]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
24 years agoThis commit was generated by cvs2svn to compensate for changes in r54816,
peter [Sun, 19 Dec 1999 13:50:37 +0000 (13:50 +0000)]
This commit was generated by cvs2svn to compensate for changes in r54816,
which included commits to RCS files with non-trunk default branches.

24 years agoImport NetBSD's mailwrapper to redirect /usr/sbin/sendmail to the user's
peter [Sun, 19 Dec 1999 13:50:37 +0000 (13:50 +0000)]
Import NetBSD's mailwrapper to redirect /usr/sbin/sendmail to the user's
chosen mailer.

Obtained from: NetBSD

24 years agoDon't ignore return value from tsleep().
phk [Sun, 19 Dec 1999 12:36:41 +0000 (12:36 +0000)]
Don't ignore return value from tsleep().

Spotted by: charnier

24 years agoRemove all vestiges of /usr programs (awk, sed, grep...) from MAKEDEV.
green [Sun, 19 Dec 1999 09:42:15 +0000 (09:42 +0000)]
Remove all vestiges of /usr programs (awk, sed, grep...) from MAKEDEV.
MAKEDEV can now run (with /sbin/chown and /bin/chgrp, still) without
any /usr.  The default PATH for MAKEDEV is now "/sbin:/bin" to prevent
against further spammage of /usr programs in MAKEDEV.  In the event
of an emergency, the variable "MAKEDEVPATH" can be used to replace
PATH, but I don't know what that emergency could be.
   Let me know if you have any problems.  I've tested it as well as
I could, which basically means everything except for RocketPorts.

Reviewed in principle by: peter

24 years agoUse symbolic values rather than constants in TARGET_DEFAULT.
obrien [Sun, 19 Dec 1999 07:40:17 +0000 (07:40 +0000)]
Use symbolic values rather than constants in TARGET_DEFAULT.

Submitted by: bde on -current

24 years agoo Hide pcic_chip functions inside file
imp [Sun, 19 Dec 1999 07:39:10 +0000 (07:39 +0000)]
o Hide pcic_chip functions inside file
o Attach children in a better way than before.
o Remove some dead code
o collapse the socket attach routines into pcic_attach

24 years agoMFC: can't stop tweaking this compulsively.
jkh [Sun, 19 Dec 1999 07:02:46 +0000 (07:02 +0000)]
MFC: can't stop tweaking this compulsively.

24 years agocosmetics missed in last commit.
jkh [Sun, 19 Dec 1999 06:50:44 +0000 (06:50 +0000)]
cosmetics missed in last commit.

24 years agoUse a unified CPP_SPEC.
obrien [Sun, 19 Dec 1999 06:42:24 +0000 (06:42 +0000)]
Use a unified CPP_SPEC.

24 years agoMore cosmetic surgery and a nasty null pointer bug in index.c fixed.
jkh [Sun, 19 Dec 1999 06:34:22 +0000 (06:34 +0000)]
More cosmetic surgery and a nasty null pointer bug in index.c fixed.

24 years agoAllow LINT to build again.
green [Sun, 19 Dec 1999 06:22:11 +0000 (06:22 +0000)]
Allow LINT to build again.

The wd(4) family had to be disabled;  wd(4) and ata(4) are mutually
exclusive, even at link level.

24 years agoSecond pass commit to introduce new ACL and Extended Attribute system
rwatson [Sun, 19 Dec 1999 06:08:07 +0000 (06:08 +0000)]
Second pass commit to introduce new ACL and Extended Attribute system
calls, vnops, vfsops, both in /kern, and to individual file systems that
require a vfsop_ array entry.

Reviewed by: eivind

24 years agoFirst pass commit to introduce new ACL and Extended Attribute system calls.
rwatson [Sun, 19 Dec 1999 05:54:46 +0000 (05:54 +0000)]
First pass commit to introduce new ACL and Extended Attribute system calls.
The second pass commit with all the supporting code will happen shortly
afterwards.

Reviewed by: eivind

24 years agoi82365_isasubr is gone, remove from building module.
imp [Sun, 19 Dec 1999 05:49:56 +0000 (05:49 +0000)]
i82365_isasubr is gone, remove from building module.

24 years agoBe paranoid about refreshing after an action.
jkh [Sun, 19 Dec 1999 04:06:46 +0000 (04:06 +0000)]
Be paranoid about refreshing after an action.

24 years agoM_PREPEND-related cleanups (unregisterifying struct mbuf *s).
green [Sun, 19 Dec 1999 01:55:37 +0000 (01:55 +0000)]
M_PREPEND-related cleanups (unregisterifying struct mbuf *s).

24 years agoI've been style(9)izing mbuf.h and making the macros more resilient
green [Sun, 19 Dec 1999 01:47:16 +0000 (01:47 +0000)]
I've been style(9)izing mbuf.h and making the macros more resilient
with guidance from EE.  This is as far as I'm going without using
inline functions.

Reviewed by: eivind

24 years ago-Wall fixes.
billf [Sun, 19 Dec 1999 00:21:51 +0000 (00:21 +0000)]
-Wall fixes.

XXX: This code is sick, and so are some of the "fixes".

24 years agoFix a broken macro usage. It had no semicolon.
green [Sat, 18 Dec 1999 23:26:26 +0000 (23:26 +0000)]
Fix a broken macro usage.  It had no semicolon.

Noticed by: eivind

24 years agoSince VOP_LOCK can be used to up and downgrade locks, it is not possible
eivind [Sat, 18 Dec 1999 23:01:52 +0000 (23:01 +0000)]
Since VOP_LOCK can be used to up and downgrade locks, it is not possible
to say anything about the lockstate before and after it.  Thus, change the
lockspec from U L U to ? ? ?.

24 years agoDo NOT unlock dvp if LOCKPARENT && vp == dvp. Does not affect any code
eivind [Sat, 18 Dec 1999 22:57:38 +0000 (22:57 +0000)]
Do NOT unlock dvp if LOCKPARENT && vp == dvp.  Does not affect any code
presently in the tree.

24 years agomake vibra16x cards announce what they are in /dev/sndstat
cg [Sat, 18 Dec 1999 22:47:23 +0000 (22:47 +0000)]
make vibra16x cards announce what they are in /dev/sndstat

24 years agofix the buffer repetition on ^c problem
cg [Sat, 18 Dec 1999 22:24:50 +0000 (22:24 +0000)]
fix the buffer repetition on ^c problem

24 years agomake the vibra16x work for playback at least, record is untested
cg [Sat, 18 Dec 1999 22:21:47 +0000 (22:21 +0000)]
make the vibra16x work for playback at least, record is untested

24 years agoCorrect the documentation of LOCKPARENT - it does not reliably result in the
eivind [Sat, 18 Dec 1999 20:50:48 +0000 (20:50 +0000)]
Correct the documentation of LOCKPARENT - it does not reliably result in the
parent being locked, but rather plays some hide and seek (does not lock if
dvp == vp).

Also add a BUGS section noting that this is undesired behaviour.

24 years agoRelax the requirements for doing WDMA2, this should enable DMA
sos [Sat, 18 Dec 1999 20:06:30 +0000 (20:06 +0000)]
Relax the requirements for doing WDMA2, this should enable DMA
on non ATA[234] compliant disks.

Move all PCI dependent stuff under #ifdef NPCI > 0

24 years agoWoops, I'm so sorry I forgot this! From the last mbuf.h change:
green [Sat, 18 Dec 1999 20:04:19 +0000 (20:04 +0000)]
Woops, I'm so sorry I forgot this!  From the last mbuf.h change:

m_mballoc_wakeup() (inline) -> MMBWAKEUP() (macro)
m_clalloc_wakeup() (inline) -> MCLWAKEUP() (macro)

Noticed by: peter

24 years agoMake this compile again.
peter [Sat, 18 Dec 1999 19:33:59 +0000 (19:33 +0000)]
Make this compile again.

GENERIC-broken-by: imp
24 years ago Fix compilation warning on alpha when converting pointer to integer
dillon [Sat, 18 Dec 1999 19:20:05 +0000 (19:20 +0000)]
Fix compilation warning on alpha when converting pointer to integer
    to generate hash index.

Reviewed by:  Andrew Gallatin <gallatin@cs.duke.edu>

24 years agoMake NDFREE unlock VP if appropriate and not asked not to, as documented in
eivind [Sat, 18 Dec 1999 17:12:15 +0000 (17:12 +0000)]
Make NDFREE unlock VP if appropriate and not asked not to, as documented in
the manpage for namei().  This capability is not yet used attempted used
anywhere in the source tree.

24 years agoRevert -lmd changes now that libcrypt doesn't expose this binutils/ld
peter [Sat, 18 Dec 1999 16:42:33 +0000 (16:42 +0000)]
Revert -lmd changes now that libcrypt doesn't expose this binutils/ld
bug any more.

24 years agoRevert the libcrypt/libmd stuff back to how it was. This should not have
peter [Sat, 18 Dec 1999 13:55:17 +0000 (13:55 +0000)]
Revert the libcrypt/libmd stuff back to how it was.  This should not have
happened as it was working around problems elsewhere (ie: binutils/ld
not doing the right thing according to the ELF design).  libcrypt has
been adjusted to not need the runtime -lmd.  It's still not quite right
(ld is supposed to work damnit) but at least it doesn't impact all the
users of libcrypt in Marcel's cross-build model.

24 years agoI've been Bruceified! As penance, I've fixed my error (improper ordering,
green [Sat, 18 Dec 1999 13:52:44 +0000 (13:52 +0000)]
I've been Bruceified!  As penance, I've fixed my error (improper ordering,
moving #ifdefs to the non-historically-correct place, indentation bugs),
and I've also completely restyle(9)d mbuf.h.  There shouldn't be any
KNF issues with it anymore, nor should there be any internal
discrepencies.
   (Oh, I also fixed a small bug, but it was benign.  The spl was not
    released in MCLALLOC if M_NOWAIT and there were no free mclusters.)

Not objected to by: bde

24 years agoCorrect the logical ID of Avance Asound 110.
tanimura [Sat, 18 Dec 1999 12:17:02 +0000 (12:17 +0000)]
Correct the logical ID of Avance Asound 110.

Submitted by: Bryan Liesner <bleez@netaxs.com>

24 years agoUse long long format specifiers and cast each parameter to long long.
marcel [Sat, 18 Dec 1999 12:12:12 +0000 (12:12 +0000)]
Use long long format specifiers and cast each parameter to long long.
This avoids the ugly #if ARCH/#else/#endif construct.

Prompted to by: bde
Could have thought of this himself: marcel

24 years agoProvide complete coverage in switch statement to quiet -Wall.
obrien [Sat, 18 Dec 1999 09:56:15 +0000 (09:56 +0000)]
Provide complete coverage in switch statement to quiet -Wall.

24 years agospell isa right on sn0 line
imp [Sat, 18 Dec 1999 07:49:54 +0000 (07:49 +0000)]
spell isa right on sn0 line

24 years agoMerge GENERIC into PCCARD
imp [Sat, 18 Dec 1999 07:47:34 +0000 (07:47 +0000)]
Merge GENERIC into PCCARD
Merge PCCARD into NEWCARD

24 years agoRemove vestiages of now obsolete zp and ze drivers.
imp [Sat, 18 Dec 1999 06:27:31 +0000 (06:27 +0000)]
Remove vestiages of now obsolete zp and ze drivers.

Forgotten by: phk

24 years agoAdd sn device: smc 91xx based ethernet controllers
imp [Sat, 18 Dec 1999 06:23:03 +0000 (06:23 +0000)]
Add sn device: smc 91xx based ethernet controllers

24 years agoDriver for the smc91xx series of ethernet chips. Ported from PAO to
imp [Sat, 18 Dec 1999 06:11:22 +0000 (06:11 +0000)]
Driver for the smc91xx series of ethernet chips.  Ported from PAO to
3.3R and then to -current.  The pccard support has been left in the
driver, but is presently non-functional because we are using the
isa_compat layer for the moment.

Obtained From: PAO
Sponsored by: Timing Solutions

24 years agoInitialize a var to quiet -Wall.
obrien [Sat, 18 Dec 1999 04:47:43 +0000 (04:47 +0000)]
Initialize a var to quiet -Wall.

24 years agoSwitch over to the OpenBSD fts.c, fixing lots of things.
green [Sat, 18 Dec 1999 04:36:14 +0000 (04:36 +0000)]
Switch over to the OpenBSD fts.c, fixing lots of things.

Obtained from: OpenBSD

24 years agoYet more cosmetic fixes (I have a little time to kill while waiting for
jkh [Sat, 18 Dec 1999 03:13:19 +0000 (03:13 +0000)]
Yet more cosmetic fixes (I have a little time to kill while waiting for
something else, so I might as well tweak 3.4's look-and-feel for the better
while I'm at it - final release build should be sometime later on tonite).

24 years agoLearn to do pointer arithmatic (doh!)
brian [Sat, 18 Dec 1999 02:31:36 +0000 (02:31 +0000)]
Learn to do pointer arithmatic (doh!)

24 years agoAdd more strategic screen clears
jkh [Sat, 18 Dec 1999 02:29:24 +0000 (02:29 +0000)]
Add more strategic screen clears

24 years agoBetter screen handling in X desktop setup.
jkh [Sat, 18 Dec 1999 02:04:02 +0000 (02:04 +0000)]
Better screen handling in X desktop setup.
Don't get cute with nested dialogs now.

24 years agoRemove duplicate entry
cpiazza [Sat, 18 Dec 1999 01:26:38 +0000 (01:26 +0000)]
Remove duplicate entry

24 years agoSpelling fixes. first change from openbsd second from netbsd
cpiazza [Sat, 18 Dec 1999 01:13:18 +0000 (01:13 +0000)]
Spelling fixes.  first change from openbsd second from netbsd

24 years agoFix some minor POSIX/SUSv2 compliance nits.
jasone [Sat, 18 Dec 1999 01:00:42 +0000 (01:00 +0000)]
Fix some minor POSIX/SUSv2 compliance nits.

PR: kern/11982

24 years agoFix bugs in the MP fragment reassembly code that can cause a panic.
archie [Fri, 17 Dec 1999 23:29:04 +0000 (23:29 +0000)]
Fix bugs in the MP fragment reassembly code that can cause a panic.

24 years agoBe more consistant in using perl vs. perl5. We were using perl5 in the
peter [Fri, 17 Dec 1999 22:07:43 +0000 (22:07 +0000)]
Be more consistant in using perl vs. perl5.  We were using perl5 in the
kernel builds so as not to confuse with perl4 when bootstrapping from old
systems.  I don't know if this is still applicable but it shouldn't hurt
to be consistant at least.

Also copy vnode_if.sh to vnode_if.pl.  Doing a 'sh vnode_if.sh' when it
was a perl script was kinda silly.

24 years agoMake a dlopen failure consistant with dlsym(). "Shouldn't happen."
peter [Fri, 17 Dec 1999 20:21:45 +0000 (20:21 +0000)]
Make a dlopen failure consistant with dlsym().  "Shouldn't happen."

24 years agopatch glitch
peter [Fri, 17 Dec 1999 20:19:28 +0000 (20:19 +0000)]
patch glitch

24 years agoRemove -lmd. Use dlopen() and dlsym() instead for calls to the MD5* and
peter [Fri, 17 Dec 1999 20:04:01 +0000 (20:04 +0000)]
Remove -lmd.  Use dlopen() and dlsym() instead for calls to the MD5* and
SHA* routines so that callers of libcrypt are not exposed to the internal
implementation.

24 years agoAdd character major 139 for the Sangoma Technologies 'wanrouter' driver.
msmith [Fri, 17 Dec 1999 19:56:39 +0000 (19:56 +0000)]
Add character major 139 for the Sangoma Technologies 'wanrouter' driver.

Contact details are:

Alex Feldman
E-mail : al.feldman@sangoma.com
Sangoma Technologies Inc.
1001 Denison st., Suite 101
Markham, L3R 2Z6    ON
Phone: (905) 474 - 1990
Fax     : (905) 388 - 2475

24 years agos/MACHINE_ARCH/TARGET_ARCH/g
marcel [Fri, 17 Dec 1999 15:50:45 +0000 (15:50 +0000)]
s/MACHINE_ARCH/TARGET_ARCH/g

The target machine is represented by TARGET_ARCH. MACHINE_ARCH always
represents the host machine. When TARGET_ARCH is not defined, it is
assumed to be equal to MACHINE_ARCH. This means that we're building a
native toolset by default. We're creating cross-compilation tools when
MACHINE_ARCH != TARGET_ARCH.

TARGET_ARCH is defined when building binutils as part of the bootstrap
build and is set to reflect the architecture we're currently cross-
building. With this change binutils is ready for cross-building.

24 years agoo Don't use long for 64 bit variables. Use u_int64_t instead.
marcel [Fri, 17 Dec 1999 15:27:21 +0000 (15:27 +0000)]
o  Don't use long for 64 bit variables. Use u_int64_t instead.
o  Fix printf format specifiers on Alpha and add i386 version.
o  Minor style fixes while there.
o  Add -Wall compiler option.

24 years agoBack up following macros by functions: ishexnumber, isideogram, isnumber,
phantom [Fri, 17 Dec 1999 15:12:21 +0000 (15:12 +0000)]
Back up following macros by functions: ishexnumber, isideogram, isnumber,
isphonogram, isrune, isspecial. Fix ordering.

Reviewed by: bde

24 years agoFix typos
phantom [Fri, 17 Dec 1999 14:48:00 +0000 (14:48 +0000)]
Fix typos

PR: docs/14858
Submitted by: OKAZAKI Tetsurou <okazaki@be.to>

24 years agoDocument -e flag.
phantom [Fri, 17 Dec 1999 14:43:33 +0000 (14:43 +0000)]
Document -e flag.

PR: docs/14936
Submitted by: Stephen J. Roznowski <sjr@home.com>

24 years agoDocument -d and -o flags
phantom [Fri, 17 Dec 1999 14:34:02 +0000 (14:34 +0000)]
Document -d and -o flags
Remove selfreference item from SEE ALSO section
Add reference to colldef(1) to SEE ALSO section

24 years agoxntpd -> ntpd.
roberto [Fri, 17 Dec 1999 13:36:40 +0000 (13:36 +0000)]
xntpd -> ntpd.

Submitted by: ru

24 years agoChange to work with recent signal changes. The signal being handled is
deischen [Fri, 17 Dec 1999 11:46:55 +0000 (11:46 +0000)]
Change to work with recent signal changes.  The signal being handled is
now added to the signal mask; this test failed because it didn't allow
for this.

24 years agoAdd libmd to DPADD and LDADD.
marcel [Fri, 17 Dec 1999 11:45:28 +0000 (11:45 +0000)]
Add libmd to DPADD and LDADD.

24 years agoDon't build usr.sbin/pcvt/keycap on anything other than i386 when
marcel [Fri, 17 Dec 1999 11:23:56 +0000 (11:23 +0000)]
Don't build usr.sbin/pcvt/keycap on anything other than i386 when
building the libraries target. pcvt is i386 specific.

24 years agoBack out previous commit, which broke mouse wheel support in moused.
sheldonh [Fri, 17 Dec 1999 10:05:36 +0000 (10:05 +0000)]
Back out previous commit, which broke mouse wheel support in moused.

Requested by: yokota

24 years agoo Separate out alloc/release from the activation
imp [Fri, 17 Dec 1999 09:00:11 +0000 (09:00 +0000)]
o Separate out alloc/release from the activation
o minor formatting nits
o remove the bus_map/unmap calls.  FreeBSD doesn't need them.
o Fix most of the compiler warnings remaining (some still remain, but they
  are releated to calling functions that are going to change, so I did
  not bother).

24 years agoInstall Flexlexer.h with mode 444.
bde [Fri, 17 Dec 1999 07:49:03 +0000 (07:49 +0000)]
Install Flexlexer.h with mode 444.

24 years agoNote that the PR associated with the previous log message is i386/7426,
jasone [Fri, 17 Dec 1999 07:48:22 +0000 (07:48 +0000)]
Note that the PR associated with the previous log message is i386/7426,
not bin/8281.

24 years agoTwo more fixes to if_detach. These are generic to all interfaces and
imp [Fri, 17 Dec 1999 06:46:07 +0000 (06:46 +0000)]
Two more fixes to if_detach.  These are generic to all interfaces and
do not pollute the interface further.

o Run if_detach at splnet().
o Creatively swipe the relevant parts of the netatm atm_nif_detach
  which will delete the relevant references to the interface going
  away.

24 years agoAdd a missing '"' in the examples.
obrien [Fri, 17 Dec 1999 06:04:37 +0000 (06:04 +0000)]
Add a missing '"' in the examples.

24 years agoAdd another strategic screen clear.
jkh [Fri, 17 Dec 1999 03:00:28 +0000 (03:00 +0000)]
Add another strategic screen clear.

24 years agoIn retrospect, msgNotify() should leave its contents on the screen
jkh [Fri, 17 Dec 1999 02:46:04 +0000 (02:46 +0000)]
In retrospect, msgNotify() should leave its contents on the screen
longer to give the user something to look at while things are happening.
Change it to do so and insert the appropriate screen saves elsewhere.

24 years ago*Really* make space accelerators work; no wonder Brian McGovern was
jkh [Fri, 17 Dec 1999 02:44:09 +0000 (02:44 +0000)]
*Really* make space accelerators work; no wonder Brian McGovern was
confused!  I fat-fingered the first commit.

24 years agonewpcm no longer requires 'controller pnp'. (And some other drivers?)
tanimura [Fri, 17 Dec 1999 02:31:36 +0000 (02:31 +0000)]
newpcm no longer requires 'controller pnp'. (And some other drivers?)

Noticed by: julian

24 years agoBump default root size to 50MB on i386 (70MB on Alpha).
jkh [Fri, 17 Dec 1999 02:02:51 +0000 (02:02 +0000)]
Bump default root size to 50MB on i386 (70MB on Alpha).

Requested incessantly by: billf

24 years agoFix handling of trailing :'s to match what other OSes do (spit out
jkh [Fri, 17 Dec 1999 01:52:15 +0000 (01:52 +0000)]
Fix handling of trailing :'s to match what other OSes do (spit out
a diagnostis).

Submitted by: Guy Harris <gharris@flashcom.net>

24 years agoChange location of temporary file from /tmp to /var/tmp. This is a
grog [Fri, 17 Dec 1999 01:18:44 +0000 (01:18 +0000)]
Change location of temporary file from /tmp to /var/tmp.  This is a
repeat of an earlier commit which apparently got lost with the last
import.  It helps solve the frequently reported problem

   pid 4032 (mail.local), uid 0 on /: file system full

(though there appears to be a lot of space) caused by idiots sending
30 MB mail messages.

Most-recently-reported-by: jahanur <jahanur@jjsoft.com>
Add $FreeBSD$ so that I can check the file back in.

Rejected-by: CVS

24 years agoFix problems with cancellation while in critical regions.
deischen [Fri, 17 Dec 1999 00:57:54 +0000 (00:57 +0000)]
Fix problems with cancellation while in critical regions.

  o Cancellation flags were not getting properly set/cleared.
  o Loops waiting for internal locks were not being exited
    correctly by a cancelled thread.
  o Minor spelling (cancelation -> cancellation) and formatting
    corrections (missing tab).

Found by: tg
Reviewed by: jasone

24 years agoFixes for signal handling:
deischen [Fri, 17 Dec 1999 00:56:36 +0000 (00:56 +0000)]
Fixes for signal handling:

  o Don't call signal handlers with the signal handler access lock
    held.
  o Remove pending signals before calling signal handlers.  If
    pending signals were not removed prior to handling them,
    invocation of the handler could cause the handler to be
    called more than once for the same signal.  Found by: JB
  o When SIGCHLD arrives, wake up all threads in PS_WAIT_WAIT
    (wait4).

PR: bin/15328
Reviewed by: jasone

24 years agoFix the example in the bugs section.
ken [Fri, 17 Dec 1999 00:28:05 +0000 (00:28 +0000)]
Fix the example in the bugs section.

24 years agoDocument NDFREE w/flags, and fix up the description of NOFOLLOW.
eivind [Thu, 16 Dec 1999 23:20:11 +0000 (23:20 +0000)]
Document NDFREE w/flags, and fix up the description of NOFOLLOW.

24 years agoDocument pthread_create().
chris [Thu, 16 Dec 1999 22:36:54 +0000 (22:36 +0000)]
Document pthread_create().

Reviewed by: jasone

24 years agoAvoid an infinite loop if the last element of the iov array passed to
jasone [Thu, 16 Dec 1999 22:35:40 +0000 (22:35 +0000)]
Avoid an infinite loop if the last element of the iov array passed to
writev() has an iov_len of 0.

PR: bin/8281

24 years agoThe function request_cleanup() had a tsleep() with PCATCH. It is
mckusick [Thu, 16 Dec 1999 22:02:09 +0000 (22:02 +0000)]
The function request_cleanup() had a tsleep() with PCATCH. It is
quite dangerous, since the process may hold locks at the point,
and if it is stopped in that tsleep the machine may hang. Because
the sleep is so short, the PCATCH is not required here, so it has
been removed. For the future, the FreeBSD team needs to decide
whether it is still reasonable to stop a process in tsleep, as that
may affect any other code that uses PCATCH while holding kernel locks.

Submitted by: Dmitrij Tejblum <tejblum@arc.hq.cti.ru>
Reviewed by: Kirk McKusick <mckusick@mckusick.com>

24 years agoMove POSIX feature constants from pthread.h to unistd.h, as required by
jasone [Thu, 16 Dec 1999 21:53:39 +0000 (21:53 +0000)]
Move POSIX feature constants from pthread.h to unistd.h, as required by
1003.1c-1995.

Undefine _POSIX_THREAD_SAFE_FUNCTIONS, since we do not implement all of the
necessary interfaces.  At least getgrgid_r(), getrnam_r(), getpwuid_r(),
getpwnam_r(), getc_unlocked(), getchar_unlocked(), putc_unlocked(),
and putchar_unlocked() are missing.  Due to a likely typo in 1003.1c-1995,
we are not technically allowed to define _POSIX_THREADS without defining
_POSIX_THREAD_SAFE_FUNCTIONS, but either way we're non-compliant, so I'm
leaving _POSIX_THREADS defined.

PR: bin/8281

24 years agoUpdate the xl driver to recognize yet another 3c905B/3c905C class NIC:
wpaul [Thu, 16 Dec 1999 18:33:57 +0000 (18:33 +0000)]
Update the xl driver to recognize yet another 3c905B/3c905C class NIC:
the 3c450-TX HomeConnect. Like the 3cSOHO100-TX OfficeConnect, this NIC
uses the same ASIC as the 3c905B/3c905C but is targeted for a particular
market segment (home users). It is somewhat less expensive than the
3c905B/3c905C ($49, according to the 3Com web site), comes with its
own custom driver kit and is bundled with various goofy Windows software
packages designed to demonstrate the niftyness of home networking (networked
game demos, etc...).

Changes are:

- Add PCI ID to list in if_xlreg.h.
- Update xl_devs table in if_xl.c.
- Update xl_choose_xcvr() to consider the HomeConnect the
  same as all the other 10baseT/100baseTX cards.

24 years agoKAME 4th patch
shin [Thu, 16 Dec 1999 18:32:01 +0000 (18:32 +0000)]
KAME 4th patch
IPv6 specific library functions addition.
(getnameinfo(), getaddrinfo(), and IPv6 transport support is not yet)

Reviewed by: freebsd-arch, cvs-committers
Obtained from: KAME project

24 years agoDon't build gensetdefs when we can use /usr/bin/gensetdefs. The latter
marcel [Thu, 16 Dec 1999 17:25:01 +0000 (17:25 +0000)]
Don't build gensetdefs when we can use /usr/bin/gensetdefs. The latter
also creates setdef{0|1}.c so there's no need to have those in the
repository. Using /usr/bin/gensetdefs has no consequences for the
output.

24 years ago Have NFS use a snapshot of boottime instead of boottime itself to
dillon [Thu, 16 Dec 1999 17:01:32 +0000 (17:01 +0000)]
Have NFS use a snapshot of boottime instead of boottime itself to
    generate the NFSv3 Version id.  boottime itself may change, sometimes
    once every tick if you are running xntpd, which really throws off
    clients.  Clients will tend to throw away what they believe to be
    stale data too often, and can get into long loops rewriting the same
    data over and over again because they believe the server has rebooted
    over and over again due to the changing version id.

Approved by: jkh

24 years agoThis commit was generated by cvs2svn to compensate for changes in r54690,
obrien [Thu, 16 Dec 1999 17:00:53 +0000 (17:00 +0000)]
This commit was generated by cvs2svn to compensate for changes in r54690,
which included commits to RCS files with non-trunk default branches.

24 years ago* The C1010 stepping B0 (Rev 1) tested OK for DT transfers without the U3EN
obrien [Thu, 16 Dec 1999 17:00:53 +0000 (17:00 +0000)]
* The C1010 stepping B0 (Rev 1) tested OK for DT transfers without the U3EN
  broken bit work-around enabled.
* Fixed a bug that made MDP not work. (However, MDP is actually not tested
  due to lack of hardware using this feature).
* Chip table changed to support the C1010 B0 w/o the U3EN bit work-around
  enabled.
* Add the SYM_SETUP_MAX_LUN, SYM_SETUP_LP_PROBE_MAP (used to tell the
  driver about chips that are to be claimed with lower priority than old
  PCI bus based driver (typically the ncr)), SYM_SETUP_SCSI_DIFF, and
  SYM_SETUP_PCI_PARITY options.

Submitted by: Gerard Roudier <groudier@club-internet.fr>

24 years agoactivate amd.4
phantom [Thu, 16 Dec 1999 16:22:48 +0000 (16:22 +0000)]
activate amd.4

24 years agoAdjust data pointers for untransmitted bytes in fifos when there's phase
luoqi [Thu, 16 Dec 1999 16:11:49 +0000 (16:11 +0000)]
Adjust data pointers for untransmitted bytes in fifos when there's phase
change during data transfer.

24 years agoAdd options used by the `sym' (Symbios SCSI) driver.
obrien [Thu, 16 Dec 1999 16:09:57 +0000 (16:09 +0000)]
Add options used by the `sym' (Symbios SCSI) driver.
SYM_SETUP_LP_PROBE_MAP Low Priority Probe Map (bits)
SYM_SETUP_SCSI_DIFF HVD support for 825a, 875, 885
SYM_SETUP_PCI_PARITY PCI parity checking
SYM_SETUP_MAX_LUN Number of LUNs supported

Submitted by: Gerard Roudier <groudier@club-internet.fr>

24 years agoAdd FAST-80 timing to the scsi syncrates table.
obrien [Thu, 16 Dec 1999 16:04:54 +0000 (16:04 +0000)]
Add FAST-80 timing to the scsi syncrates table.

Submitted by: Gerard Roudier <groudier@club-internet.fr>

24 years agoo Add games/caesar to the list of bootstrap-tools so that a
marcel [Thu, 16 Dec 1999 13:47:34 +0000 (13:47 +0000)]
o  Add games/caesar to the list of bootstrap-tools so that a
   buildworld doesn't break because the host doesn't have any
   games installed,
o  Add a new build stage: TMAKE. TMAKE builds all the build-tools
   targets in the respective makefiles. Note that these targets
   don't use the bootstrap tools,
o  Add elf2exe to the bootstrap-tools when cross-building Alpha on
   other platforms,
o  Add ${WORLDTMP}/usr/games to TMPPATH,
o  Remove ${WORLDTMP}/usr/bin even when NOCLEAN is defined. This
   prevents using any bootstrap-tools previously installed. Most
   importantly, it prevents using the cross-compiler when we still
   need the native compiler.

The current stages are BMAKE, TMAKE, XMAKE and IMAKE in that order.
BMAKE builds bootstrap-tools that either solve compatibility problems
      or are needed as cross-tools,
TMAKE builds the support tools necessary by some parts in the source
      tree and also performs the cleandir and par-obj targets,
XMAKE builds the includes, libraries and everything (resp.), and
IMAKE installs the world. This stage needs further work if it's to be
      used to install -current over -stable for example.

This is the last major update towards cross-building.

24 years agoBye bye xntpd, enter ntp.
roberto [Thu, 16 Dec 1999 12:45:40 +0000 (12:45 +0000)]
Bye bye xntpd, enter ntp.

The variable names haven't changed for compapatibility reasons.

24 years agoBye bye xntpd, enter ntp.
roberto [Thu, 16 Dec 1999 12:42:09 +0000 (12:42 +0000)]
Bye bye xntpd, enter ntp.

Reviewed by: phk