]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
18 years agoWhen sending export datagram from interrupt thread, use NG_QUEUE
glebius [Thu, 12 Jan 2006 22:48:12 +0000 (22:48 +0000)]
When sending export datagram from interrupt thread, use NG_QUEUE
in flags. When sending export datagram from expiry thread, then
use default zero flags. This removes unpleasant contention of the
interrupt thread on mutexes (usually ng_ksocket's socket buffer
mutex).

18 years agoProvide additional macros for sending netgraph items, which allow
glebius [Thu, 12 Jan 2006 22:44:04 +0000 (22:44 +0000)]
Provide additional macros for sending netgraph items, which allow
to use non-default flags for netgraph functions. Implement current
macros via new ones.

18 years ago- Print also node ID in ktr(9) messages. [1]
glebius [Thu, 12 Jan 2006 22:41:32 +0000 (22:41 +0000)]
- Print also node ID in ktr(9) messages. [1]
- Use fixed length for function name, making ktrdump(8) output
  easier to read.

Suggested by: julian [1]

18 years agoSkip format flags, when parsing ktr_desc.
glebius [Thu, 12 Jan 2006 22:32:07 +0000 (22:32 +0000)]
Skip format flags, when parsing ktr_desc.

18 years ago* fix bst.status. We mark some bits, but forgot to reset all of them
bruno [Thu, 12 Jan 2006 21:56:37 +0000 (21:56 +0000)]
* fix bst.status.  We mark some bits, but forgot to reset all of them
  before.  The symptom is that the battery inform us its charge and discharge
  at the same time...

* fix bst.rate to correctly output the (dis)charging rate.  We'll use
  the current average over one minute command and not the at_rate command.
  Note that this method is not correct if the capacity_mode is set, but
  since we don't set it ourself, it is not a problem.

  The at_rate do not give the actual rate but is used to compute the
  estimated time for (dis)charging a battery.  We should actually
  write an estimation of the actual rate using at_rate cmd and then
  perform a read to the various estimators.

Approved by: njl
MFC after: 2 days

18 years agoDo not force queueing on peer hooks. This was important only for
glebius [Thu, 12 Jan 2006 21:09:12 +0000 (21:09 +0000)]
Do not force queueing on peer hooks. This was important only for
5.0-CURRENT. And it looks like this didn't work before Julian's
revamp of netgraph queue code.

Reviewed by: julian

18 years agoInclude the bridge interface itself in the special arp handling.
thompsa [Thu, 12 Jan 2006 21:05:30 +0000 (21:05 +0000)]
Include the bridge interface itself in the special arp handling.

PR: 90973
MFC after: 1 week

18 years agoRemove old debugging leftover.
glebius [Thu, 12 Jan 2006 21:03:09 +0000 (21:03 +0000)]
Remove old debugging leftover.

Reviewed by: julian

18 years agoBelated __FreeBSD_version bump for improvements to the Linux ldconfig
brooks [Thu, 12 Jan 2006 20:27:35 +0000 (20:27 +0000)]
Belated __FreeBSD_version bump for improvements to the Linux ldconfig
support in etc/rc.d/abi.

18 years agoMark appropriate commands with NGM_READONLY and NGM_HASREPLY and
glebius [Thu, 12 Jan 2006 19:16:08 +0000 (19:16 +0000)]
Mark appropriate commands with NGM_READONLY and NGM_HASREPLY and
bump type cookie.

18 years agoNovel idea, don't print a string if it is NULL!
alfred [Thu, 12 Jan 2006 19:15:14 +0000 (19:15 +0000)]
Novel idea, don't print a string if it is NULL!

This protects people from loading _really_ old modules, like say from
5.x to a 6.x or 7.x system, like for instance right after an upgrade.

18 years agoIn the splnet(9) times netgraph(4) was synchronous and if a message
glebius [Thu, 12 Jan 2006 19:14:40 +0000 (19:14 +0000)]
In the splnet(9) times netgraph(4) was synchronous and if a message
had been replied, the reply was always delivered to the originator
synchronously.

With introduction of netgraph item callbacks and a wait channel with
mutex in ng_socket(4), we have fixed the problem with ngctl(8) returning
earlier than the command has been proceeded by target node. But still
ngctl(8) can return prior to the reply has arrived to its node.

To fix this:
 - Introduce a new flag for netgraph(4) messages - NGM_HASREPLY.
   This flag is or'ed with message like NGM_READONLY.
 - In netgraph userland library if we have sent a message with
   NGM_HASREPLY flag, then select(2) until reply comes.
 - Mark appropriate generic commands with NGM_HASREPLY flag,
   gathering them into one enum {}. Bump generic cookie.

18 years agoFix a bitwise logic error in posix_memalign().
jasone [Thu, 12 Jan 2006 18:09:25 +0000 (18:09 +0000)]
Fix a bitwise logic error in posix_memalign().

Reported by: glebius

18 years agoRemove releases now found in Groff sources.
ru [Thu, 12 Jan 2006 15:33:18 +0000 (15:33 +0000)]
Remove releases now found in Groff sources.

18 years agoPull up from the FSF branch.
ru [Thu, 12 Jan 2006 15:31:41 +0000 (15:31 +0000)]
Pull up from the FSF branch.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r154258,
ru [Thu, 12 Jan 2006 15:26:11 +0000 (15:26 +0000)]
This commit was generated by cvs2svn to compensate for changes in r154258,
which included commits to RCS files with non-trunk default branches.

18 years agoMerge support for new BSD releases from upstream:
ru [Thu, 12 Jan 2006 15:26:11 +0000 (15:26 +0000)]
Merge support for new BSD releases from upstream:

- Darwin 8.[0-3].0
- FreeBSD 4.11, 5.4, 6.0, 6.1
- NetBSD 3.0

18 years agoIn moduledir_readhints() cast the value returned by sizeof() to ssize_t
marius [Thu, 12 Jan 2006 13:18:49 +0000 (13:18 +0000)]
In moduledir_readhints() cast the value returned by sizeof() to ssize_t
when checking whether it's greater than a struct stat st_size in order
to also catch the case when st_size is -1. Previously this check didn't
trigger on sparc64 when st_size is -1 (as it's the case for a file on
a bzipfs, TFTP server etc.), causing the content of the linker hints
file to be copied to memory referenced by a null-pointer.

PR: 91231
MFC after: 1 week

18 years ago- The inline asm in this file uses output operands before all input
marius [Thu, 12 Jan 2006 11:40:39 +0000 (11:40 +0000)]
- The inline asm in this file uses output operands before all input
  operands are consumed so use the appropriate constraint modifier.
  Before this change GCC used one register for both an input and an
  unrelated output operand of in_addword(), causing the input to be
  overwritten before it was consumed and thus breaking in_addword().
  For in_cksum_hdr() and in_pseudo() this change is more or less
  cosmetic.
- Fix a misspelling in a nearby comment.

Reported & tested by: yongari
MFC after: 1 week

18 years agoFix wording in last commit.
glebius [Thu, 12 Jan 2006 10:15:51 +0000 (10:15 +0000)]
Fix wording in last commit.

Submitted by: julian

18 years agoUse posix_memalign() in valloc() rather than making assumptions about
jasone [Thu, 12 Jan 2006 09:29:38 +0000 (09:29 +0000)]
Use posix_memalign() in valloc() rather than making assumptions about
the alignment of malloc()ed memory.

Approved by: markm (mentor)

18 years agoUse posix_memalign() rather than assuming that malloc() provides adequate
jasone [Thu, 12 Jan 2006 08:01:38 +0000 (08:01 +0000)]
Use posix_memalign() rather than assuming that malloc() provides adequate
alignment.

Approved by: markm (mentor)

18 years agoExpose the posix_memalign() prototype, now that the function is implemented
jasone [Thu, 12 Jan 2006 07:58:59 +0000 (07:58 +0000)]
Expose the posix_memalign() prototype, now that the function is implemented
by libc.

18 years agoFix build without -DNDEBUG.
harti [Thu, 12 Jan 2006 07:44:40 +0000 (07:44 +0000)]
Fix build without -DNDEBUG.

Spotted by: obrien

18 years agoIn preparation for a new malloc implementation:
jasone [Thu, 12 Jan 2006 07:28:21 +0000 (07:28 +0000)]
In preparation for a new malloc implementation:

  * Add posix_memalign().

  * Move calloc() from calloc.c to malloc.c.  Add a calloc() implementation in
    rtld-elf in order to make the loader happy (even though calloc() isn't
    used in rtld-elf).

  * Add _malloc_prefork() and _malloc_postfork(), and use them instead of
    directly manipulating __malloc_lock.

Approved by: phk, markm (mentor)

18 years agoBuild shared library on behalf of bsnmpd.
ru [Thu, 12 Jan 2006 07:26:57 +0000 (07:26 +0000)]
Build shared library on behalf of bsnmpd.

18 years agoassert(3) is not used here.
obrien [Thu, 12 Jan 2006 02:53:42 +0000 (02:53 +0000)]
assert(3) is not used here.

18 years agoMove linux support to the linux section.
obrien [Thu, 12 Jan 2006 01:20:59 +0000 (01:20 +0000)]
Move linux support to the linux section.

18 years agoFix broken playback capabilities to prevent impending disaster.
ariff [Wed, 11 Jan 2006 23:22:57 +0000 (23:22 +0000)]
Fix broken playback capabilities to prevent impending disaster.
The minimum / maximum speed was way too low / high!

minspeed =   2000 - is this for real ?
maxspeed = 767999 - is this for real ?????

Wrap everything into 8000 - 48000 boundary, just to be safe.

MFC after: 3 days

18 years agoGrab the media from the passed in structure to put it into the
ambrisko [Wed, 11 Jan 2006 22:37:59 +0000 (22:37 +0000)]
Grab the media from the passed in structure to put it into the
global structure.

PR: bin/91399
Submitted by: Spencer Minear

18 years agoCreating memory file systems with softupdates enabled is pointless,
brooks [Wed, 11 Jan 2006 21:59:30 +0000 (21:59 +0000)]
Creating memory file systems with softupdates enabled is pointless,
don't do it.

PR: conf/85558
Submitted by: Ralf Wenk <RZ dash FreeBSD0605 at hs dash karlsruhe dot de>
MFC after: 5 days

18 years agoGrammar fix.
takawata [Wed, 11 Jan 2006 21:46:15 +0000 (21:46 +0000)]
Grammar fix.
Submitted by: joel@

18 years agoBe a little more read-only file system friendly when running the Linux
brooks [Wed, 11 Jan 2006 21:30:41 +0000 (21:30 +0000)]
Be a little more read-only file system friendly when running the Linux
ldconfig.  Build the cache in a temporary directory and only install it
if it's actually different that the installed one.

Also, use "cat tmp > real" to install the temporary file in the real
location to allow the real location to be a symlink to a writable
directory such as /var/run (where the file actually belongs).

MFC After: 5 days

18 years agoo Sort MLINKS.
maxim [Wed, 11 Jan 2006 21:28:55 +0000 (21:28 +0000)]
o Sort MLINKS.

18 years agoOk, I've created a test suite to avoid such regressions. Sorry for the noise.
flz [Wed, 11 Jan 2006 16:51:21 +0000 (16:51 +0000)]
Ok, I've created a test suite to avoid such regressions. Sorry for the noise.

Approved by: ssouhlal (implicit)

18 years ago- Fix another bug, it seems sometimes mail is sent to cvs-all but not cvs-ports.
flz [Wed, 11 Jan 2006 16:22:24 +0000 (16:22 +0000)]
- Fix another bug, it seems sometimes mail is sent to cvs-all but not cvs-ports.

Approved by: ssouhlal (implicit)

18 years ago- Fix search.
flz [Wed, 11 Jan 2006 16:00:38 +0000 (16:00 +0000)]
- Fix search.
- Fix author matching.

Approved by: ssouhlal (implicit)

18 years agoAdd the RB_NFIND() macro, which is useful for red-black tree searches
jasone [Wed, 11 Jan 2006 15:48:36 +0000 (15:48 +0000)]
Add the RB_NFIND() macro, which is useful for red-black tree searches
for which there may not be an exact match.

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

18 years agoAdd ktr(9) hooks to easier tracing of the netgraph item flow through
glebius [Wed, 11 Jan 2006 15:29:48 +0000 (15:29 +0000)]
Add ktr(9) hooks to easier tracing of the netgraph item flow through
netgraph.

18 years agoI wrote getnetconfig where I meant getnetpath in the previous revision.
ceri [Wed, 11 Jan 2006 13:57:15 +0000 (13:57 +0000)]
I wrote getnetconfig where I meant getnetpath in the previous revision.

18 years agoAdd a mobile phone known to work.
takawata [Wed, 11 Jan 2006 13:51:03 +0000 (13:51 +0000)]
Add a mobile phone known to work.

18 years agoCorrect insecure temporary file usage in texindex. [06:01]
cperciva [Wed, 11 Jan 2006 08:02:16 +0000 (08:02 +0000)]
Correct insecure temporary file usage in texindex. [06:01]
Correct insecure temporary file usage in ee. [06:02]
Correct a race condition when setting file permissions, sanitize file
names by default, and fix a buffer overflow when handling files
larger than 4GB in cpio. [06:03]
Fix an error in the handling of IP fragments in ipfw which can cause
a kernel panic. [06:04]

Security: FreeBSD-SA-06:01.texindex
Security: FreeBSD-SA-06:02.ee
Security: FreeBSD-SA-06:03.cpio
Security: FreeBSD-SA-06:04.ipfw

18 years ago- Locking fixes. Release lock while chn_intr().
ariff [Wed, 11 Jan 2006 08:02:15 +0000 (08:02 +0000)]
- Locking fixes. Release lock while chn_intr().
- Mark MPSAFE since most of the locking procedures already implemented.
- Turn on inverted external amplifier sense flag for selected boards.

Tested by: bland
MFC after: 1 week

18 years agoThe thr_new sysscall was already in libc, don't generate it.
davidxu [Wed, 11 Jan 2006 06:10:05 +0000 (06:10 +0000)]
The thr_new sysscall was already in libc, don't generate it.

18 years agoGet rid of the bogus IFP2FC() macro and use IFP2FWC(). IFP2FC()
brooks [Wed, 11 Jan 2006 05:37:21 +0000 (05:37 +0000)]
Get rid of the bogus IFP2FC() macro and use IFP2FWC().  IFP2FC()
attempted to cast a struct ifnet to a struct fw_com which resulted in
data corruption.

PR: kern/91307
Submitted by: Alex Semenyaka <alex at semenyaka do ru>
MFC After: 6 days

18 years ago- Remove netcat dependency by using fetch (not sure why i used nc
flz [Wed, 11 Jan 2006 02:53:54 +0000 (02:53 +0000)]
- Remove netcat dependency by using fetch (not sure why i used nc
in the first place).
- Add some XXX lines as a TODO.
- Add a cvs diff command to the generated commit script.
- Add cdiff/colordiff to the list of optional dependencies.
- Fix a problem when giving the commit mail without the headers to
mfc.pl -f, it should now work correctly.
- Bump version.

Approved by: ssouhlal (implicit)

18 years agoWhen deregistering a bus, attempt to flush out all outstanding
iedowse [Wed, 11 Jan 2006 02:06:08 +0000 (02:06 +0000)]
When deregistering a bus, attempt to flush out all outstanding
operations before returning. Point the bus at a dummy cam_sim
structure so that any CCBs will complete immediately with a
CAM_DEV_NOT_THERE status, and ensure that any xpt_schedule() calls
on the bus's devices will immediately call the peripheral's
periph_start() routine. Also repeat the async messages because
devices that were part of the way through being probed may appear
after the original AC_LOST_DEVICE was sent, and would otherwise
never go away.

These changes make it possible to deregister a bus and free the SIM
at most stages during bus probing without the usual crashes in
camisr(). In particular, plugging in a umass device and then
unplugging it as soon as the first probe messages appeared would
almost always result in a crash. Now the device just goes away with
a few CAM errors and all references to the CAM bus, target and
device are dropped correctly.

18 years agoThe interlock in taskqueue_terminate() is completely wrong for taskqueues
scottl [Wed, 11 Jan 2006 00:37:13 +0000 (00:37 +0000)]
The interlock in taskqueue_terminate() is completely wrong for taskqueues
that use spinlocks.  Remove it for now.

18 years agoSignificant performance improvements for the if_em driver:
scottl [Wed, 11 Jan 2006 00:30:25 +0000 (00:30 +0000)]
Significant performance improvements for the if_em driver:

- Only update the rx ring consumer pointer after running through the rx loop,
  not with each iteration through the loop.
- If possible, use a fast interupt handler instead of an ithread handler.  Use
  the interrupt handler to check and squelch the interrupt, then schedule a
  taskqueue to do the actual work.  This has three benefits:
  - Eliminates the 'interrupt aliasing' problem found in many chipsets by
    allowing the driver to mask the interrupt in the NIC instead of the
    OS masking the interrupt in the APIC.
  - Allows the driver to control the amount of work done in the interrupt
    handler.  This results in what I call 'adaptive polling', where you get
    the latency benefits of a quick response to interrupts with the
    interrupt mitigation and work partitioning of polling.  Polling is still
    an option in the driver, but I consider it orthogonal to this work.
  - Don't hold the driver lock in the RX handler.  The handler and all data
    associated is effectively serialized already.  This eliminates the cost of
    dropping and reaquiring the lock for every receieved packet.  The result
    is much lower contention for the driver lock, resulting in lower CPU usage
    and lower latency for interactive workloads.

The amount of work done in the taskqueue is controlled by the sysctl
dev.em.N.rx_processing_limit

and tunable
hw.em.rx_process_limit

Setting these to -1 effectively removes the limit.

The fast interrupt and taskqueue can be disabled by defining NO_EM_FASTINTR.
This work has been shown to increase fast-forwarding from ~570 kpps to
~750 kpps (note that the same NIC hardware seems unable to transmit more than
800 kpps, so this increase appears to be limited almost solely by the
hardware).  Gains have been shown in other workloads, ranging from better
performance to elimination of over-saturation livelocks.

Thanks to Andre Opperman for his time and resources from his network
performance project in performing much of the testing.  Thanks to Gleb
Smirnoff and Danny Braniss for their help in testing also.

18 years agoAdd references to fhopen, fhstat, getfh, lgetfh and fhstatfs.
grog [Tue, 10 Jan 2006 23:24:47 +0000 (23:24 +0000)]
Add references to fhopen, fhstat, getfh, lgetfh and fhstatfs.

Pointed out by: Antony Curtis <antony@mysql.com>

18 years agoDon't use the ALLOCNOW flag for tags that will only be used for static
scottl [Tue, 10 Jan 2006 22:55:35 +0000 (22:55 +0000)]
Don't use the ALLOCNOW flag for tags that will only be used for static
allocations.

18 years agoadd nfsclient/, nfs4client/, and rpc/ directories to the
rees [Tue, 10 Jan 2006 20:58:27 +0000 (20:58 +0000)]
add nfsclient/, nfs4client/, and rpc/ directories to the
top Makefile's rule to build a cscope database.

Submitted by: cel@citi.umich.edu
Approved by: alfred

18 years agoCorrect two trivial grammos.
schweikh [Tue, 10 Jan 2006 18:55:24 +0000 (18:55 +0000)]
Correct two trivial grammos.

18 years agoFix sort order.
takawata [Tue, 10 Jan 2006 18:39:23 +0000 (18:39 +0000)]
Fix sort order.
Pointed out by: ru

18 years ago- Update pretty print of multipath routes to better handle timeout of first
pav [Tue, 10 Jan 2006 16:48:52 +0000 (16:48 +0000)]
- Update pretty print of multipath routes to better handle timeout of first
  probe

  Before:
   5  *
    freebsd (195.250.137.134)  19.086 ms  24.694 ms

  After:
   5  * freebsd (195.250.137.134)  19.086 ms  24.694 ms

Fixes: bin/90098
Reported by: Jeremy Chadwick <freebsd@jdc.parodius.com>
Approved by: andre
MFC after: 1 day

18 years agoUpdate usage to reflect the fact that the -d -a now accepts -i <interface>.
brooks [Tue, 10 Jan 2006 16:15:58 +0000 (16:15 +0000)]
Update usage to reflect the fact that the -d -a now accepts -i <interface>.

Reminded by: ru
Pointy hat: brooks

18 years ago- Fix: documentation for -m option was inserted halfway thru the text of -l
pav [Tue, 10 Jan 2006 16:03:37 +0000 (16:03 +0000)]
- Fix: documentation for -m option was inserted halfway thru the text of -l
  option.

PR: docs/90359
Submitted by: Gavin Atkinson <gavin.atkinson@ury.york.ac.uk>
MFC after: 3 days

18 years ago- Xref mount_reiserfs(8)
pav [Tue, 10 Jan 2006 15:56:46 +0000 (15:56 +0000)]
- Xref mount_reiserfs(8)

PR: docs/90902
Submitted by: Scott Robbins <scottro@nyc.rr.com>
MFC after: 3 days

18 years agoHook ufoma(4) page up.
takawata [Tue, 10 Jan 2006 14:49:23 +0000 (14:49 +0000)]
Hook ufoma(4) page up.
Pointed out by: joel

18 years ago- Add a new MFC script that takes a message-id, a commit mail or a query string and
flz [Tue, 10 Jan 2006 14:33:20 +0000 (14:33 +0000)]
- Add a new MFC script that takes a message-id, a commit mail or a query string and
generates a patchset along with a commit message and a commit script if the user has
commit rights.
- Add a README file to give a quick description of each script's purpose.

Approved by: netchild

18 years agoDisable default write access by not setting the write community string.
harti [Tue, 10 Jan 2006 12:08:25 +0000 (12:08 +0000)]
Disable default write access by not setting the write community string.

PR: 91404, 91406

18 years agoThis commit was generated by cvs2svn to compensate for changes in r154184,
harti [Tue, 10 Jan 2006 11:59:31 +0000 (11:59 +0000)]
This commit was generated by cvs2svn to compensate for changes in r154184,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor fix: the routing table can change while we are fetching it from
harti [Tue, 10 Jan 2006 11:59:31 +0000 (11:59 +0000)]
Vendor fix: the routing table can change while we are fetching it from
the kernel. Instead of complaining if we get an ENOMEM (meaning it got
larger than our buffer is) reallocate the buffer and loop.

Submitted by: maxim

18 years agoThis commit was generated by cvs2svn to compensate for changes in r154182,
harti [Tue, 10 Jan 2006 11:52:51 +0000 (11:52 +0000)]
This commit was generated by cvs2svn to compensate for changes in r154182,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor fix: initialize the flag field of a newly created node to be 0.
harti [Tue, 10 Jan 2006 11:52:51 +0000 (11:52 +0000)]
Vendor fix: initialize the flag field of a newly created node to be 0.

Submitted by: jasone

18 years agoThis commit was generated by cvs2svn to compensate for changes in r154180,
harti [Tue, 10 Jan 2006 11:49:26 +0000 (11:49 +0000)]
This commit was generated by cvs2svn to compensate for changes in r154180,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor fix: make the default read and write communities NULL. This
harti [Tue, 10 Jan 2006 11:49:26 +0000 (11:49 +0000)]
Vendor fix: make the default read and write communities NULL. This
basically disables any access unless other strings are set in the config
file. Note, that there is no way to set the communities back to NULL once
they're set to something not NULL.

18 years agoThis commit was generated by cvs2svn to compensate for changes in r154178,
harti [Tue, 10 Jan 2006 11:47:56 +0000 (11:47 +0000)]
This commit was generated by cvs2svn to compensate for changes in r154178,
which included commits to RCS files with non-trunk default branches.

18 years agoVendor patch: fix a bug when parsing the include path.
harti [Tue, 10 Jan 2006 11:47:56 +0000 (11:47 +0000)]
Vendor patch: fix a bug when parsing the include path.

18 years agoAdd a (disabled) configuration line to enable the HOST-RESOURCES MIB.
harti [Tue, 10 Jan 2006 11:43:25 +0000 (11:43 +0000)]
Add a (disabled) configuration line to enable the HOST-RESOURCES MIB.

18 years agoMove the old BSD4.3 tty compatibility from (!BURN_BRIDGES && COMPAT_43)
phk [Tue, 10 Jan 2006 09:19:10 +0000 (09:19 +0000)]
Move the old BSD4.3 tty compatibility from (!BURN_BRIDGES && COMPAT_43)
to COMPAT_43TTY.

Add COMPAT_43TTY to NOTES and */conf/GENERIC

Compile tty_compat.c only under the new option.

Spit out
#warning "Old BSD tty API used, please upgrade."
if ioctl_compat.h gets #included from userland.

18 years agoMore thorough fixes to enable inverted external amplifier sense flag.
ariff [Tue, 10 Jan 2006 06:52:59 +0000 (06:52 +0000)]
More thorough fixes to enable inverted external amplifier sense flag.
Instead of dragging the entire ICH4/82801DB into this mess, select
only few boards based on pci subdevice / subvendor.

Tested by: Daisuke Orikasa <luxury-acura-3.5rl at nifty.com>
MFC after: 3 days

18 years agoAdd functions and macros and refactor code to make it easier to manage
scottl [Tue, 10 Jan 2006 06:31:12 +0000 (06:31 +0000)]
Add functions and macros and refactor code to make it easier to manage
fast taskqueues.  The following have been added:

TASKQUEUE_FAST_DEFINE() - create a global task queue.
    an arbitrary execution context.
TASKQUEUE_FAST_DEFINE_THREAD() - create a global taskqueue that uses a
    dedicated kthread.
taskqueue_create_fast() - create a local/private taskqueue.

These are all complimentary of the standard taskqueue functions.  They are
primarily useful for fast interrupt handlers that can only use spinlock for
synchronization.

I personally think that the taskqueue API is starting to get too narrow and
hairy, but fixing it will require a major redesign on the API.  Such a
redesign would be good but would break compatibility with FreeBSD 6.x, so
it really isn't desirable at this time.

Submitted by: sam

18 years ago/etc/crontab is similar enough to parse as correct if you run
brooks [Tue, 10 Jan 2006 05:56:32 +0000 (05:56 +0000)]
/etc/crontab is similar enough to parse as correct if you run
"crontab /etc/crontab", but not the same format due to the who field.
Add some limited anti-foot-shooting support and refuse to load
/etc/crontab as someone's crontab.  Users wishing shoot their foot in
this manner may copy /etc/crontab elsewhere. :)

MFC After: 1 week

18 years agoMention the -b flag in the SYNOPSIS.
brooks [Tue, 10 Jan 2006 05:44:11 +0000 (05:44 +0000)]
Mention the -b flag in the SYNOPSIS.

MFC after: 1 week

18 years agoWhen we give up on an interface, use the arp(8) command to remove all
brooks [Tue, 10 Jan 2006 05:33:33 +0000 (05:33 +0000)]
When we give up on an interface, use the arp(8) command to remove all
entries from the interface rather than using ifconfig's delete command.
This preserves non-dhclient configured addresses (though they are wiped
out when dhclient is restarted).

MFC after: 1 week

18 years agoRemove code describing pre 5.x system
imp [Tue, 10 Jan 2006 05:21:01 +0000 (05:21 +0000)]
Remove code describing pre 5.x system

18 years agoAllow the -i <interface> option with -d -a to allow all arp entries for
brooks [Tue, 10 Jan 2006 05:17:16 +0000 (05:17 +0000)]
Allow the -i <interface> option with -d -a to allow all arp entries for
a given interface to be removed.

18 years agoAllow users to add aliases to the interface.
brooks [Tue, 10 Jan 2006 04:53:20 +0000 (04:53 +0000)]
Allow users to add aliases to the interface.

PR: bin/87465 (different solution used)
MFC after: 1 week

18 years agoUse macro STATIC_LIB_REQUIRE to declare a symbol should be linked into
davidxu [Tue, 10 Jan 2006 04:53:03 +0000 (04:53 +0000)]
Use macro STATIC_LIB_REQUIRE to declare a symbol should be linked into
static binary.

18 years agoDon't free the ap object if it was never created.
scottl [Tue, 10 Jan 2006 01:55:17 +0000 (01:55 +0000)]
Don't free the ap object if it was never created.

Submitted by: jkim

18 years agoHook ufoma module to build.
takawata [Tue, 10 Jan 2006 00:54:18 +0000 (00:54 +0000)]
Hook ufoma module to build.

Pointed out by:thompsa

18 years agoAdd marker vnodes to ensure that all vnodes associated with the mount point are
tegge [Mon, 9 Jan 2006 20:42:19 +0000 (20:42 +0000)]
Add marker vnodes to ensure that all vnodes associated with the mount point are
iterated over when using MNT_VNODE_FOREACH.

Reviewed by: truckman

18 years agoFix -r causing SIGSEGV due to parse_args() being called before font_current is set.
flz [Mon, 9 Jan 2006 19:54:28 +0000 (19:54 +0000)]
Fix -r causing SIGSEGV due to parse_args() being called before font_current is set.

Approved by: krion
MFC after: 3 days

18 years agoIf the lock passed to getdirtybuf() is the softdep lock then the background
tegge [Mon, 9 Jan 2006 19:32:21 +0000 (19:32 +0000)]
If the lock passed to getdirtybuf() is the softdep lock then the background
write completed wakeup could be missed.  Close the race by grabbing the lock
normally used for protection of bp->b_xflags.

Reviewed by: truckman

18 years agoBroaden scope of softdep_worklist_busy rwlock protection of softdep processing
tegge [Mon, 9 Jan 2006 19:16:56 +0000 (19:16 +0000)]
Broaden scope of softdep_worklist_busy rwlock protection of softdep processing
to avoid some dependencies being missed by softdep_flushworklist().

Reviewed by: truckman

18 years agoAdd documents URL.
takawata [Mon, 9 Jan 2006 19:07:04 +0000 (19:07 +0000)]
Add documents URL.

18 years agoObtain mount point lock before restarting sync loop if vget() failed.
tegge [Mon, 9 Jan 2006 18:57:35 +0000 (18:57 +0000)]
Obtain mount point lock before restarting sync loop if vget() failed.

Reviewed by: truckman

18 years agoAdd man page for ufoma(4) in poor English and formatting.
takawata [Mon, 9 Jan 2006 18:56:11 +0000 (18:56 +0000)]
Add man page for ufoma(4) in poor English and formatting.

18 years agoAdd -K option to pkg_add and pkg_info.
flz [Mon, 9 Jan 2006 18:27:21 +0000 (18:27 +0000)]
Add -K option to pkg_add and pkg_info.
This option saves packages to PKGDIR (if defined or current directory by default)
as they are downloaded.
Silent a warning when -n is used and package has a +DISPLAY file.

Approved by: krion
MFC after: 1 week

18 years agoo Fix typo in the define: s/MRAK_INT_GEN/MARK_INT_GEN/. The typo
maxim [Mon, 9 Jan 2006 18:07:06 +0000 (18:07 +0000)]
o Fix typo in the define: s/MRAK_INT_GEN/MARK_INT_GEN/.  The typo
was harmless because the define is not used in coda_vfsops.c.

Submitted by: Hugo Meiland

18 years agoAdd ufoma drivers in 'files', too.
takawata [Mon, 9 Jan 2006 17:49:21 +0000 (17:49 +0000)]
Add ufoma drivers in 'files', too.

18 years agoAdd FOMA (NTT DoCoMo 3G mobile phone system) driver.
takawata [Mon, 9 Jan 2006 17:46:36 +0000 (17:46 +0000)]
Add FOMA (NTT DoCoMo 3G mobile phone system) driver.
This is based on MCPC USB mobile phone guide line (MCPC-GL005)
Some other 3G system or so will work with this driver.
Kyocera PHS terminal (a.k.a. Kyopon) is known to work, which
is now supported by umodem(4) driver.

18 years agoUpdate monitoring support:
sam [Mon, 9 Jan 2006 17:13:20 +0000 (17:13 +0000)]
Update monitoring support:
o record tsf in tx+rx frames
o switch from raw rssi to dbm for signal data and record both
  signal and noise floor data (hacked for now to assume a fixed
  noise floor; is correct with new hal)
o add monpass sysctl to control which rx'd frames are passed
  up with errors; especially useful to see frames with CRC errors
o mark 'd packets w/ a CRC error with radiotap's BADFCS flag

Also add placeholder code for calibrating the noise floor when
using newer hals.

Reviewed by: avatar
MFC after: 1 week

18 years agoadd flag to tag frames w/ a known bad FCS
sam [Mon, 9 Jan 2006 17:04:56 +0000 (17:04 +0000)]
add flag to tag frames w/ a known bad FCS

Obtained from: netbsd
MFC after: 1 week

18 years agoForced commit because I lost the commit message in the previous commit:
harti [Mon, 9 Jan 2006 13:01:26 +0000 (13:01 +0000)]
Forced commit because I lost the commit message in the previous commit:

This is the implementation of the HOSTRES MIB that was developed by
Victor during SoC 2005 with some stylification by philip@ and harti@.

Submitted by: Victor Cruceru <soc-victor@freebsd.org>
Reviewed by: philip, harti

18 years agoKill empty line.
joel [Mon, 9 Jan 2006 12:51:45 +0000 (12:51 +0000)]
Kill empty line.

18 years agoConnect the HOSTRES MIB to the build.
harti [Mon, 9 Jan 2006 12:34:07 +0000 (12:34 +0000)]
Connect the HOSTRES MIB to the build.

18 years agomsg
harti [Mon, 9 Jan 2006 12:33:45 +0000 (12:33 +0000)]
msg

18 years agoTurn on inverted external amplifier sense flags for ICH4/82801DB.
ariff [Mon, 9 Jan 2006 11:20:37 +0000 (11:20 +0000)]
Turn on inverted external amplifier sense flags for ICH4/82801DB.

PR: kern/66422, kern/75687, kern/84471
MFC after: 2 days