]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoRemove the zzz files since they have been moved to src/usr.sbin/zzz
njl [Mon, 14 Jul 2003 16:32:45 +0000 (16:32 +0000)]
Remove the zzz files since they have been moved to src/usr.sbin/zzz

21 years agoMove the zzz utility to its own directory since it is no longer dependent
njl [Mon, 14 Jul 2003 16:31:34 +0000 (16:31 +0000)]
Move the zzz utility to its own directory since it is no longer dependent
on apm(8)

21 years agoMake C++ safe
imp [Mon, 14 Jul 2003 16:31:20 +0000 (16:31 +0000)]
Make C++ safe

21 years agoApparently we prefer underscores in new options. Sorry for the churn folks.
gordon [Mon, 14 Jul 2003 16:21:16 +0000 (16:21 +0000)]
Apparently we prefer underscores in new options. Sorry for the churn folks.

Requested by: obrien

21 years agoMore changes to use __FBSDID() for setting rcsids, and fix the
gad [Mon, 14 Jul 2003 15:54:41 +0000 (15:54 +0000)]
More changes to use __FBSDID() for setting rcsids, and fix the
format of 'sccsid' lines so they consistently match style(9)
guidelines.  Inspired by recent update to lpd.c by charnier.

Reviewed by: discussed on cvs-src & with bde and obrien
MFC after: 15 days

21 years agoDocument the PHY chip statistics.
harti [Mon, 14 Jul 2003 15:08:11 +0000 (15:08 +0000)]
Document the PHY chip statistics.

21 years agoImplement statistics for the PHY chips. Statistics are hold in
harti [Mon, 14 Jul 2003 15:06:53 +0000 (15:06 +0000)]
Implement statistics for the PHY chips. Statistics are hold in
64-bit counters that wrap on overflow. They are collecte once per
second from the chips. Currently they can be retrieved via a sysctl phy_stats.
A write of an arbitrary value to the sysctl atomically retrieves the
statistics and clears them.

21 years agoMake sure ipmon is run after cleaning up in /var. Otherwise,
mtm [Mon, 14 Jul 2003 13:02:36 +0000 (13:02 +0000)]
Make sure ipmon is run after cleaning up in /var. Otherwise,
we would lose it's pid file. Also, cleanup the depencies in
rc.d/cleanvar

PR: bin/53516

21 years agoo Rename local variables, do not shadow global declarations.
maxim [Mon, 14 Jul 2003 12:43:48 +0000 (12:43 +0000)]
o Rename local variables, do not shadow global declarations.

21 years agoo Kill MINICMPLEN, there is ICMP_MINLEN already.
maxim [Mon, 14 Jul 2003 12:42:47 +0000 (12:42 +0000)]
o Kill MINICMPLEN, there is ICMP_MINLEN already.

21 years agoo Be ready to get a reply with length up to IP_MAXPACKET.
maxim [Mon, 14 Jul 2003 12:37:03 +0000 (12:37 +0000)]
o Be ready to get a reply with length up to IP_MAXPACKET.
o Warn when recieved packet length is not equal to length of the
packet we sent out. Idea from NetBSD.
o Fit the dump of packet with wrong data to 80 columns (from NetBSD).

Comments from: bde

21 years agoDocument the support for the IDT77155 Sonet/SDH PHY.
harti [Mon, 14 Jul 2003 12:13:30 +0000 (12:13 +0000)]
Document the support for the IDT77155 Sonet/SDH PHY.

Cross-reference hatm(4) and fatm(4).

21 years agoAdd support for the IDT77155 Sonet/SDH PHY.
harti [Mon, 14 Jul 2003 12:12:50 +0000 (12:12 +0000)]
Add support for the IDT77155 Sonet/SDH PHY.

21 years agoccept of empty lines when reading from a file (this fixes a bug
luigi [Mon, 14 Jul 2003 08:39:49 +0000 (08:39 +0000)]
ccept of empty lines when reading from a file (this fixes a bug
introduced in the latest commits).

Also:

* update the 'ipfw -h' output;

* allow rules of the form "100 add allow ..." i.e. with the index first.
  (requested by Paul Richards). This was an undocumented ipfw1 behaviour,
  and it is left undocumented.

and minor code cleanups.

21 years agoTake advantage of the common_source/lp.cdefs.h file to change lpr
gad [Mon, 14 Jul 2003 05:15:21 +0000 (05:15 +0000)]
Take advantage of the common_source/lp.cdefs.h file to change lpr
source to use __FBSDID() for setting rcsids.  Also fix the format
of 'sccsid' lines to consistently match style(9) guidelines.

Reviewed by: discussed with bde and obrien
MFC after: 15 days

21 years agoIntroduce a new lp.cdefs.h file, which will be used to make it
gad [Mon, 14 Jul 2003 05:13:55 +0000 (05:13 +0000)]
Introduce a new lp.cdefs.h file, which will be used to make it
somewhat easier to build this lpr on other operating systems.
This simply includes <sys/cdefs.h> when that is appropriate,
and then checks for any cdefs-ish macros that lpr uses, and
defines them if they don't already exist.  This is only a start
at making freebsd's lpr less of a hassle to port.  It is mainly
added so all of lpr can be changed to use the __FBSDID() macro,
without making it *more* of a hassle to build on other OS's.

Reviewed by: discussed with bde and obrien
MFC after: 15 days

21 years agoChange zzz to be a shell script that checks for APM or ACPI support and
njl [Mon, 14 Jul 2003 03:58:05 +0000 (03:58 +0000)]
Change zzz to be a shell script that checks for APM or ACPI support and
suspends the system appropriately.

Submitted by: Mark Santcroos <marks@ripe.net>

21 years agoDocument the existance of fla(4).
obrien [Mon, 14 Jul 2003 03:23:58 +0000 (03:23 +0000)]
Document the existance of fla(4).

21 years ago_start() needed to be written in assembly. See crt1.S.
marcel [Mon, 14 Jul 2003 03:05:42 +0000 (03:05 +0000)]
_start() needed to be written in assembly. See crt1.S.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r117535,
njl [Mon, 14 Jul 2003 02:58:00 +0000 (02:58 +0000)]
This commit was generated by cvs2svn to compensate for changes in r117535,
which included commits to RCS files with non-trunk default branches.

21 years agoRemove a multiply-defined function. The change has been submitted to the
njl [Mon, 14 Jul 2003 02:58:00 +0000 (02:58 +0000)]
Remove a multiply-defined function.  The change has been submitted to the
vendor.

21 years agoFix build breakage on ia64. The second argument of AcpiOsReadable()
marcel [Mon, 14 Jul 2003 02:42:15 +0000 (02:42 +0000)]
Fix build breakage on ia64. The second argument of AcpiOsReadable()
and AcpiOsWritable() have type ACPI_SIZE and not UINT32.

21 years agoRemove reference to max block size. dump no longer limits the block size.
grog [Mon, 14 Jul 2003 02:22:55 +0000 (02:22 +0000)]
Remove reference to max block size.  dump no longer limits the block size.

21 years agoNew release notes: ACPI-CA 20030619.
hrs [Sun, 13 Jul 2003 23:26:20 +0000 (23:26 +0000)]
New release notes:  ACPI-CA 20030619.

21 years agoRewite _start(). We cannot use a C function due to the fact that we
marcel [Sun, 13 Jul 2003 23:11:37 +0000 (23:11 +0000)]
Rewite _start(). We cannot use a C function due to the fact that we
don't call it according to the runtime specification and especially
WRT to gp this can cause trouble. The gcc 3.3.1 import broke the
ia64 runtime because the compiler saved gp prior to us being able
to set it properly. Restoring gp after the calls would then invalidate
gp and cause segmentation faults later on.
By rewriting _start() as an assembly function, we also avoided even
more gcc dependences, by trying to use gcc specific features to work
around the problem.
This version of _start() does not reference _DYNAMIC. We register the
cleanup function when it's a non-NULL pointer. The kernel will always
pass a NULL pointer and dynamic linkers may pass a non-NULL pointer.

The machine independent code to set __progname now unfortunately is
written in assembly. So be it.

21 years agoUpdate code to work with 0619 dist
njl [Sun, 13 Jul 2003 22:57:16 +0000 (22:57 +0000)]
Update code to work with 0619 dist

* Use ACPI_BUFFER as the type for AcpiGetObjectInfo
* Remove AcpiEnableEvent/AcpiClearEvent for ACPI_EVENT_FIXED (power/sleep
  buttons) as they are no longer needed
* Change calls to use the new GPE functions
* Add AcpiOs*Lock functions

21 years agoAdd include file so this builds with new acpica
njl [Sun, 13 Jul 2003 22:54:53 +0000 (22:54 +0000)]
Add include file so this builds with new acpica

Reported by: Kevin Oberman <oberman@es.net>

21 years agoFix handling of _BST in the new dist.
njl [Sun, 13 Jul 2003 22:52:51 +0000 (22:52 +0000)]
Fix handling of _BST in the new dist.

Submitted by: kochi@netbsd.org

21 years agoThis commit was generated by cvs2svn to compensate for changes in r117527,
njl [Sun, 13 Jul 2003 22:52:51 +0000 (22:52 +0000)]
This commit was generated by cvs2svn to compensate for changes in r117527,
which included commits to RCS files with non-trunk default branches.

21 years agoPatch for global lock error messages that the vendor will import eventually.
njl [Sun, 13 Jul 2003 22:50:30 +0000 (22:50 +0000)]
Patch for global lock error messages that the vendor will import eventually.
Originally committed as 1.1.1.17

21 years agoThis commit was generated by cvs2svn to compensate for changes in r117525,
njl [Sun, 13 Jul 2003 22:50:30 +0000 (22:50 +0000)]
This commit was generated by cvs2svn to compensate for changes in r117525,
which included commits to RCS files with non-trunk default branches.

21 years agoCommit local patches for files off the vendor branch.
njl [Sun, 13 Jul 2003 22:47:01 +0000 (22:47 +0000)]
Commit local patches for files off the vendor branch.

21 years agoUnmodified files from the vendor 0619 dist that should no longer be off the
njl [Sun, 13 Jul 2003 22:45:34 +0000 (22:45 +0000)]
Unmodified files from the vendor 0619 dist that should no longer be off the
vendor branch.

21 years agoACPICA import from the 0619 dist.
njl [Sun, 13 Jul 2003 22:44:13 +0000 (22:44 +0000)]
ACPICA import from the 0619 dist.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r117521,
njl [Sun, 13 Jul 2003 22:44:13 +0000 (22:44 +0000)]
This commit was generated by cvs2svn to compensate for changes in r117521,
which included commits to RCS files with non-trunk default branches.

21 years agoioctl macros and defines are now present in ioccom.h, not ioctl.h.
ceri [Sun, 13 Jul 2003 21:02:48 +0000 (21:02 +0000)]
ioctl macros and defines are now present in ioccom.h, not ioctl.h.
Update the manpage to reflect this.

PR: docs/54235
Submitted by: Karen Thode <thode12@msn.com>

21 years agoAvoid an unnecessary calculation: there is no need to subtract
robert [Sun, 13 Jul 2003 21:02:11 +0000 (21:02 +0000)]
Avoid an unnecessary calculation: there is no need to subtract
`firstaddr' from `v' if we know that the former equals zero.

21 years agoMove the MAC entry point to label ethernet-sourced mbufs with a MAC label
rwatson [Sun, 13 Jul 2003 20:32:58 +0000 (20:32 +0000)]
Move the MAC entry point to label ethernet-sourced mbufs with a MAC label
from the network interface earlier in ether_input().  At some point
(no fingers pointed), things were restructured and the labeling operation
moved later.  This wasn't a problem as BPF_MTAP() relies on the ifnet
label not the mbuf label, but there might have been other problems.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, Network Associates Laboratories

21 years agoIf mktime() or timegm() return -1, print out the resume timer value as
rwatson [Sun, 13 Jul 2003 20:06:41 +0000 (20:06 +0000)]
If mktime() or timegm() return -1, print out the resume timer value as
"unknown", rather than a conversion of -1 to a date/timestamp.

21 years agoo add read-only sysctls to view regulatory domain, country code, and
sam [Sun, 13 Jul 2003 17:07:25 +0000 (17:07 +0000)]
o add read-only sysctls to view regulatory domain, country code, and
  outdoor use controls
o use sysctl-visible values in setting up channel list

21 years agocorrect card type for Linksys *55AG
sam [Sun, 13 Jul 2003 17:03:39 +0000 (17:03 +0000)]
correct card type for Linksys *55AG

PR: bin/54190
Submitted by: Josh Elsasser <jre@vineyard.net>

21 years agoSince -march=pentium4 is supposed to be fixed with GCC 3.3, and
mux [Sun, 13 Jul 2003 10:34:09 +0000 (10:34 +0000)]
Since -march=pentium4 is supposed to be fixed with GCC 3.3, and
since people have been reporting success with it, re-enable the
pentium4 optimization.

21 years agoFix for FAIL_IF_NO_KBD case as expected.
simokawa [Sun, 13 Jul 2003 10:08:33 +0000 (10:08 +0000)]
Fix for FAIL_IF_NO_KBD case as expected.

Even if we have no AT keyboard, an AT keyboard is registered because
it's probed with KB_CONF_PROBE_ONLY flag set during console initialization.
Unregister the keyboard if it doesn't present while second probe.

This should fix USB keyboard only case without 'kbdcontrol -k /dev/kbd1'.

21 years agoEliminate non-existent word.
jkoshy [Sun, 13 Jul 2003 09:11:10 +0000 (09:11 +0000)]
Eliminate non-existent word.

Submitted by: jwd

21 years agoFixed build error with GCC 3.3.1
nyan [Sun, 13 Jul 2003 08:13:52 +0000 (08:13 +0000)]
Fixed build error with GCC 3.3.1

21 years agoFixed mouse cursor support.
nyan [Sun, 13 Jul 2003 08:09:31 +0000 (08:09 +0000)]
Fixed mouse cursor support.

Submitted by: KIYOHARA Takashi <kiyohara@kk.iij4u.or.jp>

21 years agoAdd a missing DITEM_CONTINUE flag so the options menu returns to the
murray [Sun, 13 Jul 2003 07:29:06 +0000 (07:29 +0000)]
Add a missing DITEM_CONTINUE flag so the options menu returns to the
correct place when you exit it with 'Q'.

PR: misc/47906
MFC after: 3 days

21 years agoSort entries added in r1.43.
murray [Sun, 13 Jul 2003 07:20:24 +0000 (07:20 +0000)]
Sort entries added in r1.43.

21 years agoFrom the PR:
kris [Sun, 13 Jul 2003 06:38:13 +0000 (06:38 +0000)]
From the PR:

I am the maintainer of CTM.  There is a problem that when very large deltas
are created, that the program ctm_smail, which is responsible for mailing
the deltas out, will instead create a single message that says the delta
is too large.  However, if the -q option is set, instead of placing this
message in the queue (as it would have done with the deltas), it mails it
out directly.  This conflicts with the current working of CTM in that the
email address is set as %%REPLACE-ME%% so that the created mailing pieces
can be signed by gnu-pgp, and then have the mailing address changed.

This fix means that if the -q option is set, and the delta is too large,
the "too large" message is placed in the queue.

Also, I made the "too large" message a little more up to date.

Submitted by: Stephen Montgomery-Smith <stephen@math.missouri.edu>
PR: bin/50328
MFC After: 2 weeks

21 years agoFreeBSD does not use this code, but ftpd_popen() contains a buffer overflow.
kris [Sun, 13 Jul 2003 05:59:50 +0000 (05:59 +0000)]
FreeBSD does not use this code, but ftpd_popen() contains a buffer overflow.
We might as well patch it.

Submitted by:   Przemyslaw Frasunek <venglin@freebsd.lublin.pl>
PR:             bin/23352
MFC After: The average time before an unpatched Windows 2000 server gets owned

21 years agoRemove -D_THREAD_SAFE.
deischen [Sun, 13 Jul 2003 05:35:30 +0000 (05:35 +0000)]
Remove -D_THREAD_SAFE.

Submitted by:   Craig Rodrigues <rodrigc@crodrigues.org>

21 years agoEnable the high FP registers when we call the FPSWA handler and disable
marcel [Sun, 13 Jul 2003 04:08:16 +0000 (04:08 +0000)]
Enable the high FP registers when we call the FPSWA handler and disable
them again afterwards. This fixes a disabled FP fault while in the FPSWA
handler.
While here, merge the FP fault and FP trap handling code to reduce code
duplication. Where code was different, it was not sure it should be.

Trigger case: ports/math/atlas

21 years agoCorrect some typos.
kris [Sun, 13 Jul 2003 03:49:43 +0000 (03:49 +0000)]
Correct some typos.

Submitted by: pscott@the-frontier.org
PR: misc/51909
MFC After: The average lifetime of a Star Trek redshirt

21 years agoIncluding dynamically generated osreldate.h wasn't such a good
kan [Sun, 13 Jul 2003 02:41:48 +0000 (02:41 +0000)]
Including dynamically generated osreldate.h wasn't such a good
idea after all.

Fix cross-builds and ia64 builds. gnu/lib/csu/Makefile is modified to
pre-include osreldate.h and gnu/usr.bin/cc/cc_tools/auto-host.h
will avoid including sys/param.h if __FreeBSD_version is already defined.

21 years agoNot everyone uses the base system sshd. They can use the sshd_program
mtm [Sun, 13 Jul 2003 01:49:07 +0000 (01:49 +0000)]
Not everyone uses the base system sshd. They can use the sshd_program
variable in rc.conf to have sshd from ports (or somewhere else) installed.
So, don't make the sshd_config for the base system a required file
to start the service.

PR: conf/45766

21 years agoExtend the mutex pool implementation to permit the creation and use of
truckman [Sun, 13 Jul 2003 01:22:21 +0000 (01:22 +0000)]
Extend the mutex pool implementation to permit the creation and use of
multiple mutex pools with different options and sizes.  Mutex pools can
be created with either the default sleep mutexes or with spin mutexes.
A dynamically created mutex pool can now be destroyed if it is no longer
needed.

Create two pools by default, one that matches the existing pool that
uses the MTX_NOWITNESS option that should be used for building higher
level locks, and a new pool with witness checking enabled.

Modify the users of the existing mutex pool to use the appropriate pool
in the new implementation.

Reviewed by: jhb

21 years agoCheck the status of the user command and return the proper error
ps [Sat, 12 Jul 2003 22:37:03 +0000 (22:37 +0000)]
Check the status of the user command and return the proper error
to the user.

21 years agoQuiet down ciss unless bootverbose is set.
ps [Sat, 12 Jul 2003 22:32:27 +0000 (22:32 +0000)]
Quiet down ciss unless bootverbose is set.

21 years agoDo not override FP rounding options for 32bit platforms. amd64 doesn't
kan [Sat, 12 Jul 2003 19:33:34 +0000 (19:33 +0000)]
Do not override FP rounding options for 32bit platforms. amd64 doesn't
need this.

21 years agoAdd support for the A4 Tech RFSW-35 mouse wheel. Probe is similar to
mikeh [Sat, 12 Jul 2003 18:36:04 +0000 (18:36 +0000)]
Add support for the A4 Tech RFSW-35 mouse wheel. Probe is similar to
4D Plus.

PR: 44333

21 years agoThe second argument to fgetln() is a size_t *, not an int *.
tmm [Sat, 12 Jul 2003 17:56:50 +0000 (17:56 +0000)]
The second argument to fgetln() is a size_t *, not an int *.

21 years agoPostfix, pcre, and Exim should be on the first CD in order for the recent
trhodes [Sat, 12 Jul 2003 15:35:06 +0000 (15:35 +0000)]
Postfix, pcre, and Exim should be on the first CD in order for the recent
sysinstall(8) commit to function properly.

Discussed with: bmah (while the sysinstall(8) patch was sent to re).

21 years agoGive users the ability to select an alternative MTA during the installation.
trhodes [Sat, 12 Jul 2003 15:33:09 +0000 (15:33 +0000)]
Give users the ability to select an alternative MTA during the installation.
This option adds Postfix and Exim to the list, however, qmail is not added
due to license restrictions.

Collaborated with: Simon L. Nielsen <simon@nitro.dk>
Reviewed by: jhb, re@, -audit.

21 years ago- Allow retrieval of local Configuration ROM.
simokawa [Sat, 12 Jul 2003 09:36:53 +0000 (09:36 +0000)]
- Allow retrieval of local Configuration ROM.
- Clear Configuration ROM buffer in advance for '-d' option.

21 years agoAllow retrieval of local Configuration ROM.
simokawa [Sat, 12 Jul 2003 09:34:44 +0000 (09:34 +0000)]
Allow retrieval of local Configuration ROM.

21 years agoAdd a '-T' flag to print the timestamp as numeric value instead
luigi [Sat, 12 Jul 2003 08:35:25 +0000 (08:35 +0000)]
Add a '-T' flag to print the timestamp as numeric value instead
of converting it with ctime(). This is a lot more convenient for
postprocessing.

Submitted by: "Jacob S. Barrett" <jbarrett@amduat.net>

21 years agoNew release notes: GCC 3.3.1-pre, new ipfw features, rarpd bug fix,
hrs [Sat, 12 Jul 2003 07:41:39 +0000 (07:41 +0000)]
New release notes:  GCC 3.3.1-pre, new ipfw features, rarpd bug fix,
config.{guess,sub} in the ports infrastructure.

21 years agoDocument the existence of comments in ipfw rules,
luigi [Sat, 12 Jul 2003 07:01:48 +0000 (07:01 +0000)]
Document the existence of comments in ipfw rules,
the new flags handled when reading from a file,
and clarify that only numeric values are allowed for icmptypes.

MFC after: 3 days

21 years agoIn random order:
luigi [Sat, 12 Jul 2003 06:53:16 +0000 (06:53 +0000)]
In random order:

* make the code compile with WARNS=5 (at least on i386), mostly
  by adding 'const' specifier and replacing "void *" with "char *"
  in places where pointer arithmetic was used.
  This also spotted a few places where invalid tests (e.g. uint < 0)
  were used.

* support ranges in "list" and "show" commands. Now you can say

        ipfw show 100-1000 4000-8000

  which is very convenient when you have large rulesets.

* implement comments in ipfw commands. These are implemented in the
  kernel as O_NOP commands (which always match) whose body contains
  the comment string. In userland, a comment is a C++-style comment:

        ipfw add allow ip from me to any // i can talk to everybody

  The choice of '//' versus '#' is somewhat arbitrary, but because
  the preprocessor/readfile part of ipfw used to strip away '#',
  I did not want to change this behaviour.

  If a rule only contains a comment

        ipfw add 1000 // this rule is just a comment

  then it is stored as a 'count' rule (this is also to remind
  the user that scanning through a rule is expensive).

* improve handling of flags (still to be completed).
  ipfw_main() was written thinking of 'one rule per ipfw invocation',
  and so flags are set and never cleared. With readfile/preprocessor
  support, this changes and certain flags should be reset on each
  line. For the time being, only fix handling of '-a' which
  differentiates the "list" and "show" commands.

* rework the preprocessor support -- ipfw_main() already had most
  of the parsing code, so i have moved in there the only missing
  bit (stripping away '#' and comments) and removed the parsing
  from ipfw_readfile().
  Also, add some more options (such as -c, -N, -S) to the readfile
  section.

MFC after: 3 days

21 years agoImplement comments embedded into ipfw2 instructions.
luigi [Sat, 12 Jul 2003 05:54:17 +0000 (05:54 +0000)]
Implement comments embedded into ipfw2 instructions.

Since we already had 'O_NOP' instructions which always match, all
I needed to do is allow the NOP command to have arbitrary length
(i.e. move its label in a different part of the switch() which
validates instructions).

The kernel must know nothing about comments, everything else is
done in userland (which will be described in the upcoming ipfw2.c
commit).

21 years agoAdd logic to trace across/over a trapframe. We have ABI markers in
marcel [Sat, 12 Jul 2003 04:35:09 +0000 (04:35 +0000)]
Add logic to trace across/over a trapframe. We have ABI markers in
our unwind information for functions that are entry points into the
kernel. When stepping to the next frame, the unwinder will let us
know when sych a marker was encountered. We use this to stop the
current unwind session, query the trapframe and restart a new
unwind session based on the new trapframe.

The implementation is a bit sloppy, but at this time there are
bigger fish to fry.

21 years agoThis commit was generated by cvs2svn to compensate for changes in r117465,
marcel [Sat, 12 Jul 2003 04:20:56 +0000 (04:20 +0000)]
This commit was generated by cvs2svn to compensate for changes in r117465,
which included commits to RCS files with non-trunk default branches.

21 years agoFix a nasty bug that was exposed by disabling the trace support by
marcel [Sat, 12 Jul 2003 04:20:56 +0000 (04:20 +0000)]
Fix a nasty bug that was exposed by disabling the trace support by
default. There were 2 cases where the trace code was in the else-
part of an if-statement without any compound construct to protect
against nullification. The result was that 2 unconditional statements
turned into conditional statements and wrecked havoc.

This fix has been returned to the vendor and in anticipation of a
future import committed onto the vendor branch.

21 years agoMake the system call vector name of a process accessible to user
robert [Sat, 12 Jul 2003 02:00:16 +0000 (02:00 +0000)]
Make the system call vector name of a process accessible to user
land applications by introducing the KERN_PROC_SV_NAME sysctl node,
which is searchable by PID.

21 years agoCorrect a spelling error.
ceri [Sat, 12 Jul 2003 01:16:54 +0000 (01:16 +0000)]
Correct a spelling error.

21 years agoDelete the panic part of 20030711, it has been fixed
imp [Fri, 11 Jul 2003 23:40:16 +0000 (23:40 +0000)]
Delete the panic part of 20030711, it has been fixed

21 years agogcc3.3: add dire warnings.
imp [Fri, 11 Jul 2003 22:44:59 +0000 (22:44 +0000)]
gcc3.3: add dire warnings.

21 years agoFix the gcc-3.3 boot problem. Gcc now optimizes 'int foo = 0' by moving
peter [Fri, 11 Jul 2003 21:39:25 +0000 (21:39 +0000)]
Fix the gcc-3.3 boot problem.  Gcc now optimizes 'int foo = 0' by moving
it to the bss section and skips the initialization.  This causes all
sorts of havoc because the bogus bss zero code clobbered previously set
variables.  All our supported boot loaders already zero the bss, even
kgzip for the elf case.  Since we dont generate a.out kernels, the old
a.out bootblocks and the a.out kgzip are not a factor anymore.

21 years agoRegen.
joe [Fri, 11 Jul 2003 21:17:53 +0000 (21:17 +0000)]
Regen.

21 years agoAdded a few more entries.
joe [Fri, 11 Jul 2003 21:17:29 +0000 (21:17 +0000)]
Added a few more entries.

Submitted by: Lars Eggert <larse@ISI.EDU>

21 years agoSince (x)install is pretty unhelpfull and just spits out a Usage
phk [Fri, 11 Jul 2003 20:51:16 +0000 (20:51 +0000)]
Since (x)install is pretty unhelpfull and just spits out a Usage
for a lot of unrelated error conditions, at least report the line
number where it bailed.

Don't use multiline string literals for Usage, gcc 3.3 doesn't like them.

21 years agoUnlock the fxp driver lock before calling fxp_poll() to avoid recursing on
jhb [Fri, 11 Jul 2003 20:49:36 +0000 (20:49 +0000)]
Unlock the fxp driver lock before calling fxp_poll() to avoid recursing on
the lock when using DEVICE_POLLING.

Tested by: Robin P. Blanchard <Robin.Blanchard@gactr.uga.edu>
Reviewed by: njl

21 years agoInclude osreldate.h instead of sys/param.h to get __FreeBSD_version
kan [Fri, 11 Jul 2003 18:55:58 +0000 (18:55 +0000)]
Include osreldate.h instead of sys/param.h to get __FreeBSD_version
definition. This fixes buildworld breakage on ia64.

21 years agoLower WARNS to 3 so that this still compiles on non x86 architectures.
mux [Fri, 11 Jul 2003 17:15:19 +0000 (17:15 +0000)]
Lower WARNS to 3 so that this still compiles on non x86 architectures.

21 years agoFix typo for BURN_BRIDGES' file, it should be opt_global.h
imp [Fri, 11 Jul 2003 17:04:37 +0000 (17:04 +0000)]
Fix typo for BURN_BRIDGES' file, it should be opt_global.h

21 years agoForgot that the rescue subdir needs to be a conditional for the
gordon [Fri, 11 Jul 2003 17:01:58 +0000 (17:01 +0000)]
Forgot that the rescue subdir needs to be a conditional for the
build-tools bit.

21 years agoTurn rescue back on, conditional to NORESCUE. We seem to be split on
gordon [Fri, 11 Jul 2003 16:57:43 +0000 (16:57 +0000)]
Turn rescue back on, conditional to NORESCUE. We seem to be split on
using underscores or not, so I just randomly picked a style. I think
I have the logic correct, but if someone wants to give it a once over
that would be good.

Tim submitted a patch to fix the cross-building issues which I tested
with a tinderbox run for sparc64.

Submitted by: Tim Kientzle <kientzle@acm.org>

21 years agoNFS support should be conditional on LOADER_NFS_SUPPORT, not
tmm [Fri, 11 Jul 2003 16:12:50 +0000 (16:12 +0000)]
NFS support should be conditional on LOADER_NFS_SUPPORT, not
LOADER_NET_SUPPORT.

21 years agoForced commit to note that the previous commit fixed this PR:
mux [Fri, 11 Jul 2003 14:20:50 +0000 (14:20 +0000)]
Forced commit to note that the previous commit fixed this PR:

PR: i386/40957

21 years agoHuge cleanup of the rarpd(8) code :
mux [Fri, 11 Jul 2003 14:13:21 +0000 (14:13 +0000)]
Huge cleanup of the rarpd(8) code :
- Use getifaddrs() instead of rolling our own buggy one.  Previously,
  rarpd(8) would fail to see some interfaces because of a hardcoded limit.
  It now successfully sees any interface in the system, and this also makes
  the code _much_ simpler.
- Replace strncpy() calls with strlcpy() calls.  Some uses of strncpy()
  were bogus ; the code wasn't ensuring that the string was NUL terminated.
- Don't try to guard about select() FD_* macros being undefined.
- Use IF_NAMESIZE and ETHER_ADDR_LEN macros where appropriate.
- Add static keywords to function definitions for consistency, since
  the prototypes have it (I wonder why GCC didn't complain about this).
- Remove compat code for very old BSD versions and SunOS.
- Remove code for systems not having the dirent.h header.
- The code is now WARNS=5 clean so mark it as such.
- Don't add -DTFTP_DIR="/tftpboot" to the build command line since it's
  the default.

MFC after: 2 weeks

21 years agoFix if_wi breakage.
simokawa [Fri, 11 Jul 2003 13:50:31 +0000 (13:50 +0000)]
Fix if_wi breakage.

21 years agoIf a thread is sending signal to its process, if the thread can handle
davidxu [Fri, 11 Jul 2003 13:42:23 +0000 (13:42 +0000)]
If a thread is sending signal to its process, if the thread can handle
the signal itself, it should get it without looking for other threads.

21 years agoadd asus wl100
imp [Fri, 11 Jul 2003 09:25:18 +0000 (09:25 +0000)]
add asus wl100

21 years agoAdd support for asus wl100
imp [Fri, 11 Jul 2003 09:19:04 +0000 (09:19 +0000)]
Add support for asus wl100

PR: 49033/53249

21 years agosync to pccarddev 1.55
imp [Fri, 11 Jul 2003 09:17:07 +0000 (09:17 +0000)]
sync to pccarddev 1.55

21 years agoremove stray ASUS
imp [Fri, 11 Jul 2003 09:16:40 +0000 (09:16 +0000)]
remove stray ASUS

21 years agosync to pccarddev 1.54
imp [Fri, 11 Jul 2003 09:15:45 +0000 (09:15 +0000)]
sync to pccarddev 1.54

21 years agoAdd asus wl100
imp [Fri, 11 Jul 2003 09:15:21 +0000 (09:15 +0000)]
Add asus wl100

PR: 49033, 53249

21 years agoAdd a body directive before the first instruction in epc_syscall().
marcel [Fri, 11 Jul 2003 08:52:48 +0000 (08:52 +0000)]
Add a body directive before the first instruction in epc_syscall().
This results in a zero length prologue and a body that covers the
whole function. This is more correct.

21 years agoRemove a gratuitous align directive after the endp directive for
marcel [Fri, 11 Jul 2003 08:49:26 +0000 (08:49 +0000)]
Remove a gratuitous align directive after the endp directive for
IVT entries.