]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
20 years agoReplace deprecated NG_NODELEN with the new NG_NODESIZ. There is one
Hartmut Brandt [Mon, 26 Jan 2004 15:19:43 +0000 (15:19 +0000)]
Replace deprecated NG_NODELEN with the new NG_NODESIZ. There is one
problem here still to be solved: the sockaddr_hci has still a 16 byte
field for the node name. The code currently does not correctly use the
length field in the sockaddr to handle the address length, so
node names get truncated to 15 characters when put into a sockaddr_hci.

20 years agoAvoid dereferencing null pointers in fsck_ffs. (pfatal may return,
Colin Percival [Mon, 26 Jan 2004 15:05:30 +0000 (15:05 +0000)]
Avoid dereferencing null pointers in fsck_ffs. (pfatal may return,
so it isn't a safe way of handling [mc]alloc failures.)

PR: misc/61800
Approved by: rwatson (mentor)

20 years agoGet rid of the old *LEN constants in favour of the new *SIZ constants
Hartmut Brandt [Mon, 26 Jan 2004 14:57:49 +0000 (14:57 +0000)]
Get rid of the old *LEN constants in favour of the new *SIZ constants
that also include the trailing \0 byte.

20 years agoDefine the new command NGM_SOURCE_START_NOW to allow generation of
Hartmut Brandt [Mon, 26 Jan 2004 14:54:39 +0000 (14:54 +0000)]
Define the new command NGM_SOURCE_START_NOW to allow generation of
traffic for non-ethernet hooks. This commit should have been packaged
with the commit to ng_source.c.

20 years agoMake ng_source to work with non-ethernet interfaces. We do this by
Hartmut Brandt [Mon, 26 Jan 2004 14:53:16 +0000 (14:53 +0000)]
Make ng_source to work with non-ethernet interfaces. We do this by
introducing a START_NOW command. This command does not send
and GET_IFINDEX message downstream (to wait for the response from
the ETHERNET node), but directly starts the sending process. This allows
one to generate traffic as input for any hook on any node.

20 years agoDeclare a function to silence a warning.
Hartmut Brandt [Mon, 26 Jan 2004 14:48:21 +0000 (14:48 +0000)]
Declare a function to silence a warning.

20 years agoShould use the non-locking versions of the ifqueue macros to
Hartmut Brandt [Mon, 26 Jan 2004 14:46:35 +0000 (14:46 +0000)]
Should use the non-locking versions of the ifqueue macros to
fiddle around with private queues, because their mutex is not
needed. All this processing should be protected by the netgraph
locking.

20 years agoReplace a call to bzero() with an M_ZERO flag. Replace the MALLOC() with
Hartmut Brandt [Mon, 26 Jan 2004 14:44:36 +0000 (14:44 +0000)]
Replace a call to bzero() with an M_ZERO flag. Replace the MALLOC() with
malloc().

20 years agoThe version in the type description must be the ABI version, not
Hartmut Brandt [Mon, 26 Jan 2004 14:14:09 +0000 (14:14 +0000)]
The version in the type description must be the ABI version, not
the netgraph version.

Correct the return type of a function: it wants to return an error
code, so it cannot be void.

20 years agoGet rid of the deprecated *LEN constants in favour of the new
Hartmut Brandt [Mon, 26 Jan 2004 14:05:31 +0000 (14:05 +0000)]
Get rid of the deprecated *LEN constants in favour of the new
*SIZ constants that include the trailing \0 byte.

20 years agoMake a message less scary (based on user feedback)
Eivind Eklund [Mon, 26 Jan 2004 13:45:21 +0000 (13:45 +0000)]
Make a message less scary (based on user feedback)

20 years agoDocument the interface method if_input().
Yaroslav Tykhiy [Mon, 26 Jan 2004 12:51:00 +0000 (12:51 +0000)]
Document the interface method if_input().

20 years agomdoc cleanup.
Dag-Erling Smørgrav [Mon, 26 Jan 2004 12:50:30 +0000 (12:50 +0000)]
mdoc cleanup.

Submitted by: ru

20 years agoAdd support for WCCPv2. It should be enablem manually using link2
Maxim Sobolev [Mon, 26 Jan 2004 12:33:56 +0000 (12:33 +0000)]
Add support for WCCPv2. It should be enablem manually using link2
ifconfig(8) flag since header for version 2 is the same but IP payload
is prepended with additional 4-bytes field.

Inspired by: Roman Synyuk <roman@univ.kiev.ua>
MFC after: 2 weeks

20 years agoFixed some style bugs.
Yoshihiro Takahashi [Mon, 26 Jan 2004 12:28:40 +0000 (12:28 +0000)]
Fixed some style bugs.

20 years agoGet rid of the deprecated *LEN constants and use the new *SIZ
Hartmut Brandt [Mon, 26 Jan 2004 12:24:07 +0000 (12:24 +0000)]
Get rid of the deprecated *LEN constants and use the new *SIZ
(that include the trailing \0) constants instead.

20 years ago(whilespace-only)
Maxim Sobolev [Mon, 26 Jan 2004 12:21:59 +0000 (12:21 +0000)]
(whilespace-only)

Kill trailing spaces.

20 years agoAdd support for virtual interfaces. These have no phy chip and, hence, we
Hartmut Brandt [Mon, 26 Jan 2004 12:17:49 +0000 (12:17 +0000)]
Add support for virtual interfaces. These have no phy chip and, hence, we
need to handle interfaces without phy specially.

20 years agoAdd a device type for virtual interfaces.
Hartmut Brandt [Mon, 26 Jan 2004 12:13:11 +0000 (12:13 +0000)]
Add a device type for virtual interfaces.

20 years agoClearly state that upon SIOCSIFFLAGS, ifp->if_flags is updated
Yaroslav Tykhiy [Mon, 26 Jan 2004 12:08:23 +0000 (12:08 +0000)]
Clearly state that upon SIOCSIFFLAGS, ifp->if_flags is updated
at the common ifioctl() level.

20 years agoAdd an ATM sub-type for virtual interfaces.
Hartmut Brandt [Mon, 26 Jan 2004 11:52:32 +0000 (11:52 +0000)]
Add an ATM sub-type for virtual interfaces.

20 years agoExtend the description of SIOCSIFCAP.
Yaroslav Tykhiy [Mon, 26 Jan 2004 11:48:26 +0000 (11:48 +0000)]
Extend the description of SIOCSIFCAP.

20 years agoNew -e flag, modifies the behavior of the -u uid flag to use the
Diomidis Spinellis [Mon, 26 Jan 2004 11:11:36 +0000 (11:11 +0000)]
New -e flag, modifies the behavior of the -u uid flag to use the
effective user id, instead of the real user id.

MFC after: 2 weeks

20 years agoUpdate the upgrade instructions to account for the fact, that
Hartmut Brandt [Mon, 26 Jan 2004 10:28:31 +0000 (10:28 +0000)]
Update the upgrade instructions to account for the fact, that
there is no need anymore to re-create files. Also give a hint to
look for changes needed to the Makefiles.

20 years agoMake this WARNS=3 clean.
Ruslan Ermilov [Mon, 26 Jan 2004 10:27:18 +0000 (10:27 +0000)]
Make this WARNS=3 clean.

20 years agoAdd a note to say that ULE is now the default scheduler in GENERIC.
Alex Dupre [Mon, 26 Jan 2004 09:57:04 +0000 (09:57 +0000)]
Add a note to say that ULE is now the default scheduler in GENERIC.

PR: docs/61870
Submitted by: Xin LI <delphij@frontfree.net>
Approved by: imp

20 years agoFix problem where kgdb misses trap frames.
David E. O'Brien [Mon, 26 Jan 2004 09:18:47 +0000 (09:18 +0000)]
Fix problem where kgdb misses trap frames.

Reported by: iedowse

20 years agoUse proper name.
David E. O'Brien [Mon, 26 Jan 2004 09:17:26 +0000 (09:17 +0000)]
Use proper name.

20 years agoAvoid possible panic on shutdown: if there are still some devices
Bill Paul [Mon, 26 Jan 2004 08:36:18 +0000 (08:36 +0000)]
Avoid possible panic on shutdown: if there are still some devices
attached when shutting down, kill our kthreads, but don't destroy
the mutex pool and uma zone resources since the driver shutdown
routine may need them later.

20 years agoRevert changes of rev 1.55; replace with strong encouragement to
Wes Peters [Mon, 26 Jan 2004 07:55:30 +0000 (07:55 +0000)]
Revert changes of rev 1.55; replace with strong encouragement to
read fetch(3) as well.

Requested by: des (maintainer)

20 years agoFix sorting in the device list
Christian Brueffer [Mon, 26 Jan 2004 07:21:28 +0000 (07:21 +0000)]
Fix sorting in the device list

20 years agoPrevious commit did the reverse of what the log message said (and of
Dag-Erling Smørgrav [Mon, 26 Jan 2004 07:09:12 +0000 (07:09 +0000)]
Previous commit did the reverse of what the log message said (and of
what I intended to do).  Really remove my middle name.

20 years agostyle.Makefile(5)
David E. O'Brien [Mon, 26 Jan 2004 06:10:26 +0000 (06:10 +0000)]
style.Makefile(5)

20 years agoFollow existing GDB style for frame_info vars.
David E. O'Brien [Mon, 26 Jan 2004 06:07:33 +0000 (06:07 +0000)]
Follow existing GDB style for frame_info vars.

20 years agoRemove trailing white space.
David E. O'Brien [Mon, 26 Jan 2004 06:05:04 +0000 (06:05 +0000)]
Remove trailing white space.

20 years agoUse malloc() instead of alloca() to allocate a (potentially) large buffer
Stephen McKay [Mon, 26 Jan 2004 04:27:22 +0000 (04:27 +0000)]
Use malloc() instead of alloca() to allocate a (potentially) large buffer
for storing the "diff -n" output.  Some files (eg ports/INDEX,v) are too
big nowadays to fit on the stack.

Submitted by: Stephen Montgomery-Smith <stephen@math.missouri.edu>

20 years agoThe log file is more useful when line buffered.
Stephen McKay [Mon, 26 Jan 2004 04:08:43 +0000 (04:08 +0000)]
The log file is more useful when line buffered.

Submitted by: Stephen Montgomery-Smith <stephen@math.missouri.edu>

20 years agoBe much more strict about parsing tagged log messages from /dev/klog;
Ian Dowse [Mon, 26 Jan 2004 00:46:46 +0000 (00:46 +0000)]
Be much more strict about parsing tagged log messages from /dev/klog;
if the line doesn't match ^<%d>, then treat it as a regular kernel
printf line. Previously if a kernel printf message started with "<"
it would be interpreted as a log message, often with LOG_EMERG
level. This was triggered by some printfs in sys/dev/aic7xxx/, and
can also happen with the partial lines that result if syslogd cannot
keep up with the rate of arrival of kernel messages.

Reviewed by: dwmalone
MFC after: 1 week

20 years agoFix a panic in dsp_clone when trying to access a sound
Mathew Kanner [Sun, 25 Jan 2004 22:46:22 +0000 (22:46 +0000)]
Fix a panic in dsp_clone when trying to access a sound
device that doesn't exists.  I'm using my discretion and
committing without mentor approval since Seigo is away.

Noticed by: Maxime Henrion <mux@freebsd.org>

20 years agoTrim IDE/ATA device list in favor of the list in the ata.4 manpage
Christian Brueffer [Sun, 25 Jan 2004 21:57:57 +0000 (21:57 +0000)]
Trim IDE/ATA device list in favor of the list in the ata.4 manpage

20 years agoNew release note: PFIL_HOOKS in GENERIC.
Bruce A. Mah [Sun, 25 Jan 2004 21:54:24 +0000 (21:54 +0000)]
New release note:  PFIL_HOOKS in GENERIC.

20 years agoCleanup CFLAGS in libsdp(3) Makefile
Maksim Yevmenkin [Sun, 25 Jan 2004 21:33:59 +0000 (21:33 +0000)]
Cleanup CFLAGS in libsdp(3) Makefile
Connect sdpd(8) to the build

Reviewed by: ru, imp (mentor)

20 years ago- Existing code would ignore pccard_ether_delay when more then 9 seconds
Doug Ambrisko [Sun, 25 Jan 2004 19:52:16 +0000 (19:52 +0000)]
- Existing code would ignore pccard_ether_delay when more then 9 seconds
- If there was an exiting dhclient running on the same interface as the
  new iface that left and returned then dhclient would be told to use the
  same interface twice.  Dhclient would fail and exit after getting
  confused.  Use "sort -u" on them to ensure no duplicates.
  This is a mostly a race condition on suspend and resume and how things
  happen to occur.
- Check for netmask being set on an interface rather then up.  An interface
  can be up but not configured.

Reviewed by: mbr

20 years agoFix the case where .OBJDIR != .CURDIR.
Dag-Erling Smørgrav [Sun, 25 Jan 2004 19:11:43 +0000 (19:11 +0000)]
Fix the case where .OBJDIR != .CURDIR.

20 years agoUpdate Eterm entry to Eterm 0.9.2
Olivier Houchard [Sun, 25 Jan 2004 19:04:50 +0000 (19:04 +0000)]
Update Eterm entry to Eterm 0.9.2

Submitted by: Geraud CONTINSOUZAS <geraud.continsouzas@wanadoo.fr>
MFC after: 3 days

20 years agoThis setup was used for amd64 and powerpc before they reached the point
Dag-Erling Smørgrav [Sun, 25 Jan 2004 19:04:37 +0000 (19:04 +0000)]
This setup was used for amd64 and powerpc before they reached the point
where a normal buildworld would succeed.  It is no longer required.

20 years agoWhen aborting fork() due to a failure, if using MAC, make sure to clean
Robert Watson [Sun, 25 Jan 2004 18:42:18 +0000 (18:42 +0000)]
When aborting fork() due to a failure, if using MAC, make sure to clean
up the p_label field.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, McAfee Research

20 years agoavoid duplicate free.
Hajimu UMEMOTO [Sun, 25 Jan 2004 17:18:12 +0000 (17:18 +0000)]
avoid duplicate free.
(though KAME doesn't decide how to fix it yet, I once commit it
to be in time for 5.2.1-RELEASE.)

Submitted by: itojun

20 years agoAdd SIS964 to the list of supported chipsets
Christian Brueffer [Sun, 25 Jan 2004 16:13:49 +0000 (16:13 +0000)]
Add SIS964 to the list of supported chipsets

Submitted by: bmah

20 years agoBack out rev. 1.40
Christian Brueffer [Sun, 25 Jan 2004 16:12:09 +0000 (16:12 +0000)]
Back out rev. 1.40

The hardware notes were wrong WRT supported chipsets (northbridges are not
ata related)

Submitted by: bmah

20 years agoFixed some style bugs (insertion sort errors, tab lossage, and ornation
Bruce Evans [Sun, 25 Jan 2004 15:27:23 +0000 (15:27 +0000)]
Fixed some style bugs (insertion sort errors, tab lossage, and ornation
of EOF).

20 years agoRegister the uart(4)'s spin lock with witness(4).
Ruslan Ermilov [Sun, 25 Jan 2004 15:04:37 +0000 (15:04 +0000)]
Register the uart(4)'s spin lock with witness(4).

20 years agoCorrect the name of a loader(8) tunable.
Ruslan Ermilov [Sun, 25 Jan 2004 14:03:55 +0000 (14:03 +0000)]
Correct the name of a loader(8) tunable.

20 years agoUpdate the "overview of FreeBSD changes to OpenSSH-portable" to reflect
Dag-Erling Smørgrav [Sun, 25 Jan 2004 13:09:56 +0000 (13:09 +0000)]
Update the "overview of FreeBSD changes to OpenSSH-portable" to reflect
reality.

20 years agoRemove trailing whitespace.
Dag-Erling Smørgrav [Sun, 25 Jan 2004 12:32:56 +0000 (12:32 +0000)]
Remove trailing whitespace.

20 years agoReplace description of the mutex profiling code with a reference to
Dag-Erling Smørgrav [Sun, 25 Jan 2004 12:31:59 +0000 (12:31 +0000)]
Replace description of the mutex profiling code with a reference to
the newly committed manual page.

20 years agoAdd more implentation notes based on the comments in sys/conf/NOTES.
Dag-Erling Smørgrav [Sun, 25 Jan 2004 12:30:35 +0000 (12:30 +0000)]
Add more implentation notes based on the comments in sys/conf/NOTES.

20 years agoFixed a bug that previous revision has introduced -- we missed
Ruslan Ermilov [Sun, 25 Jan 2004 12:28:36 +0000 (12:28 +0000)]
Fixed a bug that previous revision has introduced -- we missed
the double quotes ("" and '') as a separate argument.

Reported by: ache

The fix in this and previous revisions combined is functionally
equivalent to the below patch against rev. 1.27 but the code is
now much easier to follow:

%%%
Index: str.c
===================================================================
RCS file: /home/ncvs/src/usr.bin/make/str.c,v
retrieving revision 1.27
diff -u -r1.27 str.c
--- str.c 28 Oct 2002 23:33:57 -0000 1.27
+++ str.c 25 Jan 2004 12:09:21 -0000
@@ -168,7 +168,7 @@
  inquote = (char) ch;
  /* Don't miss "" or '' */
  if (start == NULL && p[1] == inquote) {
- start = t + 1;
+ start = t;
  break;
  }
  }
%%%

20 years agoAdd a cross-reference to MUTEX_PROFILING(9).
Dag-Erling Smørgrav [Sun, 25 Jan 2004 12:24:23 +0000 (12:24 +0000)]
Add a cross-reference to MUTEX_PROFILING(9).

20 years agoAdd a manual page for the mutex profiling code.
Dag-Erling Smørgrav [Sun, 25 Jan 2004 12:23:53 +0000 (12:23 +0000)]
Add a manual page for the mutex profiling code.

20 years agoI don't normally use my middle name, so remove it from attributions in
Dag-Erling Smørgrav [Sun, 25 Jan 2004 11:39:42 +0000 (11:39 +0000)]
I don't normally use my middle name, so remove it from attributions in
man pages (though not from copyright notices).  While I'm here, add email
addresses where appropriate.

20 years agoDhclient dumps core on suspend/resume cycles. If the client isn't active
Martin Blapp [Sun, 25 Jan 2004 11:30:50 +0000 (11:30 +0000)]
Dhclient dumps core on suspend/resume cycles.  If the client isn't active
then we need to go to the reboot state or in state_bound it will core on
the de-reference of client -> active -> options since client -> active = NULL.

While we are here, fix the indentation.

Submitted by: Doug Ambrisko <ambrisko@ambrisko.com>

20 years agoMove LongRun support out of identcpu.c, where it hardly belongs, into its
Maxim Sobolev [Sun, 25 Jan 2004 09:59:16 +0000 (09:59 +0000)]
Move LongRun support out of identcpu.c, where it hardly belongs, into its
own file and make it opt-in, not mandatory, depending on CPU_ENABLE_LONGRUN
config(8) option.

Discussed with: nate
MFC after:      2 weeks

20 years agoMove LongRun support out of identcpu.c, where it hardly belongs, into its
Maxim Sobolev [Sun, 25 Jan 2004 09:59:15 +0000 (09:59 +0000)]
Move LongRun support out of identcpu.c, where it hardly belongs, into its
own file and make it opt-in, not mandatory, depending on CPU_ENABLE_LONGRUN
config(8) option.

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:
Discussed with: nate
MFC after:      2 weeks

20 years ago - sched_strict has been dead for a long time now. Get rid of it.
Jeff Roberson [Sun, 25 Jan 2004 08:58:14 +0000 (08:58 +0000)]
 - sched_strict has been dead for a long time now.  Get rid of it.

20 years ago - Clean up KASSERTS.
Jeff Roberson [Sun, 25 Jan 2004 08:57:38 +0000 (08:57 +0000)]
 - Clean up KASSERTS.

20 years ago - Correct function names listed in KASSERTs. These were copied from other
Jeff Roberson [Sun, 25 Jan 2004 08:21:46 +0000 (08:21 +0000)]
 - Correct function names listed in KASSERTs.  These were copied from other
   code and it was sloppy of me not to adjust these sooner.

20 years ago - Now that both schedulers support temporary cpu pinning use this rather
Jeff Roberson [Sun, 25 Jan 2004 08:04:45 +0000 (08:04 +0000)]
 - Now that both schedulers support temporary cpu pinning use this rather
   than the switchin functions to guarantee that we're operating with the
   correct tlb entry.
 - Remove the post copy/zero tlb invalidations.  It is faster to invalidate
   an entry that is known to exist and so it is faster to invalidate after
   use.  However, some architectures implement speculative page table
   prefetching so we can not be guaranteed that the invalidated entry is still
   invalid when we re-enter any of these functions.  As a result of this we
   must always invalidate before use to be safe.

20 years ago - Implement cpu pinning and binding. This is acomplished by keeping a per-
Jeff Roberson [Sun, 25 Jan 2004 08:00:04 +0000 (08:00 +0000)]
 - Implement cpu pinning and binding.  This is acomplished by keeping a per-
   cpu run queue that is only used for pinned or bound threads.

Submitted by: Chris Bradfield <chrisb@ation.org>

20 years ago - Use a unique string for the sched_setup SYSINIT and rename sched_setup to
Jeff Roberson [Sun, 25 Jan 2004 07:49:45 +0000 (07:49 +0000)]
 - Use a unique string for the sched_setup SYSINIT and rename sched_setup to
   synch_setup.  The schedulers use the sched_setup function name.

20 years ago - Compile 4BSD in LINT since ULE will be tested by GENERIC kernel builds.
Jeff Roberson [Sun, 25 Jan 2004 07:47:57 +0000 (07:47 +0000)]
 - Compile 4BSD in LINT since ULE will be tested by GENERIC kernel builds.
 - Fix the formatting on the ULE options line, I didn't notice that a space
   was used normally.

Reported by: bde

20 years agoDescribe EOVERFLOW case
Andrey A. Chernov [Sun, 25 Jan 2004 06:24:08 +0000 (06:24 +0000)]
Describe EOVERFLOW case

20 years agoNew release notes: SCHED_ULE default, cdboot buggy BIOS workaround,
Bruce A. Mah [Sun, 25 Jan 2004 06:23:10 +0000 (06:23 +0000)]
New release notes:  SCHED_ULE default, cdboot buggy BIOS workaround,
dc(4) sparc64/OFW fix, sk(4) multicast fix, IPsec bugfix, ata(4)
bugfixes, GEOM fix, NFSv4 panic fix, NSS large groups fix, CVS
security fix.

Most (but not all) of these are 5.2-RELEASE errata items.

20 years agoMFi386 revision 1.230
Alan Cox [Sun, 25 Jan 2004 06:19:37 +0000 (06:19 +0000)]
MFi386 revision 1.230
 - Move smp_topology to subr_smp.c so that it is defined on all architectures.

20 years agos/freebsd.org/FreeBSD.org/
Bruce A. Mah [Sun, 25 Jan 2004 05:08:44 +0000 (05:08 +0000)]
s/freebsd.org/FreeBSD.org/

20 years ago - Don't define DETECT_DEADLOCK. I don't know that this code has detected
Jeff Roberson [Sun, 25 Jan 2004 05:03:14 +0000 (05:03 +0000)]
 - Don't define DETECT_DEADLOCK.  I don't know that this code has detected
   a deadlock in several years.  Furthermore, the IPI code is currently
   protected by a seperate spinlock.  This only served to make IPIs twice as
   expensive as they had to be which severely slowed down the IPI heavy ULE
   scheduler.

20 years ago - Add a flags parameter to mi_switch. The value of flags may be SW_VOL or
Jeff Roberson [Sun, 25 Jan 2004 03:54:52 +0000 (03:54 +0000)]
 - Add a flags parameter to mi_switch.  The value of flags may be SW_VOL or
   SW_INVOL.  Assert that one of these is set in mi_switch() and propery
   adjust the rusage statistics.  This is to simplify the large number of
   users of this interface which were previously all required to adjust the
   proper counter prior to calling mi_switch().  This also facilitates more
   switch and locking optimizations.
 - Change all callers of mi_switch() to pass the appropriate paramter and
   remove direct references to the process statistics.

20 years agoAdd some basic support for measuring sleep mutex contention to the
Robert Watson [Sun, 25 Jan 2004 01:59:27 +0000 (01:59 +0000)]
Add some basic support for measuring sleep mutex contention to the
mutex profiling code.  As with existing mutex profiling, measurement
is done with respect to mtx_lock() instances in the code, as opposed
to specific mutexes.  In particular, measure two things:

(1) Lock contention.  How often did this mtx_lock() call get made and
    have to sleep (or almost sleep) waiting for the lock.  This helps
    identify the "victims" of contention.

(2) Hold contention.  How often, while the lock was held by a thread
    as a result of this mtx_lock(), did another thread try to acquire
    the same mutex.  This helps identify the causes of contention.

I'm currently exploring adding measurement of "time waited for the
lock", but the current implementation has proven useful to me so far
so I figured I'd commit it so others could try it out.  Note that this
increases the size of mutexes when MUTEX_PROFILING is enabled, so you
might find you need to further bump UMA_BOOT_PAGES.  Fixes welcome.

The once over: des, others

20 years agoCorrect KASSERT() in ndis_destroy(): ndis_mtx is a pointer now.
Bill Paul [Sun, 25 Jan 2004 00:13:07 +0000 (00:13 +0000)]
Correct KASSERT() in ndis_destroy(): ndis_mtx is a pointer now.
Also add KASSERT() for ndis_intrmtx().

20 years agoAdd Sandberg USB to Serial Link (model number 133-08) to the list of
Simon L. B. Nielsen [Sat, 24 Jan 2004 22:12:06 +0000 (22:12 +0000)]
Add Sandberg USB to Serial Link (model number 133-08) to the list of
supported devices.

MFC after: 1 week

20 years agoDeal with MOD_FREQUENCY before MOD_OFFSET because the latter is the
Poul-Henning Kamp [Sat, 24 Jan 2004 21:48:43 +0000 (21:48 +0000)]
Deal with MOD_FREQUENCY before MOD_OFFSET because the latter is the
one which runs the actual update.  This fixes a bug where there were
a delay in applying the frequency adjustment.  In extreme cases this
could result in marginal stability of the kernel-pll.

20 years ago - Remove local changes that leaked into my last commit.
Jeff Roberson [Sat, 24 Jan 2004 21:45:25 +0000 (21:45 +0000)]
 - Remove local changes that leaked into my last commit.

Spotted by: juli

20 years ago - Recruit some new ULE users by making it the default scheduler in GENERIC.
Jeff Roberson [Sat, 24 Jan 2004 21:38:52 +0000 (21:38 +0000)]
 - Recruit some new ULE users by making it the default scheduler in GENERIC.
   ULE will be in a probationary period to determine whether it will be left
   as the default in 5.3 which would likely mean the rest of the 5.x series.

20 years ago - ULE is not exactly experimental anymore. Change some comments and enable
Jeff Roberson [Sat, 24 Jan 2004 21:32:53 +0000 (21:32 +0000)]
 - ULE is not exactly experimental anymore.  Change some comments and enable
   it in LINT.

20 years ago1. Statically initialize swap_pager_full and swap_pager_almost_full to the
Alan Cox [Sat, 24 Jan 2004 21:31:06 +0000 (21:31 +0000)]
1. Statically initialize swap_pager_full and swap_pager_almost_full to the
   full state.  (When swap is added their state will change appropriately.)
2. Set swap_pager_full and swap_pager_almost_full to the full state when
   the last swap device is removed.
Combined these changes eliminate nonsense messages from the kernel on swap-
less machines.

Item 2 submitted by: Divacky Roman <xdivac02@stud.fit.vutbr.cz>
Prodding by: phk

20 years agoRegen after rev. 1.161 of usbdevs.
Maxim Sobolev [Sat, 24 Jan 2004 21:24:36 +0000 (21:24 +0000)]
Regen after rev. 1.161 of usbdevs.

20 years agoAdd support for Crystalfontz CFA-631 USB LCD (uftdi(4) driver).
Maxim Sobolev [Sat, 24 Jan 2004 21:23:12 +0000 (21:23 +0000)]
Add support for Crystalfontz CFA-631 USB LCD (uftdi(4) driver).

For some very unclear reason this device contains a FTDI 8U232AM USB->COM
adapter, but reports different device id than original 8U232AM. At the same
time, it reports vendor id of FTDI.

Sponsored by: Porta Software Ltd
MFC after: 2 weeks

20 years ago- Move performance-controlling sysctls into hw.p4tcc.* tree;
Maxim Sobolev [Sat, 24 Jan 2004 21:13:13 +0000 (21:13 +0000)]
- Move performance-controlling sysctls into hw.p4tcc.* tree;

Suggested by:   nate

- get rid of "magick" values in code and make sysctl's reflecting reality
  on processor versions which have one or another frequency "forbidden"
  due to errata.

MFC after:      2 weeks

20 years ago- Move performance-controlling sysctls into hw.p4tcc.* tree;
Maxim Sobolev [Sat, 24 Jan 2004 21:13:12 +0000 (21:13 +0000)]
- Move performance-controlling sysctls into hw.p4tcc.* tree;

Suggested by:   nate

- get rid of "magick" values in code and make sysctl's reflecting reality
  on processor versions which have one or another frequency "forbidden"
  due to errata.

PR:
Submitted by:
Reviewed by:
Approved by:
Obtained from:
MFC after:      2 weeks

20 years agoMove the test used to determine whether IPFilter is loaded or not
Maxime Henrion [Sat, 24 Jan 2004 20:40:11 +0000 (20:40 +0000)]
Move the test used to determine whether IPFilter is loaded or not
into its own function to avoid a small duplication of code.

20 years agoRename the makefile variable MIBS to BMIBS so that it does not
Hartmut Brandt [Sat, 24 Jan 2004 20:14:03 +0000 (20:14 +0000)]
Rename the makefile variable MIBS to BMIBS so that it does not
conflict with the environment variable MIBS that is used by net-snmp.

20 years agoRename the MIBS makefile variable to BMIBS so that it does not conflict
Hartmut Brandt [Sat, 24 Jan 2004 20:12:30 +0000 (20:12 +0000)]
Rename the MIBS makefile variable to BMIBS so that it does not conflict
with the environment variable MIBS that is used by net-snmp.

20 years ago - Move smp_topology to subr_smp.c so that it is defined on all architectures.
Jeff Roberson [Sat, 24 Jan 2004 19:52:48 +0000 (19:52 +0000)]
 - Move smp_topology to subr_smp.c so that it is defined on all architectures.

20 years agoo Pass a correct argument to errx(3).
Maxim Konovalov [Sat, 24 Jan 2004 19:20:09 +0000 (19:20 +0000)]
o Pass a correct argument to errx(3).

PR: bin/61846
Submitted by: Eugene Grosbein
MFC after: 1 week

20 years agoInstead of bogusly complaining about odd file names, handle them properly
Dag-Erling Smørgrav [Sat, 24 Jan 2004 17:32:43 +0000 (17:32 +0000)]
Instead of bogusly complaining about odd file names, handle them properly
by escaping all suspicious characters.

MFC after: 3 days

20 years agoTry to apply consistent indentation.
Dag-Erling Smørgrav [Sat, 24 Jan 2004 15:35:04 +0000 (15:35 +0000)]
Try to apply consistent indentation.

20 years agoInclude the hostname in the history file.
Dag-Erling Smørgrav [Sat, 24 Jan 2004 15:06:36 +0000 (15:06 +0000)]
Include the hostname in the history file.

20 years agoAdd PFIL_HOOKS to the GENERIC kernel configuration, primarily so
Jacques Vidrine [Sat, 24 Jan 2004 14:59:51 +0000 (14:59 +0000)]
Add PFIL_HOOKS to the GENERIC kernel configuration, primarily so
that one can load the IPFilter module (which requires PFIL_HOOKS).

Requested by: Many, for over a year

20 years agoBump the date.
Dag-Erling Smørgrav [Sat, 24 Jan 2004 14:53:25 +0000 (14:53 +0000)]
Bump the date.

20 years agoBump version number to reflect the addition of cvsup support.
Dag-Erling Smørgrav [Sat, 24 Jan 2004 14:53:11 +0000 (14:53 +0000)]
Bump version number to reflect the addition of cvsup support.