]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
23 years agoMinor cleanup: sort includes, fix name.
dd [Mon, 25 Jun 2001 01:21:13 +0000 (01:21 +0000)]
Minor cleanup: sort includes, fix name.

23 years agoFix crypto targets to be user-overridable and also obey the
jkh [Mon, 25 Jun 2001 00:43:10 +0000 (00:43 +0000)]
Fix crypto targets to be user-overridable and also obey the
NO_SENDMAIL flag if set.  The whole NO_FOO mechanism in /usr/src
is pretty bogus and needs to be re-examined in the context of a
larger argument about modularity, but that's something for another
time.

Submitted by: Peter Pentchev <roam@orbitel.bg>

23 years agoXref to extattr(2) and extattr(3).
dd [Sun, 24 Jun 2001 23:59:22 +0000 (23:59 +0000)]
Xref to extattr(2) and extattr(3).

23 years agoAdd a link to extattr.2 (from extattr_get_file.2). The other names
dd [Sun, 24 Jun 2001 23:58:27 +0000 (23:58 +0000)]
Add a link to extattr.2 (from extattr_get_file.2).  The other names
are too specific and too long, and extattr.2 makes a nice point of
reference for "extattr-related syscalls".

23 years agoAdd a manual page for extattr_string_to_namespace and
dd [Sun, 24 Jun 2001 23:54:46 +0000 (23:54 +0000)]
Add a manual page for extattr_string_to_namespace and
extattr_namespace_to_string.

Reviewed by: rwatson

23 years agoNuke unused variables.
dd [Sun, 24 Jun 2001 23:41:57 +0000 (23:41 +0000)]
Nuke unused variables.

23 years agoMerge from ipfw.8 (1.57).
kuriyama [Sun, 24 Jun 2001 23:40:09 +0000 (23:40 +0000)]
Merge from ipfw.8 (1.57).
  o Sync with netinet6/ip6_fw.c (1.12).

MFC after: 10 days

23 years agoNuke unused variables.
dd [Sun, 24 Jun 2001 23:31:07 +0000 (23:31 +0000)]
Nuke unused variables.

23 years agoInclude missing header files whicih define functions for which gcc has
dd [Sun, 24 Jun 2001 23:25:59 +0000 (23:25 +0000)]
Include missing header files whicih define functions for which gcc has
builtints (e.g., exit, strcmp).

23 years agoInclude missing header files which define functions for which gcc has
dd [Sun, 24 Jun 2001 23:22:14 +0000 (23:22 +0000)]
Include missing header files which define functions for which gcc has
builtints (e.g., exit, strcmp).

23 years agoInclude missing header files which define functions for which gcc has
dd [Sun, 24 Jun 2001 23:04:23 +0000 (23:04 +0000)]
Include missing header files which define functions for which gcc has
builtins (e.g., exit, strcmp).

23 years agoMerge from netinet/ip_fw.c (1.117 -> 1.118).
kuriyama [Sun, 24 Jun 2001 22:41:00 +0000 (22:41 +0000)]
Merge from netinet/ip_fw.c (1.117 -> 1.118).
  o Use syslog(3) interface for logging.

Reviewed by: ume
MFC after: 10 days

23 years agoXref extattr_set_file(2).
dd [Sun, 24 Jun 2001 21:14:46 +0000 (21:14 +0000)]
Xref extattr_set_file(2).

23 years agoXref extattr_get_file(2).
dd [Sun, 24 Jun 2001 21:13:58 +0000 (21:13 +0000)]
Xref extattr_get_file(2).

23 years agoAdd a namespace argument to one of the examples.
dd [Sun, 24 Jun 2001 21:13:01 +0000 (21:13 +0000)]
Add a namespace argument to one of the examples.

23 years agomdoc(7) police: the BUGS section should go after HISTORY and AUTHORS.
dd [Sun, 24 Jun 2001 20:58:20 +0000 (20:58 +0000)]
mdoc(7) police: the BUGS section should go after HISTORY and AUTHORS.

23 years agoThe fact that phk wrote this is not a bug!
dd [Sun, 24 Jun 2001 20:56:39 +0000 (20:56 +0000)]
The fact that phk wrote this is not a bug!

23 years agoremove IN6_IS_ADDR_ANY macro (outside of standard, #if 0'ed for a long time)
ume [Sun, 24 Jun 2001 20:43:01 +0000 (20:43 +0000)]
remove IN6_IS_ADDR_ANY macro (outside of standard, #if 0'ed for a long time)

Obtained from: KAME
MFC after: 10 days

23 years agoReword the sentence that I screwed up in the last commit.
dd [Sun, 24 Jun 2001 20:34:31 +0000 (20:34 +0000)]
Reword the sentence that I screwed up in the last commit.

Submitted by: Erik Trulsson <ertr1013@student.uu.se>

23 years agoAdd missing includes and sort includes.
dd [Sun, 24 Jun 2001 20:28:19 +0000 (20:28 +0000)]
Add missing includes and sort includes.

23 years agoInclude missing header files.
dd [Sun, 24 Jun 2001 20:26:00 +0000 (20:26 +0000)]
Include missing header files.

23 years agodisallow setsockopt(IPV6_V6ONLY) for already bound sockets.
ume [Sun, 24 Jun 2001 20:25:38 +0000 (20:25 +0000)]
disallow setsockopt(IPV6_V6ONLY) for already bound sockets.

Obtained from: KAME
MFC after: 10 days

23 years agoInclude missing header files which define functions for which gcc has
dd [Sun, 24 Jun 2001 20:25:23 +0000 (20:25 +0000)]
Include missing header files which define functions for which gcc has
builtints (e.g., exit, strcmp).

23 years agoInclude stdlib.h for exit(3).
dd [Sun, 24 Jun 2001 20:05:30 +0000 (20:05 +0000)]
Include stdlib.h for exit(3).

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoInclude missing header files which define functions for which gcc has
dd [Sun, 24 Jun 2001 19:50:42 +0000 (19:50 +0000)]
Include missing header files which define functions for which gcc has
builtins (e.g., exit, strcmp).

23 years agoInclude missing header files which define functions for which gcc has
dd [Sun, 24 Jun 2001 19:41:18 +0000 (19:41 +0000)]
Include missing header files which define functions for which gcc has
builtins (e.g., exit).

Submitted by: Mike Barcroft <mike@q9media.com>

23 years agoSilence warning and set WARNS=2.
dd [Sun, 24 Jun 2001 18:56:00 +0000 (18:56 +0000)]
Silence warning and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>
Reviewed by: md5(1)

23 years agoConstify and set WARNS=2.
dd [Sun, 24 Jun 2001 18:51:40 +0000 (18:51 +0000)]
Constify and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>
Reviewed by: md5(1)

23 years agoConstify, add a prototype, and set WARNS=2.
dd [Sun, 24 Jun 2001 18:43:33 +0000 (18:43 +0000)]
Constify, add a prototype, and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>
Reviewed by: md5(1)

23 years agoConstify and set WARNS=2.
dd [Sun, 24 Jun 2001 18:41:30 +0000 (18:41 +0000)]
Constify and set WARNS=2.

Submitted by: Mike Barcroft <mike@q9media.com>
Reviewed by: md5(1)

23 years agofix oopslet for PROC_UNLOCK_NOSWITCH
mjacob [Sun, 24 Jun 2001 18:23:17 +0000 (18:23 +0000)]
fix oopslet for PROC_UNLOCK_NOSWITCH

23 years agoUse strdup(3) instead of reimplementing it inline.
dd [Sun, 24 Jun 2001 18:21:52 +0000 (18:21 +0000)]
Use strdup(3) instead of reimplementing it inline.

23 years agocleanup some of the XPORT_FC code
mjacob [Sun, 24 Jun 2001 18:19:24 +0000 (18:19 +0000)]
cleanup some of the XPORT_FC code

23 years agoAdd a transport settings variant for Fibre Channel.
mjacob [Sun, 24 Jun 2001 18:17:45 +0000 (18:17 +0000)]
Add a transport settings variant for Fibre Channel.

23 years agoAdd CAM_NEW_TRAN_CODE as an option (to be included in opt_cam.h).
mjacob [Sun, 24 Jun 2001 18:15:58 +0000 (18:15 +0000)]
Add CAM_NEW_TRAN_CODE as an option (to be included in opt_cam.h).

23 years agoReword a sentence to make it proper English.
dd [Sun, 24 Jun 2001 18:11:42 +0000 (18:11 +0000)]
Reword a sentence to make it proper English.

23 years agoClean up the addrmerge() function, which was over-complicated and
iedowse [Sun, 24 Jun 2001 15:03:06 +0000 (15:03 +0000)]
Clean up the addrmerge() function, which was over-complicated and
contained a number of memory leaks. The changes include:

 - Add a comment describing what addrmerge() does.
 - Deal with 0.0.0.0./::. or AF_LOCAL callers correctly.
 - Use rpcbind_get_conf() instead of getnetconfigent() so we don't
   have to remember to free the returned netconfig struct.
 - Make just one pass through the ifaddrs list; we can pick up a fallback
   interface address in the same pass as the netmask comparison.
 - Define and use SA2SIN* macros to avoid the need for loads of
   protocol-specific local variables.
 - Use mostly protocol-independent code for building the netbuf version
   of the address to be returned.
 - Use the common cleanup code for virtually all error and non-error
   cases, fixing a number of memory leaks.

23 years agoon icmp6 node information query (FQDN), do not return hostnames with
ume [Sun, 24 Jun 2001 14:59:34 +0000 (14:59 +0000)]
on icmp6 node information query (FQDN), do not return hostnames with
two dots (like "foo..bar").  0-length labels are not distinguishable
with multiple name replies.

Obtained from: KAME
MFC after: 10 days

23 years agodecrease warning
ume [Sun, 24 Jun 2001 14:57:20 +0000 (14:57 +0000)]
decrease warning

Obtained from: KAME
MFC after: 10 days

23 years agoNuke the comment about MIP6. We don't have MIP6 code, yet.
ume [Sun, 24 Jun 2001 14:55:18 +0000 (14:55 +0000)]
Nuke the comment about MIP6.  We don't have MIP6 code, yet.

MFC after: 10 days

23 years agoinject outbound packet to BPF.
ume [Sun, 24 Jun 2001 14:52:55 +0000 (14:52 +0000)]
inject outbound packet to BPF.

Submitted by: itojun
Obtained from: KAME
MFC after: 10 days

23 years agoConstify, staticize and add a missing include; add WARNS?=2.
des [Sun, 24 Jun 2001 12:38:28 +0000 (12:38 +0000)]
Constify, staticize and add a missing include; add WARNS?=2.

Submitted by: Mike Barcroft <mike@q9media.com>
MFC after: 2 weeks

23 years agoDon't dereference a NULL pointer if we fail to get a sendfilebuf.
dwmalone [Sun, 24 Jun 2001 12:27:30 +0000 (12:27 +0000)]
Don't dereference a NULL pointer if we fail to get a sendfilebuf.

23 years agoAdd prototypes and turn on WARNS=?2.
dwmalone [Sun, 24 Jun 2001 12:23:54 +0000 (12:23 +0000)]
Add prototypes and turn on WARNS=?2.

23 years agoAllow getcred sysctl to work in jailed root processes. Processes can
dwmalone [Sun, 24 Jun 2001 12:18:27 +0000 (12:18 +0000)]
Allow getcred sysctl to work in jailed root processes. Processes can
only do getcred calls for sockets which were created in the same jail.
This should allow the ident to work in a reasonable way within jails.

PR: 28107
Approved by: des, rwatson

23 years agoGet rid of some constness warnings.
dwmalone [Sun, 24 Jun 2001 12:08:47 +0000 (12:08 +0000)]
Get rid of some constness warnings.

23 years agoDon't add -Wall, as it's a compiler specific flag.
dwmalone [Sun, 24 Jun 2001 09:20:42 +0000 (09:20 +0000)]
Don't add -Wall, as it's a compiler specific flag.

23 years agoFix most of the warnings given by WARNS=2.
dwmalone [Sun, 24 Jun 2001 09:20:07 +0000 (09:20 +0000)]
Fix most of the warnings given by WARNS=2.

23 years agoFix asm constraints for atomic_cmpset_32. This fix may also be needed
benno [Sun, 24 Jun 2001 06:36:28 +0000 (06:36 +0000)]
Fix asm constraints for atomic_cmpset_32.  This fix may also be needed
elsewhere.

23 years agoAfter exhaustive discussions and some meandering and confusion, enough
dillon [Sun, 24 Jun 2001 05:24:41 +0000 (05:24 +0000)]
After exhaustive discussions and some meandering and confusion, enough
people are on track with the cause and effect of this, and although
fixing this severely degenerate case appears to violate the letter of
POSIX.1-200x, Bruce and I (and enough others) agree that it should be
comitted.

So, this patch generates an ENOENT error for any attempt to do a path lookup
through an empty symlink (e.g. open(), stat()).

Submitted by: "Andrey A. Chernov" <ache@nagual.pp.ru>
Reviewed by: bde
Discussed exhaustively on: freebsd-current
Previously committed to: NetBSD 4 years ago

23 years agoReset timestamps of battery info. and status by thier notify handler.
iwasaki [Sun, 24 Jun 2001 02:39:08 +0000 (02:39 +0000)]
Reset timestamps of battery info. and status by thier notify handler.

Suggested by:   takawata

23 years ago- Swap order of "S4B" and "S5" in sleep_state_names. They already
iwasaki [Sun, 24 Jun 2001 02:37:38 +0000 (02:37 +0000)]
- Swap order of "S4B" and "S5" in sleep_state_names.  They already
  changed in ACPICA actypes.h.
- Use ACPI_S_STATES_MAX instead of ACPI_STATE_S5.

23 years agoDon't call exit(3) from a signal handler.
dd [Sun, 24 Jun 2001 01:55:17 +0000 (01:55 +0000)]
Don't call exit(3) from a signal handler.

Obtained from: OpenBSD

23 years agomdoc(7) police: use the Fn macro where appropriate.
dd [Sun, 24 Jun 2001 01:38:20 +0000 (01:38 +0000)]
mdoc(7) police: use the Fn macro where appropriate.

23 years agoRemove duplicate words.
dd [Sun, 24 Jun 2001 01:34:38 +0000 (01:34 +0000)]
Remove duplicate words.

23 years agoCorrect a typo.
dd [Sun, 24 Jun 2001 01:30:38 +0000 (01:30 +0000)]
Correct a typo.

23 years agomdoc(7) police: sort xrefs.
dd [Sun, 24 Jun 2001 01:17:07 +0000 (01:17 +0000)]
mdoc(7) police: sort xrefs.

23 years agoAdd myself.
jasone [Sun, 24 Jun 2001 00:45:03 +0000 (00:45 +0000)]
Add myself.

23 years agoFix some return-value brain-damage in forward_register(). This
iedowse [Sat, 23 Jun 2001 21:54:47 +0000 (21:54 +0000)]
Fix some return-value brain-damage in forward_register(). This
function has a return type of u_int32_t, into which it was somehow
supposed to encode:
 *  A valid 32-bit XID (which could be any value including 0).
 *  0, meaning a duplicate request.
 *  -1, meaning a malloc failed (!);
We now ensure that all XIDs are non-zero, and pass the XID out via
a pointer argument.

In forward_find() and free_slot_by_xid(), remove an unnecessary
and confusing test for a negative result from an unsigned modulo
operation, but add an unnecessary cast to highlight why.

23 years agoUse the new CLSET_CONNECT clnt_control() request on UDP clients to
iedowse [Sat, 23 Jun 2001 19:57:20 +0000 (19:57 +0000)]
Use the new CLSET_CONNECT clnt_control() request on UDP clients to
ensure that we never proceed with the mount() syscall if the server
is replying from the wrong source address. Previously the userland
RPC call to the remote nfsd would succeed, but the kernel uses
connect() so it would not see the replies, resulting in a hung
mount.

23 years agoAdd a new clnt_control() request `CLSET_CONNECT' that controls
iedowse [Sat, 23 Jun 2001 19:43:21 +0000 (19:43 +0000)]
Add a new clnt_control() request `CLSET_CONNECT' that controls
whether or not connect(2) is used for UDP client sockets. The default
is not to connect(), so existing clients will see no change in
behaviour.

The use of connect(2) for UDP clients has a number of advantages:
only replies from the intended address are received, and ICMP errors
pertaining to the connection are reported back to the application.

23 years agoAdd initializetion of NVCTL register with EEPROM stored value.
semenu [Sat, 23 Jun 2001 19:30:26 +0000 (19:30 +0000)]
Add initializetion of NVCTL register with EEPROM stored value.
This fix hazardous very slow work for one of my cards.

MFC after: 1 week

23 years agoremove -g -O0, that was only for debugging...
ben [Sat, 23 Jun 2001 19:24:18 +0000 (19:24 +0000)]
remove -g -O0, that was only for debugging...

23 years agoDon't dereference a pointer after freeing that pointer.
ben [Sat, 23 Jun 2001 18:07:18 +0000 (18:07 +0000)]
Don't dereference a pointer after freeing that pointer.

PR: 27990
Reviewed by: dd, phk
Tested by: dd
MFC after: 2 weeks

23 years agothe turtle beach santacruz implements ac97 eapd in an inverted sense -
cg [Sat, 23 Jun 2001 18:00:06 +0000 (18:00 +0000)]
the turtle beach santacruz implements ac97 eapd in an inverted sense -
inform the ac97 code of this.

23 years agoReduce field width for some columns so that the default output fits in <= 79
schweikh [Sat, 23 Jun 2001 17:52:10 +0000 (17:52 +0000)]
Reduce field width for some columns so that the default output fits in <= 79
columns. 80 columns + newline is a problem for some terminals including syscons.

Requested by: bde
MFC after: 1 week

23 years agoReplace bzero() of struct ip with explicit zeroing of structure members,
jlemon [Sat, 23 Jun 2001 17:44:27 +0000 (17:44 +0000)]
Replace bzero() of struct ip with explicit zeroing of structure members,
which is faster.

23 years agoadd defines and ifdefs so this code will compile on 4.x
cg [Sat, 23 Jun 2001 17:36:51 +0000 (17:36 +0000)]
add defines and ifdefs so this code will compile on 4.x

add spls so this code will work on 4.x

23 years agoThree people known to be born on December 18!
ru [Sat, 23 Jun 2001 17:34:26 +0000 (17:34 +0000)]
Three people known to be born on December 18!

Submitted by: cwt

23 years agomove a private struct definition from ac97.h to ac97.c
cg [Sat, 23 Jun 2001 17:32:40 +0000 (17:32 +0000)]
move a private struct definition from ac97.h to ac97.c

add proper handling of cards which take eapd=1 to mean 'produce sound'
instead of 'external amplifier power down'

23 years agoAdd netstat(1) knob to reset net.inet.{ip|icmp|tcp|udp|igmp}.stats.
ru [Sat, 23 Jun 2001 17:17:59 +0000 (17:17 +0000)]
Add netstat(1) knob to reset net.inet.{ip|icmp|tcp|udp|igmp}.stats.
For example, ``netstat -s -p ip -z'' will show and reset IP stats.

PR: bin/17338

23 years agoadd sndstat.c so module users have /dev/sndstat again
cg [Sat, 23 Jun 2001 17:09:38 +0000 (17:09 +0000)]
add sndstat.c so module users have /dev/sndstat again

23 years ago- Fix space allocation for mbstat structure
bmilekic [Sat, 23 Jun 2001 17:04:17 +0000 (17:04 +0000)]
- Fix space allocation for mbstat structure
- Make sure to try hw.ncpu if kern.smp.cpus doesn't exist (i.e. on UP) to
  get number of CPUs.

23 years agoMake sure to try hw.ncpu if kern.smp.cpus doesn't exist (i.e. on UP) when
bmilekic [Sat, 23 Jun 2001 17:03:27 +0000 (17:03 +0000)]
Make sure to try hw.ncpu if kern.smp.cpus doesn't exist (i.e. on UP) when
getting number of CPUs.

23 years agoBy default link statically (like we did in the port) - it provides 8-10%
sobomax [Sat, 23 Jun 2001 12:29:14 +0000 (12:29 +0000)]
By default link statically (like we did in the port) - it provides 8-10%
faster comperssion, which in the bzip2's case could be a big win.

23 years agoAdd sysctl interface (Read-only) for temprature, AC-line and Battery.
iwasaki [Sat, 23 Jun 2001 10:38:25 +0000 (10:38 +0000)]
Add sysctl interface (Read-only) for temprature, AC-line and Battery.
Patches for acpi_cmbat.c submitted by Munehiro Matsuda.

23 years agoHonor -s -s (don't show zero stats) with -r, untangle SYNOPSIS further.
ru [Sat, 23 Jun 2001 09:18:57 +0000 (09:18 +0000)]
Honor -s -s (don't show zero stats) with -r, untangle SYNOPSIS further.
(usage() still is not synchronized with SYNOPSIS, intentionally.)

23 years agoAdd myself.
rse [Sat, 23 Jun 2001 09:16:29 +0000 (09:16 +0000)]
Add myself.

23 years agoFixed DPADD.
ru [Sat, 23 Jun 2001 09:10:48 +0000 (09:10 +0000)]
Fixed DPADD.

Submitted by: bde

23 years agoDocument "normalizing" behavior of mktime(3).
ru [Sat, 23 Jun 2001 08:57:40 +0000 (08:57 +0000)]
Document "normalizing" behavior of mktime(3).

PR: bin/28313
Reviewed by: bde, wollman
MFC after: 1 week

23 years agoMerged from sys/i386/isa/npx.c revision 1.102.
kato [Sat, 23 Jun 2001 08:39:13 +0000 (08:39 +0000)]
Merged from sys/i386/isa/npx.c revision 1.102.

23 years agoAverage age is 28 years and 11 months.
ru [Sat, 23 Jun 2001 08:36:11 +0000 (08:36 +0000)]
Average age is 28 years and 11 months.

Submitted by: hrs, marko

23 years agoMerged from sys/i386/i386/machdep.c revisions 1.456 and 1.457.
kato [Sat, 23 Jun 2001 08:30:13 +0000 (08:30 +0000)]
Merged from sys/i386/i386/machdep.c revisions 1.456 and 1.457.

23 years agoRemoved pcm and sbc devices.
kato [Sat, 23 Jun 2001 08:18:55 +0000 (08:18 +0000)]
Removed pcm and sbc devices.

23 years agoMerged from sys/conf/files.i386 revisions 1.362 and 1.363.
kato [Sat, 23 Jun 2001 08:10:50 +0000 (08:10 +0000)]
Merged from sys/conf/files.i386 revisions 1.362 and 1.363.

23 years agoMerged from sys/boot/i386/btx/btx/btx.s revision 1.25.
kato [Sat, 23 Jun 2001 08:04:14 +0000 (08:04 +0000)]
Merged from sys/boot/i386/btx/btx/btx.s revision 1.25.

23 years agoMake sure the average age keeps on decreasing down that slippery slope.
green [Sat, 23 Jun 2001 07:24:18 +0000 (07:24 +0000)]
Make sure the average age keeps on decreasing down that slippery slope.

23 years agoRemove duplicate word.
dd [Sat, 23 Jun 2001 07:02:33 +0000 (07:02 +0000)]
Remove duplicate word.

23 years agoBring in fix from rev 1.32 that got lost during the conversion to kqueue.
jlemon [Sat, 23 Jun 2001 06:54:28 +0000 (06:54 +0000)]
Bring in fix from rev 1.32 that got lost during the conversion to kqueue.
This prevents an indefinte timeout in case the kevent call is interrupted
for some reason.

PR: 26665
MFC in: 2 weeks

23 years agoConvert inb/outb to bus_space.
nsouch [Sat, 23 Jun 2001 06:52:54 +0000 (06:52 +0000)]
Convert inb/outb to bus_space.

Submitted by: jcm@FreeBSD-uk.eu.org

23 years agoTranslate various ppbus sequences into microsequences to limit
nsouch [Sat, 23 Jun 2001 06:51:52 +0000 (06:51 +0000)]
Translate various ppbus sequences into microsequences to limit
overhead of abstraction layers.

Submitted by: jcm@FreeBSD-uk.eu.org

23 years agoMake the similar changes as in our keyinfo, i.e. allow user to get his own
ache [Sat, 23 Jun 2001 04:48:59 +0000 (04:48 +0000)]
Make the similar changes as in our keyinfo, i.e. allow user to get his own
sequence and allow root to get everybody's one. Before this change user
can't get his own sequence, root required.

23 years agogive up and join the party (along with thomas Moestl (same bday))
julian [Sat, 23 Jun 2001 04:37:31 +0000 (04:37 +0000)]
give up and join the party (along with thomas Moestl (same bday))
but maybe this file shouldn't be so widly distributed :-)

23 years agoEliminate the allocation of a tcp template structure for each
silby [Sat, 23 Jun 2001 03:21:46 +0000 (03:21 +0000)]
Eliminate the allocation of a tcp template structure for each
connection.  The information contained in a tcptemp can be
reconstructed from a tcpcb when needed.

Previously, tcp templates required the allocation of one
mbuf per connection.  On large systems, this change should
free up a large number of mbufs.

Reviewed by: bmilekic, jlemon, ru
MFC after: 2 weeks

23 years agoFix a number of bugs in the implementation of the WEP related status
brooks [Fri, 22 Jun 2001 23:35:24 +0000 (23:35 +0000)]
Fix a number of bugs in the implementation of the WEP related status
functions in ifconfig.  "ifconfig an0" should output the correct
status now.  Also, make the read and write functions both more
robust and more consistant.  This should stop most of the incorrect
size complaints and eliminate the possiability of panics from firmware
that increases resource sizes.

PR: kern/27826
Reviewed by: imp, jlemon
Submitted by: Doug Ambrisko <ambrisko@ambrisko.com>
David Wolfskill <dhw@whistle.com>

23 years ago- Lock CURSIG() with the proc lock to close the signal race with psignal.
jhb [Fri, 22 Jun 2001 23:11:26 +0000 (23:11 +0000)]
- Lock CURSIG() with the proc lock to close the signal race with psignal.
- Grab Giant around ktrace points.
- Clean up KTR_PROC tracepoints to not display the value of
  sched_lock.mtx_lock as it isn't really needed anymore and just obfuscates
  the messages.
- Add a few if conditions to replace gotos.
- Ensure that every msleep KTR event ends up with a matching msleep resume
  KTR event (this was broken when we didn't do a mi_switch()).
- Only note via ktrace that we resumed from a switch once rather than twice
  in several places in msleep().
- Remove spl's rom asleep and await as the proc lock and sched_lock provide
  all the needed locking.
- In mawait() add in a needed ktrace point for noting that we are about to
  switch out.

23 years ago- Lock CURSIG with the proc lock and don't release the proc lock until
jhb [Fri, 22 Jun 2001 23:06:38 +0000 (23:06 +0000)]
- Lock CURSIG with the proc lock and don't release the proc lock until
  after grabbing the sched lock to close a race.
- Lock ktrace points with Giant.

23 years ago- Grab the proc lock around CURSIG and postsig(). Don't release the proc
jhb [Fri, 22 Jun 2001 23:05:11 +0000 (23:05 +0000)]
- Grab the proc lock around CURSIG and postsig().  Don't release the proc
  lock until after grabbing the sched_lock to avoid CURSIG racing with
  psignal.
- Don't grab Giant for addupc_task() as it isn't needed.

Reported by: tegge (signal race), bde (addupc_task a while back)

23 years ago- Change CURSIG() and postsig() to require that the proc lock is held
jhb [Fri, 22 Jun 2001 23:02:37 +0000 (23:02 +0000)]
- Change CURSIG() and postsig() to require that the proc lock is held
  rather than grabbing it and releasing it themselves.  This allows callers
  of these functions to get the lock to close race conditions.
- Grab Giant around ktrace in postsig.
- Count the switches performed on SIGSTOP's as involuntary context switches
  in the resource usage stats.

Reported by: tegge (signal race), bde (missing csw stats)

23 years agoNow how many files would a new committer have to commit?
tanimura [Fri, 22 Jun 2001 22:53:22 +0000 (22:53 +0000)]
Now how many files would a new committer have to commit?