0000 (21:25 +0000)]
Try to silence Coverity by adding (void) in front of function call.
Also add a comment, explaining why return value is not being checked.

Requested by: netchild
MFC after: 1 week

17 years agoNew sentence -> new line
brueffer [Wed, 28 Mar 2007 21:21:22 +0000 (21:21 +0000)]
New sentence -> new line

17 years agoVarious buglets fixed (from submitter):
imp [Wed, 28 Mar 2007 21:18:45 +0000 (21:18 +0000)]
Various buglets fixed (from submitter):

The changes to getstr() is so that the character that is
passed in to it, is also processed just as the rest. I also
removed one of the getc() calls otherwise you loose every
second character.

I also changed the strcpy of kname, so that it only happens if
kname is '\0'. This is so that one can pass a kernel in
through /boot.config.

The last change to boot2.c is in parse(). If you tried to type
a kernel name to boot, the first character was lost, the arg--
fix that.

Submitted by: jhay

17 years agoFix problem where memcmp would return true in the case where the
imp [Wed, 28 Mar 2007 21:15:50 +0000 (21:15 +0000)]
Fix problem where memcmp would return true in the case where the
character after the character that was the same.

Submitted by: jhay

17 years agoFix off by one error in length of the string.
imp [Wed, 28 Mar 2007 21:12:43 +0000 (21:12 +0000)]
Fix off by one error in length of the string.

Submitted by: jhay

17 years agoRemove the run promotion/demotion machinery. Replace it with red-black
jasone [Wed, 28 Mar 2007 19:55:07 +0000 (19:55 +0000)]
Remove the run promotion/demotion machinery.  Replace it with red-black
trees that track all non-full runs for each bin.  Use the red-black
trees to be able to guarantee that each new allocation is placed in the
lowest address available in any non-full run.  This change completes the
transition to allocating from low addresses in order to reduce the
retention of sparsely used chunks.

If the run in current use by a bin becomes empty, deallocate the run
rather than retaining it for later use.  The previous behavior had the
tendency to spread empty runs across multiple chunks, thus preventing
the release of chunks that were completely unused.

Generalize base_chunk_alloc() (and rename it to base_pages_alloc()) to
handle allocation sizes larger than the chunk size, so that it is
possible to support chunk sizes that are smaller than an arena object.

Reduce the minimum chunk size from 64kB to 8kB.

Optimize tracking of addresses for deleted chunks.

Fix a statistics bug for huge allocations.

17 years agoo Add MosChip 9835 PCI Dual UART PCI id.
maxim [Wed, 28 Mar 2007 18:56:27 +0000 (18:56 +0000)]
o Add MosChip 9835 PCI Dual UART PCI id.

PR: kern/110967
Submitted by: Joseph Terner
MFC after: 1 month

17 years agoChange macro in order to refer to FreeBSD 7.0 instead of 7.0BSD.
jasone [Wed, 28 Mar 2007 18:55:44 +0000 (18:55 +0000)]
Change macro in order to refer to FreeBSD 7.0 instead of 7.0BSD.

Reported by: Michal Mertl <mime@traveller.cz>

17 years agoFor embedded UARTs compatible with the ns8250 family it is possible
marcel [Wed, 28 Mar 2007 18:34:59 +0000 (18:34 +0000)]
For embedded UARTs compatible with the ns8250 family it is possible
that the driver clock is identical to the processor or bus clock.
This is the case for the PowerQUICC processor. When the clock is
high enough, overflows happen in the calculation of the time it
takes to send 1/10 of a character, used in delay loops. Fix the
overflows so as to fix bugs in the delay loops that can cause either
insufficient delays or excessive delays.

17 years agoWhen we match UARTs found during bus-enumeration with UARTs used for
marcel [Wed, 28 Mar 2007 18:26:12 +0000 (18:26 +0000)]
When we match UARTs found during bus-enumeration with UARTs used for
system devices (i.e. console, debug port or keyboard), don't stop
after the first match. Find them all and keep track of the last.
The reason for this change is that the low-level console is always
added to the list of system devices first, with other devices added
later. Since new devices are added to the list at the head, we have
the console always at the end. When a debug port is using the same
UART as the console, we would previously mark the "newbus" UART as
a debug port instead of as a console. This would later result in a
panic because no "newbus" device was associated with the console.
By matching all possible system devices we would mark the "newbus"
UART as a console and not as a debug port.
While it is arguably better to be able to mark a "newbus" UART as
both console and debug port, this fix is lightweight and allows
a single UART to be used as the console as well as a debug port
with only the aesthetic bug of not telling the user about it also
being a debug port.

Now that we match all possible system devices, update the rclk of
the system devices with the rclk that was obtained through the
bus attachment. It is generally true that clock information is
more reliable when obtained from the parent bus than by means of
some hardcoded or assumed value used early in the boot. This by
virtue of having more context information.

MFC after: 1 month

17 years agoAdd support for the 82562GX chip within if_fxp.
remko [Wed, 28 Mar 2007 18:10:50 +0000 (18:10 +0000)]
Add support for the 82562GX chip within if_fxp.

PR: 110251
Submitted by: Vyacheslav Vovk
Approved by: imp (mentor)
MFC After: 3 days

17 years agoAdd support for the RTL8110SC driver.
remko [Wed, 28 Mar 2007 18:07:12 +0000 (18:07 +0000)]
Add support for the RTL8110SC driver.

PR: 110804
Submitted by: Daan Vreeken
Sponsored by: Vitsch Electronics (patch)
Approved by: imp (mentor)
MFC After: 3 days

17 years agoAdd method enabled() to the SCC interface. This method can be used
marcel [Wed, 28 Mar 2007 18:05:17 +0000 (18:05 +0000)]
Add method enabled() to the SCC interface. This method can be used
by driver backends to mark individual channels as enabled or not.
The default implementation of this method always mark channels as
enabled.
This method is currently not used, but is added with the PowerQUICC
in mind where the 2nd SCC channel can be disabled.

17 years agoDon't derference a pointer before setting it.
mjacob [Wed, 28 Mar 2007 16:34:11 +0000 (16:34 +0000)]
Don't derference a pointer before setting it.
Very Pointy Dunce Cap T o: me.
Submitted by: Marcel
MFC after: 3 days

17 years agoDo not build the libedit support if NO_LIBPTHREAD is defined.
glebius [Wed, 28 Mar 2007 14:04:35 +0000 (14:04 +0000)]
Do not build the libedit support if NO_LIBPTHREAD is defined.

PR: bin/104478
Submitted by: Mike Durian <durian shadetreesoftware.com>

17 years agoBump maximum number of interface hooks to the maximum possible value.
glebius [Wed, 28 Mar 2007 13:59:13 +0000 (13:59 +0000)]
Bump maximum number of interface hooks to the maximum possible value.
This will increase the memory consumption for more than 1 Mb, but this
is required for operation on multiinterface access concentrators running
mpd.

Requested by: Alexander Motin

17 years agoWhen blackholing do a 'dropunlock' in the new world order to prevent the
andre [Wed, 28 Mar 2007 12:58:13 +0000 (12:58 +0000)]
When blackholing do a 'dropunlock' in the new world order to prevent the
INP_INFO_LOCK from leaking.

Reported by: ache
Found by: rwatson

17 years agonet.link.ether.inet.prune_intvl no longer exists.
glebius [Wed, 28 Mar 2007 11:06:50 +0000 (11:06 +0000)]
net.link.ether.inet.prune_intvl no longer exists.

17 years agoReflect netcat and bzip2 updates.
delphij [Wed, 28 Mar 2007 08:13:23 +0000 (08:13 +0000)]
Reflect netcat and bzip2 updates.

17 years agoThis commit was generated by cvs2svn to compensate for changes in r167984,
delphij [Wed, 28 Mar 2007 07:58:30 +0000 (07:58 +0000)]
This commit was generated by cvs2svn to compensate for changes in r167984,
which included commits to RCS files with non-trunk default branches.