]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agodocument the various bus_space_*_stream_? functions...
jmg [Sat, 22 Apr 2006 03:58:41 +0000 (03:58 +0000)]
document the various bus_space_*_stream_? functions...

18 years agoFix a typo.
jkoshy [Sat, 22 Apr 2006 03:04:24 +0000 (03:04 +0000)]
Fix a typo.

Submitted by: Pawel Worach <pawel.worach at gmail dot com>
MFC after: 1 week

18 years agoPrint td_name instead of p_comm if td_name is non-empty for
jhb [Fri, 21 Apr 2006 20:40:43 +0000 (20:40 +0000)]
Print td_name instead of p_comm if td_name is non-empty for
'show turnstile' and 'show sleepq'.

18 years agoUse LIST_FOREACH().
jhb [Fri, 21 Apr 2006 20:39:51 +0000 (20:39 +0000)]
Use LIST_FOREACH().

18 years agoo Do recrack(arguments) for commands which actually take NAME as
maxim [Fri, 21 Apr 2006 20:33:16 +0000 (20:33 +0000)]
o Do recrack(arguments) for commands which actually take NAME as
arguments so we do not coredump at "help foo", "back bar" and such.

o Be consistent and print argc - 1 as a command arguments number in
all cases.

PR: bin/37096
Submitted by: Joshua Goodall
MFC after: 1 month

18 years agoFree allocated environment variables after use.
jkim [Fri, 21 Apr 2006 19:35:03 +0000 (19:35 +0000)]
Free allocated environment variables after use.

Coverity ID: 366
Found by: Coverity Prevent

18 years agoDon't try to kill embryonic processes in killpg1(). This prevents
ps [Fri, 21 Apr 2006 19:26:21 +0000 (19:26 +0000)]
Don't try to kill embryonic processes in  killpg1().  This prevents
a race condition between fork() and kill(pid,sig) with pid < 0 that
can cause a kernel panic.

Submitted by: up
MFC after: 3 weeks

18 years agoRemove unnecessary assignment.
jkim [Fri, 21 Apr 2006 19:06:29 +0000 (19:06 +0000)]
Remove unnecessary assignment.

Coverity ID: 553
Found by: Coverity Prevent

18 years agoRedo some code based upon issues found by Coverity.
mjacob [Fri, 21 Apr 2006 18:46:35 +0000 (18:46 +0000)]
Redo some code based upon issues found by Coverity.

18 years agoChange some order of the way we do some target mode ops.
mjacob [Fri, 21 Apr 2006 18:31:21 +0000 (18:31 +0000)]
Change some order of the way we do some target mode ops.

Found by Coverity.

18 years agoSome more gratuitous format and name changes.
mjacob [Fri, 21 Apr 2006 18:30:01 +0000 (18:30 +0000)]
Some more gratuitous format and name changes.

Pull in some target mode changes from a private branch.
Pull in some more RELENG_4 compilation changes.

A lot of lines changed, but not much content change yet.

18 years agoMention the minidump feature.
delphij [Fri, 21 Apr 2006 16:22:01 +0000 (16:22 +0000)]
Mention the minidump feature.

18 years agoIn nexus_teardown_intr(), actually remove the handler.
marcel [Fri, 21 Apr 2006 16:12:28 +0000 (16:12 +0000)]
In nexus_teardown_intr(), actually remove the handler.

MFC after: 1 day

18 years agoMFp4:
imp [Fri, 21 Apr 2006 13:58:51 +0000 (13:58 +0000)]
MFp4:
Make getc(0) polling (this has the happy side effect of saving
8 bytes in the generated code).

18 years agoMFp4: Loop forever getting characters... A kludge for now.
imp [Fri, 21 Apr 2006 13:56:26 +0000 (13:56 +0000)]
MFp4: Loop forever getting characters...  A kludge for now.

18 years agoMFp4: Integrate support for building tag support for linux booting.
imp [Fri, 21 Apr 2006 13:51:47 +0000 (13:51 +0000)]
MFp4: Integrate support for building tag support for linux booting.

18 years agoMFp4:
imp [Fri, 21 Apr 2006 13:44:58 +0000 (13:44 +0000)]
MFp4:
Init eeprom here now that we don't init the eeprom in arm_init.s
and read in the second 8k of the data.
Tweak formatting.

18 years agoMove from arm_init.s -> .S. #ifdef linux vs freebsd defaults for the
imp [Fri, 21 Apr 2006 13:40:26 +0000 (13:40 +0000)]
Move from arm_init.s -> .S.  #ifdef linux vs freebsd defaults for the
boot loader (linux booting helps regression testing).  No repo copy
because of limited history.

18 years agoAdd minidump.h for pc98.
nyan [Fri, 21 Apr 2006 12:13:59 +0000 (12:13 +0000)]
Add minidump.h for pc98.

18 years agoMFi386: revision 1.554 (add minidump_machdep.c).
nyan [Fri, 21 Apr 2006 12:13:11 +0000 (12:13 +0000)]
MFi386: revision 1.554 (add minidump_machdep.c).

18 years agoMFi386: revision 1.625.
nyan [Fri, 21 Apr 2006 12:12:21 +0000 (12:12 +0000)]
MFi386: revision 1.625.

18 years agoBecause optarg and optind are defined in unistd.h,
delphij [Fri, 21 Apr 2006 09:39:51 +0000 (09:39 +0000)]
Because optarg and optind are defined in unistd.h,
remove the defination here.

18 years agoAllow for nmbclusters and maxsockets to be increased via sysctl.
ps [Fri, 21 Apr 2006 09:25:40 +0000 (09:25 +0000)]
Allow for nmbclusters and maxsockets to be increased via sysctl.
An eventhandler is used to update all the various zones that depend
on these values.

18 years agoMFp4: formatting nits and minor style changes that likely won't matter.
imp [Fri, 21 Apr 2006 07:45:57 +0000 (07:45 +0000)]
MFp4: formatting nits and minor style changes that likely won't matter.

18 years agoAllow one to enable building tags. Check the size of the boot imag.
imp [Fri, 21 Apr 2006 07:41:36 +0000 (07:41 +0000)]
Allow one to enable building tags.  Check the size of the boot imag.

18 years agoMFp4:
imp [Fri, 21 Apr 2006 07:29:14 +0000 (07:29 +0000)]
MFp4:

On the KB9202 go ahead and enable the flash controller so the boot
loader can access the parallel flash.

18 years agoUpdate for rev 1.205 of vm_mmap.c.
trhodes [Fri, 21 Apr 2006 07:22:55 +0000 (07:22 +0000)]
Update for rev 1.205 of vm_mmap.c.

Submitted by: fanf

18 years agoMFp4:
imp [Fri, 21 Apr 2006 07:19:22 +0000 (07:19 +0000)]
MFp4:

Make this compile, assuming that you have linux installed in a
sensible place.  tag_list is disabled by default, since we don't
distribute linux, but it is desirable to allow the boot loader to boot
Linux or FreeBSD (mostly for testing).

18 years agoIt seems that POSIX would rather ENODEV returned in place of EINVAL when
trhodes [Fri, 21 Apr 2006 07:17:25 +0000 (07:17 +0000)]
It seems that POSIX would rather ENODEV returned in place of EINVAL when
trying to mmap() an fd that isn't a normal file.

Reference: http://www.opengroup.org/onlinepubs/009695399/functions/mmap.html
Submitted by: fanf

18 years agoRemove what I believe are two useless ifdefs. If a user or administrator
trhodes [Fri, 21 Apr 2006 07:14:25 +0000 (07:14 +0000)]
Remove what I believe are two useless ifdefs.  If a user or administrator
enables multilabel, or any option for that matter, most likely they have
a reason.  This will allow users to see that mulilabel is enabled via an
issued "mount" command and remove an annoying warning - printed only when
a MAC kernel is not installed - on boot up.

Discussed with: green, brueffer, Samy Al Bahra.
Probably ran past: csjp (though I can't remember).

18 years agoMFp4:
imp [Fri, 21 Apr 2006 07:12:12 +0000 (07:12 +0000)]
MFp4:
Minor diffs to reduce diffs to p4.

18 years agospi flash updating boot loader. upload this file for the recovery
imp [Fri, 21 Apr 2006 06:51:08 +0000 (06:51 +0000)]
spi flash updating boot loader.  upload this file for the recovery
xmodem download.  Then download the image you want in the flash.
This will burn the image into the flash.  You must then reset the
unit and the new flash image will be used for booting...

18 years agoiic eeprom updating boot loader. upload this file for the recovery
imp [Fri, 21 Apr 2006 06:48:52 +0000 (06:48 +0000)]
iic eeprom updating boot loader.  upload this file for the recovery
xmodem download.  Then download the image you want in the eeprom.
This will burn the image into the eeprom.  You must then reset the
unit and the new eeprom image will be used for booting...

18 years agoAdd boot0iic and boot0spi. These are 'shortcut' boot loaders designed
imp [Fri, 21 Apr 2006 06:43:32 +0000 (06:43 +0000)]
Add boot0iic and boot0spi.  These are 'shortcut' boot loaders designed
to replace either a bad iic eeprom with one you download with xmodem,
or likewise with a spi flash.

18 years agoOops. Minidumps were developed on 6.x, in without the small pv entry code.
peter [Fri, 21 Apr 2006 04:50:18 +0000 (04:50 +0000)]
Oops.  Minidumps were developed on 6.x, in without the small pv entry code.
Add some strategic dump_add_page()/dump_drop_page() lines to include pv
chunks in the minidumps - these operate in the direct map region like UMA.

18 years agoTeach libkvm how to read minidumps. It turns out that minidumps are
peter [Fri, 21 Apr 2006 04:32:51 +0000 (04:32 +0000)]
Teach libkvm how to read minidumps.  It turns out that minidumps are
far more convenient for libkvm to work with because of the page table
block at the beginning.  As a result, the MD code is smaller.

libkvm will automatically detect old vs mini dumps on i386 and amd64.

libkvm will handle i386 PAE and non-PAE modes.  There is a PAE flag in
the i386 minidump header to signal the width of the entries in the
page table block.

Other convenient values are also present, such as kernbase and the direct
map addresses on amd64.

18 years agoMerge minidumps from amd64 where they were originally developed.
peter [Fri, 21 Apr 2006 04:28:43 +0000 (04:28 +0000)]
Merge minidumps from amd64 where they were originally developed.

Major differences:
 * since there is no direct map region, there is no custom uma memory
   allocator to modify to include its pages in the dumps.
 * Various data entries are reduced from 64 bit to 32 bit to match the
   native size.

dump_add_page() and dump_drop_page() are still present in case one wants to
arrange for arbitary pages to be dumped.  This is of marginal use though
because libkvm+kgdb cannot address physical memory that isn't mapped into
kvm.

18 years agoIntroduce minidumps. Full physical memory crash dumps are still available
peter [Fri, 21 Apr 2006 04:24:50 +0000 (04:24 +0000)]
Introduce minidumps.  Full physical memory crash dumps are still available
via the debug.minidump sysctl and tunable.

Traditional dumps store all physical memory.  This was once a good thing
when machines had a maximum of 64M of ram and 1GB of kvm.  These days,
machines often have many gigabytes of ram and a smaller amount of kvm.
libkvm+kgdb don't have a way to access physical ram that is not mapped
into kvm at the time of the crash dump, so the extra ram being dumped
is mostly wasted.

Minidumps invert the process.  Instead of dumping physical memory in
in order to guarantee that all of kvm's backing is dumped, minidumps
instead dump only memory that is actively mapped into kvm.

amd64 has a direct map region that things like UMA use.  Obviously we
cannot dump all of the direct map region because that is effectively
an old style all-physical-memory dump.  Instead, introduce a bitmap
and two helper routines (dump_add_page(pa) and dump_drop_page(pa)) that
allow certain critical direct map pages to be included in the dump.
uma_machdep.c's allocator is the intended consumer.

Dumps are a custom format.  At the very beginning of the file is a header,
then a copy of the message buffer, then the bitmap of pages present in
the dump, then the final level of the kvm page table trees (2MB mappings
are expanded into a 4K page mappings), then the sparse physical pages
according to the bitmap.  libkvm can now conveniently access the kvm
page table entries.

Booting my test 8GB machine, forcing it into ddb and forcing a dump
leads to a 48MB minidump.  While this is a best case, I expect minidumps
to be in the 100MB-500MB range.  Obviously, never larger than physical
memory of course.

minidumps are on by default.  It would want be necessary to turn them off
if it was necessary to debug corrupt kernel page table management as that
would mess up minidumps as well.

Both minidumps and regular dumps are supported on the same machine.

18 years agoSpelling/grammar fixes.
jkoshy [Fri, 21 Apr 2006 01:33:04 +0000 (01:33 +0000)]
Spelling/grammar fixes.

MFC after: 1 week

18 years ago- ipmi is a device, not an option (1)
brueffer [Thu, 20 Apr 2006 18:29:59 +0000 (18:29 +0000)]
- ipmi is a device, not an option (1)
- use standard formatting for authors

Submitted by: Pawel Worach <pawel.worach@gmail.com> (1)

18 years agoAfter some discussion we believe that having SERVERS to REQUIRE:
delphij [Thu, 20 Apr 2006 12:30:12 +0000 (12:30 +0000)]
After some discussion we believe that having SERVERS to REQUIRE:
ldconfig would provide necessary protection for named as well,
so remove the dependency here.

Approved by: flz

18 years agoFix x.fmt
ache [Thu, 20 Apr 2006 11:00:52 +0000 (11:00 +0000)]
Fix x.fmt

PR:     88222
Submitted by:   Anders Lindquist <andersl@saaf.se>

18 years agogeli(8) provides keys on newsession time, so remove CRD_F_KEY_EXPLICIT flag
pjd [Thu, 20 Apr 2006 06:33:46 +0000 (06:33 +0000)]
geli(8) provides keys on newsession time, so remove CRD_F_KEY_EXPLICIT flag
as HW crypto drivers don't support it.

18 years agopadlock(4) doesn't support explicitly provided keys yet.
pjd [Thu, 20 Apr 2006 06:31:44 +0000 (06:31 +0000)]
padlock(4) doesn't support explicitly provided keys yet.
Return an error instead of encrypting/decrypting data with a wrong key.

18 years agoadd debug msg
sam [Thu, 20 Apr 2006 05:03:21 +0000 (05:03 +0000)]
add debug msg

18 years agoSet the rid for the resoruce obtained from rman_reserve_resource.
imp [Thu, 20 Apr 2006 04:21:27 +0000 (04:21 +0000)]
Set the rid for the resoruce obtained from rman_reserve_resource.

18 years agoSet the rid for any resource obtained from rman_reserve_resource.
imp [Thu, 20 Apr 2006 04:20:41 +0000 (04:20 +0000)]
Set the rid for any resource obtained from rman_reserve_resource.

Reviewed by: wollman, jmg (as were the other commits fixing this problem)

18 years agoSet the rid for any resource obtained from rman_resource_reserve.
imp [Thu, 20 Apr 2006 04:19:10 +0000 (04:19 +0000)]
Set the rid for any resource obtained from rman_resource_reserve.

18 years agoSet the rid of the resource obtained from rman_reserve_resource.
imp [Thu, 20 Apr 2006 04:18:30 +0000 (04:18 +0000)]
Set the rid of the resource obtained from rman_reserve_resource.

18 years agoSet the rid for a resoruce allocated with rman_reserve_resource.
imp [Thu, 20 Apr 2006 04:16:34 +0000 (04:16 +0000)]
Set the rid for a resoruce allocated with rman_reserve_resource.

18 years agoSet the rid for any resources that are allocated at the level where
imp [Thu, 20 Apr 2006 04:16:02 +0000 (04:16 +0000)]
Set the rid for any resources that are allocated at the level where
they first emerge from the rman_* system.

18 years agoWhen returning a resource that we've allocated with rman_reserve_resource,
imp [Thu, 20 Apr 2006 04:12:02 +0000 (04:12 +0000)]
When returning a resource that we've allocated with rman_reserve_resource,
go ahead and set the rid for that resource.

18 years agoSet the rid of the resource we're about to return to the user.
imp [Thu, 20 Apr 2006 04:10:27 +0000 (04:10 +0000)]
Set the rid of the resource we're about to return to the user.

18 years agoFree another memory leak when dealing with disk notification.
ps [Thu, 20 Apr 2006 03:05:02 +0000 (03:05 +0000)]
Free another memory leak when dealing with disk notification.

18 years agoconst'ify resource_spec to note that we won't be changing anything while
jmg [Thu, 20 Apr 2006 01:44:16 +0000 (01:44 +0000)]
const'ify resource_spec to note that we won't be changing anything while
releasing resources... also, NULL out the resources as we free them...

18 years agor_spare1 and r_spare2 aren't needed. They aren't used. They can't be
imp [Wed, 19 Apr 2006 21:25:55 +0000 (21:25 +0000)]
r_spare1 and r_spare2 aren't needed.  They aren't used.  They can't be
accessed from outside of subr_rman.c.  Remove them.

Reviewed by: jmg (in theory)

18 years agoRemove debounce code in mpt_intr. After some reflection
mjacob [Wed, 19 Apr 2006 21:17:33 +0000 (21:17 +0000)]
Remove debounce code in mpt_intr. After some reflection
and watching a debounce followed by a timeout, I think
I'm forced to conclude that it was not a good idea.

18 years agoAdd MLINKS for rwlock(9).
jhb [Wed, 19 Apr 2006 21:09:39 +0000 (21:09 +0000)]
Add MLINKS for rwlock(9).

18 years agoDocument rw_try_upgrade() and rw_downgrade().
jhb [Wed, 19 Apr 2006 21:09:17 +0000 (21:09 +0000)]
Document rw_try_upgrade() and rw_downgrade().

18 years agoImplement rw_try_upgrade() and rw_downgrade(). rw_try_upgrade() makes a
jhb [Wed, 19 Apr 2006 21:06:52 +0000 (21:06 +0000)]
Implement rw_try_upgrade() and rw_downgrade().  rw_try_upgrade() makes a
single attempt at upgrading a read lock to a write lock, and rw_downgrade()
converts curthread's write lock into a read lock.

18 years agoVarious nits and fixups so that the text reads smoother in several places.
jhb [Wed, 19 Apr 2006 20:43:35 +0000 (20:43 +0000)]
Various nits and fixups so that the text reads smoother in several places.
Also, more closely match the prototypes in sys/rwlock.h and add a BUGS
section.

18 years agoTrim a couple of xrefs.
jhb [Wed, 19 Apr 2006 20:41:56 +0000 (20:41 +0000)]
Trim a couple of xrefs.

18 years agoFree some previously leaked memory on module unload.
ps [Wed, 19 Apr 2006 19:56:10 +0000 (19:56 +0000)]
Free some previously leaked memory on module unload.

18 years agoRemove sab(4).
marcel [Wed, 19 Apr 2006 19:39:35 +0000 (19:39 +0000)]
Remove sab(4).

18 years agoAdd a missing .Pp call.
brueffer [Wed, 19 Apr 2006 19:21:42 +0000 (19:21 +0000)]
Add a missing .Pp call.

MFC after: 3 days

18 years agoIndentation fixes
brueffer [Wed, 19 Apr 2006 19:09:21 +0000 (19:09 +0000)]
Indentation fixes

18 years agoDocument how the backoff delay is calculated.
brueffer [Wed, 19 Apr 2006 17:46:27 +0000 (17:46 +0000)]
Document how the backoff delay is calculated.

Submitted by: markus
MFC after: 3 days

18 years agoMFp4 (checkpoint of work in progress):
imp [Wed, 19 Apr 2006 17:16:49 +0000 (17:16 +0000)]
MFp4 (checkpoint of work in progress):

o Use a directory layout that is more akin to the i386 boot layout.
o Create a libat91 for library routines that are used by one or more
  of the boot loaders.
o Create bootiic for booting from an iic part.
o Create bootspi for booting from an spi part.
o Optimize the size of many of these routines (especially emac.c).  Except
  for the emac.c optimizations, all these have been tested.
o eliminate the inc directory, libat91 superceeds it.
o Move linker.cfg up a layer to allow it to be shared.

18 years agoRe-add option -C to usage().
ru [Wed, 19 Apr 2006 12:19:06 +0000 (12:19 +0000)]
Re-add option -C to usage().

18 years agoDocument the -C option of top(1), which has been added to the source
keramida [Wed, 19 Apr 2006 11:47:51 +0000 (11:47 +0000)]
Document the -C option of top(1), which has been added to the source
and the usage message of top, but not to the manpage.

Pointy hat to:  keramida
Reminded by:    ru
MFC after:      1 week

18 years agoAdd -u and -U which functions like the jail(8) that allow
delphij [Wed, 19 Apr 2006 10:12:10 +0000 (10:12 +0000)]
Add -u and -U which functions like the jail(8) that allow
commands to be executed under a different credential.

PR: bin/95777
Submitted by: clsung

18 years agoCorrect a local information leakage bug affecting AMD FPUs.
cperciva [Wed, 19 Apr 2006 07:00:19 +0000 (07:00 +0000)]
Correct a local information leakage bug affecting AMD FPUs.

Security: FreeBSD-SA-06:14.fpu

18 years agoMake ldconfig as SERVER dependency. This makes it possible
delphij [Wed, 19 Apr 2006 05:10:34 +0000 (05:10 +0000)]
Make ldconfig as SERVER dependency.  This makes it possible
for some early starting services from the ports collection
to have their shared objects available before start.

Reviewed by: freebsd-rc (dougb, brooks)
MFC After: 3 days

18 years agoFxi tpyo.
mjacob [Tue, 18 Apr 2006 22:43:46 +0000 (22:43 +0000)]
Fxi tpyo.

Noticed by: maxim

18 years agoAdd ISP_DEFAULT_ROLES as a config option.
mjacob [Tue, 18 Apr 2006 22:24:55 +0000 (22:24 +0000)]
Add ISP_DEFAULT_ROLES as a config option.

18 years agoActually check to see if XPT_CALC_GEOMETRY had an error
mjacob [Tue, 18 Apr 2006 22:01:59 +0000 (22:01 +0000)]
Actually check to see if XPT_CALC_GEOMETRY had an error
and pick a bogus geometry if it had so we don't later
get a divide by zero trap.

18 years agoUpdate 2300 f/w. This will probably be the last
mjacob [Tue, 18 Apr 2006 21:55:30 +0000 (21:55 +0000)]
Update 2300 f/w. This will probably be the last
update here before we switch to the new f/w loading
framework.

18 years agoDo more guarding against zero divide traps for the geom subroutine.
mjacob [Tue, 18 Apr 2006 21:53:39 +0000 (21:53 +0000)]
Do more guarding against zero divide traps for the geom subroutine.

18 years agoIn receiving a new ATIO, don't record the associated CCB in the target
mjacob [Tue, 18 Apr 2006 21:52:00 +0000 (21:52 +0000)]
In receiving a new ATIO, don't record the associated CCB in the target
state structure. This field is only for CCBs that are associated with
actions that are occurring on the HBA (i.e., XPT_CONT_IO actions).

This way we also don't get confused when the upstream listener stalls
try and look at a CCB which has already been freed (by CAM).

18 years agoCorrect usage example.
ceri [Tue, 18 Apr 2006 21:37:24 +0000 (21:37 +0000)]
Correct usage example.

PR: docs/95059
Submitted by: Jos Backus
MFC after: 4 days

18 years ago'owner' is not used without SMP. Fix kernel build for such kernel
wkoszek [Tue, 18 Apr 2006 20:32:42 +0000 (20:32 +0000)]
'owner' is not used without SMP. Fix kernel build for such kernel
configurations.

Approved by: jhb

18 years agoIf we're doing a try-alloc of a pv entry and give up early, do not forget
peter [Tue, 18 Apr 2006 20:17:32 +0000 (20:17 +0000)]
If we're doing a try-alloc of a pv entry and give up early, do not forget
to reduce the pv_entry_count counter.  This was found by Tor Egge.  In the
same email, Tor also pointed out the pv_stats problem in the previous
commit, but I'd forgotten about it until I went looking for this email
about this allocation problem.

18 years agopv_entry_count is more than a statistic. It is used for resource limiting.
peter [Tue, 18 Apr 2006 20:11:00 +0000 (20:11 +0000)]
pv_entry_count is more than a statistic.  It is used for resource limiting.
Do not compile out its counter updates if pv entry stats are turned off.

18 years agoSome grammar/language improvements, more to come. Also note some
brueffer [Tue, 18 Apr 2006 19:35:48 +0000 (19:35 +0000)]
Some grammar/language improvements, more to come.  Also note some
merged items.

18 years agoo Describe what actually boot(8) -m mutes.
maxim [Tue, 18 Apr 2006 19:00:02 +0000 (19:00 +0000)]
o Describe what actually boot(8) -m mutes.

PR: docs/67893
Submitted by: Chris Pepper, Daniel Gerzo
MFC after: 2 weeks

18 years agoAdaptively spin before blocking on the turnstile if an rwlock is write
jhb [Tue, 18 Apr 2006 18:27:54 +0000 (18:27 +0000)]
Adaptively spin before blocking on the turnstile if an rwlock is write
locked.  In general the adaptive spinning is similar to the same code
for mutexes with some extra trickiness in rw_wunlock_hard().  Specifically,
even though both wait bits might be set and we might have a turnstile with
at least one waiting thread, there might not be any threads blocked on the
queue we are not waking up (they might all be spinning), and we should
only preserve the waiting flag for the queue we aren't waking up if there
are in fact threads blocked on that queue.  Secondly, there might not be
any threads blocked on the queue we have chosen to waken threads from
(there might only be threads blocked on the other queue and the threads
for this queue are all spinning) in which case we disown the turnstile
instead of doing a braodcast and unpend.

18 years agoUpdate comments to mention that each turnstile contains two queues and to
jhb [Tue, 18 Apr 2006 18:21:38 +0000 (18:21 +0000)]
Update comments to mention that each turnstile contains two queues and to
describe turnstile_disown() and turnstile_empty().

18 years ago- Bring back turnstile_empty() which can check to see if an individual
jhb [Tue, 18 Apr 2006 18:16:54 +0000 (18:16 +0000)]
- Bring back turnstile_empty() which can check to see if an individual
  queue on a turnstile is empty.
- Add a turnstile_disown() function that allows a thread to give up
  ownership of a turnstile w/o waking up any waiters.

18 years agoBring SYNOPSIS in line with usage().
ru [Tue, 18 Apr 2006 15:26:58 +0000 (15:26 +0000)]
Bring SYNOPSIS in line with usage().

18 years ago- Move _rc_subr_loaded=: at the end of the file, to be consistent with NetBSD.
flz [Tue, 18 Apr 2006 15:16:55 +0000 (15:16 +0000)]
- Move _rc_subr_loaded=: at the end of the file, to be consistent with NetBSD.
- Sync with latest version from NetBSD.

'In order to handle some perl scripts running as daemons, add a
pattern which also matches "`basename $interpreter`: $command" in the
ps listing.'

Approved by: cperciva (mentor)
Obtained from: NetBSD
MFC after: 1 week

18 years ago- Add new ntpd_config variable so that people can override it in rc.conf.
flz [Tue, 18 Apr 2006 15:02:24 +0000 (15:02 +0000)]
- Add new ntpd_config variable so that people can override it in rc.conf.
- Add default value in /etc/defaults/rc.conf.
- Add documentation bits to rc.conf(5).

Approved by: cperciva (mentor)
MFC after: 1 week

18 years agoFix storing offset of already synchronized data. Offset in entire array was
pjd [Tue, 18 Apr 2006 13:52:11 +0000 (13:52 +0000)]
Fix storing offset of already synchronized data. Offset in entire array was
stored in metadata instead of an offset in single disk.
After reboot/crash synchronization process started from a wrong offset
skipping (not synchronizing) part of the component which can lead to data
corrutpion (when synchronization process was interrupted on initial
synchronization) or other strange situations like 'graid3 status' showing
value more than 100%.

Reported, reviewed and tested by: ru
Reported by: Dmitry Morozovsky <marck@rinet.ru>
MFC after: 1 day

18 years agofix "ipf -Z" reporting rubbish and possibly panic'ing box
darrenr [Tue, 18 Apr 2006 13:24:14 +0000 (13:24 +0000)]
fix "ipf -Z" reporting rubbish and possibly panic'ing box

MFC after: 4 days

18 years agoFix an obvious bug.
ru [Tue, 18 Apr 2006 12:37:44 +0000 (12:37 +0000)]
Fix an obvious bug.

Submitted by: Alex Lyashkov <umka@sevcity.net>

18 years agoAdd ldconfig to the list of requirements for named, needed to use bind
flz [Tue, 18 Apr 2006 10:35:05 +0000 (10:35 +0000)]
Add ldconfig to the list of requirements for named, needed to use bind
from ports. The effect is that ldconfig is now started right after
mountcritremote. Everything else is left unchanged.

PR: conf/68916
Submitted by: JD Bronson <jd@aurora.org>
Approved by: cperciva (mentor)
MFC after: 1 week

18 years agoMerge rev. 1.240 of ip_output.c, so that IPFIREWALL_FORWARD_EXTENDED
glebius [Tue, 18 Apr 2006 09:20:16 +0000 (09:20 +0000)]
Merge rev. 1.240 of ip_output.c, so that IPFIREWALL_FORWARD_EXTENDED
kernel option will affect both forwarding methods - classic and fast.

18 years agoIn vfs_hash_get(): mount point should never be changed
delphij [Tue, 18 Apr 2006 08:05:08 +0000 (08:05 +0000)]
In vfs_hash_get(): mount point should never be changed
so explicitly constify the mp parameter.

Reviewed by: phk

18 years agoReorder the DRM_*_AGP enums to match linux and what our code expects, fixing
anholt [Tue, 18 Apr 2006 06:14:43 +0000 (06:14 +0000)]
Reorder the DRM_*_AGP enums to match linux and what our code expects, fixing
i915 attachment.

Submitted by: Jonathan Fosburgh <jonathan@fosburgh.org>

18 years agoCorrect a typo: In order to make fixmount(8) quite, one should
delphij [Tue, 18 Apr 2006 03:33:17 +0000 (03:33 +0000)]
Correct a typo: In order to make fixmount(8) quite, one should
use -q, not -v.

18 years agoremove DRIVER_MODULE lines that are useless... pcf doesn't exist (only
jmg [Mon, 17 Apr 2006 22:33:42 +0000 (22:33 +0000)]
remove DRIVER_MODULE lines that are useless...  pcf doesn't exist (only
as pcf_ebus and pcf_isa, they should probably be fixed back to pcf),
and bti2c doesn't exist, bktr has smbus or iicbb as children..

Brought to you by: http://people.FreeBSD.org/~jmg/driver.pdf