]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
26 years agoFix a recursive kernel_map lock problem in vm_zone allocator.
John Dyson [Mon, 15 Dec 1997 05:16:09 +0000 (05:16 +0000)]
Fix a recursive kernel_map lock problem in vm_zone allocator.
PR: 5298

26 years ago- Fix some very minor typos.
Nate Williams [Mon, 15 Dec 1997 04:38:15 +0000 (04:38 +0000)]
- Fix some very minor typos.

26 years agoAdd support for poll(2) on files. vop_nopoll() now returns POLLNVAL
Garrett Wollman [Mon, 15 Dec 1997 03:09:59 +0000 (03:09 +0000)]
Add support for poll(2) on files.   vop_nopoll() now returns POLLNVAL
if one of the new poll types is requested; hopefully this will not break
any existing code.  (This is done so that programs have a dependable
way of determining whether a filesystem supports the extended poll types
or not.)

The new poll types added are:

POLLWRITE - file contents may have been modified
POLLNLINK - file was linked, unlinked, or renamed
POLLATTRIB - file's attributes may have been changed
POLLEXTEND - file was extended

Note that the internal operation of poll() means that it is impossible
for two processes to reliably poll for the same event (this could
be fixed but may not be worth it), so it is not possible to rewrite
`tail -f' to use poll at this time.

26 years agoOops, I forgot to set `cflag' in some cases when initializing the
Kazutaka YOKOTA [Mon, 15 Dec 1997 02:57:45 +0000 (02:57 +0000)]
Oops, I forgot to set `cflag' in some cases when initializing the
mouse port.

26 years agoConsult sa_len before trampling it with MSG_COMPAT set.
Mike Smith [Mon, 15 Dec 1997 02:29:11 +0000 (02:29 +0000)]
Consult sa_len before trampling it with MSG_COMPAT set.
PR:             kern/5291
Submitted by:   pb@fasterix.freenix.org (Pierre Beyssac)

26 years agoAdd support for low resolution SMP kernel profiling.
Tor Egge [Mon, 15 Dec 1997 02:18:35 +0000 (02:18 +0000)]
Add support for low resolution SMP kernel profiling.

  - A nonprofiling version of s_lock (called s_lock_np) is used
    by mcount.

  - When profiling is active, more registers are clobbered in
    seemingly simple assembly routines. This means that some
    callers needed to save/restore extra registers.

  - The stack pointer must have space for a 'fake' return address
    in idle, to avoid stack underflow.

26 years agoFix recursion problem which occurs when a signal is received during
John Birrell [Mon, 15 Dec 1997 02:12:42 +0000 (02:12 +0000)]
Fix recursion problem which occurs when a signal is received during
a malloc. The signal handler creates a thread which requires a malloc...
For now, the only thing to do is to block signals. When we move user
pthreads to use the kernel threads, mutexes will be implemented in kernel
space and then malloc can revert.

26 years agoDon't forward hardclock or statclock to stopped cpus. Disable forwarding
Tor Egge [Mon, 15 Dec 1997 01:14:10 +0000 (01:14 +0000)]
Don't forward hardclock or statclock to stopped cpus. Disable forwarding
when a panic has occured.

26 years agoExplicitly use the _IOC macro directly. This changes the ioctl command
Sean Eric Fagan [Mon, 15 Dec 1997 00:29:41 +0000 (00:29 +0000)]
Explicitly use the _IOC macro directly.  This changes the ioctl command
numbers, so any programs using it need to be recompiled.

26 years agoMake gzipped dynamically linked executables work again. There was
John Polstra [Sun, 14 Dec 1997 19:36:24 +0000 (19:36 +0000)]
Make gzipped dynamically linked executables work again.  There was
an old bug here that failed to copy the a.out header into memory
properly.  It didn't matter until changes were made recently to
the dynamic linker.

26 years agoOptimize rm(1) usage.
Wolfram Schneider [Sun, 14 Dec 1997 15:38:47 +0000 (15:38 +0000)]
Optimize rm(1) usage.

26 years agoSync with sys/i386/i386/machdep.c revision 1.279.
KATO Takenori [Sun, 14 Dec 1997 12:31:44 +0000 (12:31 +0000)]
Sync with sys/i386/i386/machdep.c revision 1.279.

26 years agotypo fix.
Jordan K. Hubbard [Sun, 14 Dec 1997 12:30:59 +0000 (12:30 +0000)]
typo fix.
PR: 5289
Submitted by: Studded@dal.net

26 years agoSync with sys/i386/conf/files.i386 revision 1.182.
KATO Takenori [Sun, 14 Dec 1997 12:30:54 +0000 (12:30 +0000)]
Sync with sys/i386/conf/files.i386 revision 1.182.

26 years agoSlight improvement to the vm_zone stats output. Also, some other superficial
John Dyson [Sun, 14 Dec 1997 05:17:44 +0000 (05:17 +0000)]
Slight improvement to the vm_zone stats output.  Also, some other superficial
cleanups.

26 years agoAs described by the submitter:
Mike Smith [Sun, 14 Dec 1997 03:17:54 +0000 (03:17 +0000)]
As described by the submitter:

- emulate Linux IP_HDRINCL behaviour in sendto(): byte order fixed
  Note that we do an extra getsockopt() on every sendto()
  to check if the option is set because we don't keep state
  in the emulator code. Is there a better way to implement
  this?
- correct a bug (value of "name" not passed) with
  getsockopt()

Submitted by: pb@fasterix.freenix.org (Pierre Beyssac)

26 years agoAs described by the submitter:
Mike Smith [Sun, 14 Dec 1997 03:15:21 +0000 (03:15 +0000)]
As described by the submitter:

... fix a bug with orecvfrom() or recvfrom() called with
the MSG_COMPAT flag on kernels compiled with the COMPAT_43 option.
The symptom is that the fromaddr is not correctly returned.

This affects the Linux emulator.

Submitted by: pb@fasterix.freenix.org (Pierre Beyssac)

26 years agoAfter one of my analysis passes to evaluate methods for SMP TLB mgmt, I
John Dyson [Sun, 14 Dec 1997 02:11:23 +0000 (02:11 +0000)]
After one of my analysis passes to evaluate methods for SMP TLB mgmt, I
noticed some major enhancements available for UP situations.  The number
of UP TLB flushes is decreased much more than significantly with these
changes.  Since a TLB flush appears to cost minimally approx 80 cycles,
this is a "nice" enhancement, equiv to eliminating between 40 and 160
instructions per TLB flush.

Changes include making sure that kernel threads all use the same PTD,
and eliminate unneeded PTD switches at context switch time.

26 years agoAllow slow SCSI CDROMs more time to start up.
James Raynard [Sun, 14 Dec 1997 00:32:33 +0000 (00:32 +0000)]
Allow slow SCSI CDROMs more time to start up.
PR: 5181
Submitted by: Wilko Bulte <wilko@yedi.iaf.nl>

26 years agoBring in NetBSD compatible & augmented RCSID/COPYRIGHT macros
Paul Traina [Sat, 13 Dec 1997 20:39:14 +0000 (20:39 +0000)]
Bring in NetBSD compatible & augmented RCSID/COPYRIGHT macros

26 years agoSync with NetBSD v1.3.
Paul Traina [Sat, 13 Dec 1997 20:38:21 +0000 (20:38 +0000)]
Sync with NetBSD v1.3.

Keep I8N1 and quarantine port code.  Replace our security fix with NetBSD's
(their idea was better).

26 years agoOops..The default behaviour should be *not* to log all succeeded
Guido van Rooij [Sat, 13 Dec 1997 19:50:14 +0000 (19:50 +0000)]
Oops..The default behaviour should be *not* to log all succeeded
requests. If you want this extra loggin, add the -l option.

26 years agoAllow TMPDIR to be set and actually *used*; if it not set, it defaults
Sean Eric Fagan [Sat, 13 Dec 1997 18:18:12 +0000 (18:18 +0000)]
Allow TMPDIR to be set and actually *used*; if it not set, it defaults
to /tmp.

26 years agoFixed .PATH statement so that minigzip.c can be found when
Bruce Evans [Sat, 13 Dec 1997 15:57:55 +0000 (15:57 +0000)]
Fixed .PATH statement so that minigzip.c can be found when
${.OBJDIR} != ${.CURDIR}.

26 years agoFixed EOF handing.
Bruce Evans [Sat, 13 Dec 1997 13:49:59 +0000 (13:49 +0000)]
Fixed EOF handing.

1. SS_CANTRCVMORE was initially set on the wrong socket, so reads
when there has never been a writer on the socket did not return 0.
Note that such reads are only possible if the fifo was opened in
(O_RDONLY | O_NONBLOCK) mode.

2. SS_CANTSENDMORE was initially set on the wrong socket, but this
was harmless because the wrong socket is never sent from and there
is no need to set the flag initially on the right socket (since open
in (O_WRONLY | O_NONBLOCK) mode fails if there is no reader...).

3. SS_CANTRCVMORE was cleared when read() returns.  This broke the
case where read() returns 0 - subsequent reads are supposed to
return 0 until a writer appears.  There is no need to clear the
flag when read() returns, since it is cleared correctly when a
writer appears.

26 years agoRestored fifo_pathconf() from rev.1.32. vop_stdpathconf() is too
Bruce Evans [Sat, 13 Dec 1997 12:58:09 +0000 (12:58 +0000)]
Restored fifo_pathconf() from rev.1.32.  vop_stdpathconf() is too
general to be of much use.  Using it here weakened the _PC_MAX_CANON,
_PC_MAX_INPUT and _PC_VDISABLE cases.

fifo_pathconf() is not quite correct either.  _PC_CHOWN_RESTRICTED
and _PC_LINK_MAX should be handled by the host file system.  For
directories, the host file system should let us handle _PC_PIPE_BUF.

26 years agoRestored ufs_pathconf() from rev.1.61. vop_stdpathconf() is too
Bruce Evans [Sat, 13 Dec 1997 12:30:34 +0000 (12:30 +0000)]
Restored ufs_pathconf() from rev.1.61.  vop_stdpathconf() is too
general to be of much use.  Using it here broke the _PC_NAME_MAX,
_PC_NO_TRUNC and _PC_PATH_MAX cases, and weakened the _PC_MAX_CANON,
_PC_MAX_INPUT and _PC_VDISABLE cases.

26 years agoDelete "typedef ... date" (see style(9)).
Wolfgang Helbig [Sat, 13 Dec 1997 11:51:16 +0000 (11:51 +0000)]
Delete "typedef ... date" (see style(9)).
In the man page Use ".Pp" instead of blank lines, adopt English
and stress that the Julian->Gregorian switch took place at
different dates in different countries.
Suggested by: Garrett.

26 years agoAdd minigzip
Mike Smith [Sat, 13 Dec 1997 09:59:32 +0000 (09:59 +0000)]
Add minigzip

26 years agoThis framework builds the 'minigzip' sample program supplied with
Mike Smith [Sat, 13 Dec 1997 09:57:48 +0000 (09:57 +0000)]
This framework builds the 'minigzip' sample program supplied with
libz.  It is a non-GPL-polluted minimal implementation of the 'gzip'
command, yielding a 12k executable (vs ~100k for gzip).

26 years agoSubmitted by: Peter Hawkins <peter@clari.net.au>
Daniel O'Callaghan [Sat, 13 Dec 1997 06:34:47 +0000 (06:34 +0000)]
Submitted by: Peter Hawkins <peter@clari.net.au>
Spelling police.

26 years agoFix a segmentation violation that happened if "auth" was in the
John Polstra [Sat, 13 Dec 1997 05:27:29 +0000 (05:27 +0000)]
Fix a segmentation violation that happened if "auth" was in the
options file but there was no suitable secret in the secrets file.

26 years agoChange the ioctls for procfs around a bit; in particular, whever possible,
Sean Eric Fagan [Sat, 13 Dec 1997 03:13:49 +0000 (03:13 +0000)]
Change the ioctls for procfs around a bit; in particular, whever possible,
change from

ioctl(fd, PIOC<foo>, &i);

to

ioctl(fd, PIOC<foo>, i);

This is going from the _IOW to _IO ioctl macro.  The kernel, procctl, and
truss must be in synch for it all to work (not doing so will get errors about
inappropriate ioctl's, fortunately).  Hopefully I didn't forget anything :).

26 years agoAllow random IP number allocation to peer.
Brian Somers [Sat, 13 Dec 1997 02:37:33 +0000 (02:37 +0000)]
Allow random IP number allocation to peer.
Validate the peers suggested IP by attempting to make a routing table
entry.
Give up IPCP negotiation if the peer NAKs us with an unusable IP.
Always SIOCDIFADDR then SIOCAIFADDR when configuring the tun device.
Using SIOCSIFDSTADDR allows duplicate dst addresses (which we don't
want)!!!
Allow up to 200 interface names (was 50) (now that ppp can play server
properly).
Up the version number (1.5 -> 1.6).

Cosmetic:
  Log unexpected CCP packets in the CCP log rather than the ERROR log.
  Log unexpected Config Reqs in the appropriate LCP/IPCP/CCP log rather
  than the ERROR log.
  Log failed route additions and deletions with WARN, not TCPIP.
  Log the option id and length for unrecognised IPCP options.
  Change some .Sq to .Ar in the man page.

26 years agoAdd needed #include.
Tor Egge [Fri, 12 Dec 1997 21:45:23 +0000 (21:45 +0000)]
Add needed #include.

Problem found by: Bruce Evans <bde@zeta.org.au>

26 years agoMake code slightly more robust towards change by calculating size
Eivind Eklund [Fri, 12 Dec 1997 18:38:22 +0000 (18:38 +0000)]
Make code slightly more robust towards change by calculating size
in-place.

Obtained from:  OpenBSD ar_subs.c rev 1.7 by Tood Miller <millert@openbsd.org)

26 years agoMake this lint more cleanly
Eivind Eklund [Fri, 12 Dec 1997 18:24:17 +0000 (18:24 +0000)]
Make this lint more cleanly

Obtained from: OpenBSD file rev 1.4 by mickey

26 years agoMerge from OpenBSD:
Eivind Eklund [Fri, 12 Dec 1997 18:11:29 +0000 (18:11 +0000)]
Merge from OpenBSD:
>sprintf -> snprintf paranoia

Obtainted from: OpenBSD src/bin/stty/print.c v1.5
by Todd Miller <millert@openbsd.org>

26 years agoFixed printing of and comparison with d_typename[]. It is not always
Bruce Evans [Fri, 12 Dec 1997 18:01:57 +0000 (18:01 +0000)]
Fixed printing of and comparison with d_typename[].  It is not always
null terminated.

26 years agoMF22 (oops, I committed it to the wrong tree yesterday).
Bruce Evans [Fri, 12 Dec 1997 17:58:31 +0000 (17:58 +0000)]
MF22 (oops, I committed it to the wrong tree yesterday).

Changed lots of %ld's in format strings back to %d.  %ld was right
when daddr_t was long, but Lite2 changed daddr_t to int32_t which
is implemented as `int' on i386's.

26 years agoFix(?) some style consistancy breakage and do some other nit-picking on
Peter Wemm [Fri, 12 Dec 1997 14:14:44 +0000 (14:14 +0000)]
Fix(?) some style consistancy breakage and do some other nit-picking on
the SUIDDIR changes.

26 years agoI've been using these tweaks to enable the sound driver to talk to the
Peter Wemm [Fri, 12 Dec 1997 14:08:50 +0000 (14:08 +0000)]
I've been using these tweaks to enable the sound driver to talk to the
(mutant) Crystal CSS4236 chip on the Intel PR440FX SMP motherboard.

XXX this uses some rather ugly PnP bootstrap code that is *NOT* compatable
with 'controller pnp0' or *ANY* other PnP devices.  If you use some other
PnP devices, enabling css0 will burn your house down. :-]  The
"simplified" PnP init sequence directly blats your config(8) settings onto
the chip.  I'm pretty sure 'css0' will conflict with 'mss0', this whole
area desperately needs a cleanup.

I have been using the following with some success on the PR440FX:
controller   snd0
device css0  at isa? port 0x534 irq 5 drq 1 flags 0x08 vector adintr
device opl0  at isa? port 0x388
device mpu0  at isa? port 0x330 irq 10 vector mpuintr

26 years agoChange select(0, NULL, ...) to usleep()
Andrey A. Chernov [Fri, 12 Dec 1997 11:47:28 +0000 (11:47 +0000)]
Change select(0, NULL, ...) to usleep()

26 years agoEleven days were eliminated by the Gregorian Reformation in 1752, not ten.
Wolfgang Helbig [Fri, 12 Dec 1997 10:25:19 +0000 (10:25 +0000)]
Eleven days were eliminated by the Gregorian Reformation in 1752, not ten.
Submittee by:

26 years agoWe have had support for running the kernel daemons as threads for
John Dyson [Fri, 12 Dec 1997 04:00:59 +0000 (04:00 +0000)]
We have had support for running the kernel daemons as threads for
quite a while, but forgot to do so.  For now, this code supports
most daemons  running as kernel threads in UP kernels, and as
full processes in SMP.  We will soon be able to run them as
threads in SMP, but not yet.

26 years agoFix a problem with procfs_exit() that resulted in missing some procfs
Sean Eric Fagan [Fri, 12 Dec 1997 03:33:43 +0000 (03:33 +0000)]
Fix a problem with procfs_exit() that resulted in missing some procfs
nodes; this also apparantly caused a panic in some circumstances.
Also, since procfs_exit() is getting rid of the nodes when a process
exits, don't bother checking for the process' existance in procfs_inactive().

26 years agoFixed the usual multiplication overflow bug in an lseek() offset calculation.
Bruce Evans [Thu, 11 Dec 1997 07:27:06 +0000 (07:27 +0000)]
Fixed the usual multiplication overflow bug in an lseek() offset calculation.
The scan for bad sectors was broken for offsets >= 4G on 32-bit systems.

26 years agoMerged from Lite2 (fix misformattings in copyright).
Bruce Evans [Thu, 11 Dec 1997 07:12:10 +0000 (07:12 +0000)]
Merged from Lite2 (fix misformattings in copyright).

26 years agoPut [+format] at the end of the usage message.
Brian Somers [Thu, 11 Dec 1997 02:38:56 +0000 (02:38 +0000)]
Put [+format] at the end of the usage message.
Make `date -?' output (pretty much) the same as
the man page
PR: 5269

26 years agoFix the prototype for swapout_procs();
John Dyson [Thu, 11 Dec 1997 02:10:55 +0000 (02:10 +0000)]
Fix the prototype for swapout_procs();
Submitted by: dima@best.net

26 years agoRemove simultaneous include of <sys/param.h> and <sys/types.h>.
Eivind Eklund [Wed, 10 Dec 1997 22:18:54 +0000 (22:18 +0000)]
Remove simultaneous include of <sys/param.h> and <sys/types.h>.
Reorder includes to be alphabetical some places since I already was in
here.

26 years agoFix some style bugs.
Guido van Rooij [Wed, 10 Dec 1997 20:33:59 +0000 (20:33 +0000)]
Fix some style bugs.
Submitted by: bruce

26 years agoMerge from OpenBSD:
Eivind Eklund [Wed, 10 Dec 1997 17:52:49 +0000 (17:52 +0000)]
Merge from OpenBSD:
> Error out if someone tries to mv a mount point.  Old behavior was to
> move all files contained in the mounted filesystem to the dest. dir
> which could be quite nasty.  Personally, I think rename(2) should
> return EPERM or EINVAL instead of EXDEV.

Obtained from: OpenBSD mv.c rev 1.6 by Todd Miller <millert@openbsd.org>

26 years agoSync with sys/i386/conf/files.i386, majors.i386 and options.i386
KATO Takenori [Wed, 10 Dec 1997 09:28:59 +0000 (09:28 +0000)]
Sync with sys/i386/conf/files.i386, majors.i386 and options.i386
revisions 1.181, 1.23 and 1.64, respectively.

26 years agodocument some things that others and I have done to the tree... these
John-Mark Gurney [Wed, 10 Dec 1997 07:41:24 +0000 (07:41 +0000)]
document some things that others and I have done to the tree... these
include the PnP and Luigi's Sound code...

in the security section, talk about the f00f bug being fixed...

26 years agoQuiet some lint.
John Dyson [Wed, 10 Dec 1997 04:14:23 +0000 (04:14 +0000)]
Quiet some lint.

26 years agonatd 1_10 => 1_11
Brian Somers [Wed, 10 Dec 1997 02:14:57 +0000 (02:14 +0000)]
natd 1_10 => 1_11
  Cosmetic style changes
  Use u_short for port values.
Submitted by: Ari Suutari <ari@suutari.iki.fi>

26 years agoSync with original source: add FreeBSD 2.2.5, NetBSD 1.3, and OpenBSD 2.2
Wolfram Schneider [Tue, 9 Dec 1997 22:53:06 +0000 (22:53 +0000)]
Sync with original source: add FreeBSD 2.2.5, NetBSD 1.3, and OpenBSD 2.2

26 years agoAdd LIBZ
Wolfram Schneider [Tue, 9 Dec 1997 20:17:49 +0000 (20:17 +0000)]
Add LIBZ

26 years agoLog all failed mount attempts.
Guido van Rooij [Tue, 9 Dec 1997 18:43:44 +0000 (18:43 +0000)]
Log all failed mount attempts.
Also add a flag (-l) so mountd will also log all succeeded requests
to mountd.

26 years agoeliminated a previously unnoticde compile warning about use of
Jamil J. Weatherbee [Tue, 9 Dec 1997 12:41:13 +0000 (12:41 +0000)]
eliminated a previously unnoticde compile warning about use of
__inline, not good to use anyway ?!?!!

26 years agothe alog.c file respectively, added in wrong order by accident
Jamil J. Weatherbee [Tue, 9 Dec 1997 12:07:50 +0000 (12:07 +0000)]
the alog.c file respectively, added in wrong order by accident

26 years agoadd entry in LINT for alog driver
Jamil J. Weatherbee [Tue, 9 Dec 1997 12:04:49 +0000 (12:04 +0000)]
add entry in LINT for alog driver
added line to files.i386 to compile in alog.c optionally as a driver

26 years agoSync with sys/i386/isa/mse.c and syscons.c revisions 1.36 and 1.242,
KATO Takenori [Tue, 9 Dec 1997 11:58:02 +0000 (11:58 +0000)]
Sync with sys/i386/isa/mse.c and syscons.c revisions 1.36 and 1.242,
respectively.

26 years agoRemove obsolete psm driver options: PSM_ACCEL, PSM_EMULATION,
Kazutaka YOKOTA [Tue, 9 Dec 1997 11:56:19 +0000 (11:56 +0000)]
Remove obsolete psm driver options: PSM_ACCEL, PSM_EMULATION,
PSM_CHECKSYNC.

26 years agoSync with sys/i386/i386/userconfig.c revision 1.98.
KATO Takenori [Tue, 9 Dec 1997 11:55:25 +0000 (11:55 +0000)]
Sync with sys/i386/i386/userconfig.c revision 1.98.

26 years agoadding alogio.h to system include files
Jamil J. Weatherbee [Tue, 9 Dec 1997 11:43:04 +0000 (11:43 +0000)]
adding alogio.h to system include files

26 years agoexcuse me, adding alog.4 man page now forgot to commit previously
Jamil J. Weatherbee [Tue, 9 Dec 1997 11:38:02 +0000 (11:38 +0000)]
excuse me, adding alog.4 man page now forgot to commit previously

26 years agoAdded alog.4 man page to /usr/src/share/man/man4/man4.i386/ directory
Jamil J. Weatherbee [Tue, 9 Dec 1997 11:36:36 +0000 (11:36 +0000)]
Added alog.4 man page to /usr/src/share/man/man4/man4.i386/ directory
Altered Makefile in that directory so that make builds the alog.4 man page

26 years agoAdded major 86 for the alog driver.
Jamil J. Weatherbee [Tue, 9 Dec 1997 10:51:11 +0000 (10:51 +0000)]
Added major 86 for the alog driver.

26 years agoBack out last commit re dmesg.boot
Daniel O'Callaghan [Tue, 9 Dec 1997 10:06:49 +0000 (10:06 +0000)]
Back out last commit re dmesg.boot

26 years agoMake rc record boottime dmesg to /var/log/dmesg.boot, so the data
Daniel O'Callaghan [Tue, 9 Dec 1997 07:22:04 +0000 (07:22 +0000)]
Make rc record boottime dmesg to /var/log/dmesg.boot, so the data
is preserved after it scrolls out of the dmesg buffer.

26 years agoCode to prevent a panic caused by procfs_exit(). Note that i don't know
Sean Eric Fagan [Tue, 9 Dec 1997 05:03:41 +0000 (05:03 +0000)]
Code to prevent a panic caused by procfs_exit().  Note that i don't know
what is teh root cause -- but, sometimes, a procfs vnode in pfshead is
apparantly corrupt (or a UFS vnode instead).  Without this patch, I can
get it to panic by doing (in csh)

while (1)
ps auxwww
end

and it will panic when the PID's wrap.  With it, it does not panic.
Yes -- I know that this is NOT the right way to fix it.  But I haven't
been able to get it to panic yet (which confuses me).  I am going to
be looking into the vgone() code now, as that may be a part of it.

26 years agoThe improvements to clock statistics by Tor Egge
Steve Passe [Mon, 8 Dec 1997 23:00:24 +0000 (23:00 +0000)]
The improvements to clock statistics by Tor Egge
Wrappered and enabled by the define BETTER_CLOCK (on by default in smpyests.h)

Reviewed by: smp@csn.net
Submitted by: Tor Egge <Tor.Egge@idi.ntnu.no>

26 years agoThe improvements to clock statistics by Tor Egge
Steve Passe [Mon, 8 Dec 1997 22:59:39 +0000 (22:59 +0000)]
The improvements to clock statistics by Tor Egge
Wrappered and enabled by the define BETTER_CLOCK (on by default in smpyests.h)

apic_vector.s also contains a small change I (smp) made to eliminate
the double level INT problem.  It seems stable, but I haven't the tools
in place to prove it fixes the problem.

Reviewed by: smp@csn.net
Submitted by: Tor Egge <Tor.Egge@idi.ntnu.no>

26 years agoMention pftp as an alternative command to enable passive mode.
Ollivier Robert [Mon, 8 Dec 1997 22:09:44 +0000 (22:09 +0000)]
Mention pftp as an alternative command to enable passive mode.

26 years agoA couple of fixes from bruce: first of all, psignal is a void (stupid
Sean Eric Fagan [Mon, 8 Dec 1997 22:09:39 +0000 (22:09 +0000)]
A couple of fixes from bruce:  first of all, psignal is a void (stupid
me; unfortunately, also makes it hard ot check for errors); second, I had
managed to forget a change to PIOCSFL (it should be _IOW, not _IOR) I had
in my local copy, and Bruce called me on it.

Submitted by: bde

26 years agoThis is a hack. Decode IEEE 802.1Q VLAN tagging so that we can decode
Garrett Wollman [Mon, 8 Dec 1997 21:42:35 +0000 (21:42 +0000)]
This is a hack.  Decode IEEE 802.1Q VLAN tagging so that we can decode
tagged traffic according to the encapsulated protocol.  It needs in
addition modifications to the filter generator which would deal with
checking the ethertype and vlan header as required.

26 years agochange ``ingored'' to ``ignored'' in two places. I noticed this typo
Gary Jennejohn [Mon, 8 Dec 1997 21:36:25 +0000 (21:36 +0000)]
change ``ingored'' to ``ignored'' in two places. I noticed this typo
when I ran tconv to test APE (A Programmer's Editor).

26 years agoAdd -P option to cp(1) for local to local copying.
Wolfram Schneider [Mon, 8 Dec 1997 21:02:36 +0000 (21:02 +0000)]
Add  -P option to cp(1) for local to local copying.
Pointed out by: bde

26 years agoDeal with inflate() returning avail_in == avail_out == 0
Brian Somers [Mon, 8 Dec 1997 20:22:12 +0000 (20:22 +0000)]
Deal with inflate() returning avail_in == avail_out == 0
We must call inflate again in case there's any pending output
despite our input buffer being empty.  If the output buffer
is in fact already flushed, inflate() returns Z_BUF_ERROR.
There isn't really an error !

26 years agoCorrect usage of `add' and `delete'.
Brian Somers [Mon, 8 Dec 1997 20:09:10 +0000 (20:09 +0000)]
Correct usage of `add' and `delete'.

26 years agoRemoved the annoying "apic_ipi might be stuck" message.
Steve Passe [Mon, 8 Dec 1997 18:36:02 +0000 (18:36 +0000)]
Removed the annoying "apic_ipi might be stuck" message.
Added commentary about the real problem and what needs to be done.

26 years agoExplicitly state that the -t option is required only when the
Kazutaka YOKOTA [Mon, 8 Dec 1997 11:54:42 +0000 (11:54 +0000)]
Explicitly state that the -t option is required only when the
moused command is not able to detect the appropriate protocol for the
give mouse automatically.

Suggested by: sos

26 years agoadd process id to tmp files... this prevents two runs from stomping
John-Mark Gurney [Mon, 8 Dec 1997 09:00:47 +0000 (09:00 +0000)]
add process id to tmp files...  this prevents two runs from stomping
over each other's tmp files...  (usr.bin/truss uncovered this bug)

26 years agoUse full path in synopsis. Sort #includes. Use .Tn for NIS.
Philippe Charnier [Mon, 8 Dec 1997 07:49:56 +0000 (07:49 +0000)]
Use full path in synopsis. Sort #includes. Use .Tn for NIS.

26 years agoSync with diffs I found in kerberised versions: -Wall, no `;' in macros.
Philippe Charnier [Mon, 8 Dec 1997 07:46:53 +0000 (07:46 +0000)]
Sync with diffs I found in kerberised versions: -Wall, no `;' in macros.

26 years agoCorrect a path.
Philippe Charnier [Mon, 8 Dec 1997 07:44:20 +0000 (07:44 +0000)]
Correct a path.

26 years agoUse Pa for files and Ar for arguments.
Philippe Charnier [Mon, 8 Dec 1997 07:43:13 +0000 (07:43 +0000)]
Use Pa for files and Ar for arguments.

26 years agoMFC: no \n in syslog strings. Change -P to -p in flags. EOF -> -1. Use err(3).
Philippe Charnier [Mon, 8 Dec 1997 07:41:13 +0000 (07:41 +0000)]
MFC: no \n in syslog strings. Change -P to -p in flags. EOF -> -1. Use err(3).

26 years ago- Changed strcmp to strncmp for checking the CIS manufacturer strings,
Nate Williams [Mon, 8 Dec 1997 06:35:07 +0000 (06:35 +0000)]
- Changed strcmp to strncmp for checking the CIS manufacturer strings,
  since we only store CIS_MAXSTR data, and the users may stick the
  'entire' CIS string returned from the card in /etc/pccard.conf and
  cause the comparison to (bogusly) fail.

Submitted by: Brad Karp <karp@eecs.harvard.edu>

26 years agoCorrect prototypes to match POSIX. Correct return code for aio_cancel.
John Dyson [Mon, 8 Dec 1997 02:18:25 +0000 (02:18 +0000)]
Correct prototypes to match POSIX.   Correct return code for aio_cancel.
Submitted by: Alex Nash <nash@mcs.com>

26 years agoUse at_exit() to invoke procfs_exit() instead of calling it directly.
Sean Eric Fagan [Mon, 8 Dec 1997 01:06:36 +0000 (01:06 +0000)]
Use at_exit() to invoke procfs_exit() instead of calling it directly.
Note that an unload facility should be used to call rm_at_exit() (if
procfs is being loaded as an LKM and is subsequently removed), but it
was non-obvious how to do this in the VFS framework.

Reviewed by: Julian Elischer

26 years agoVarious of the ISP users have commented that the 1.41 version of the
John Dyson [Mon, 8 Dec 1997 00:59:08 +0000 (00:59 +0000)]
Various of the ISP users have commented that the 1.41 version of the
nfs_bio.c code worked better than the 1.44.  This commit reverts
the important parts of 1.44 to 1.41, and we will fix it when we
can get a handle on the problem.

26 years agoFix PAP, CHAP & LQR req (I broke the byte ordering when
Brian Somers [Sun, 7 Dec 1997 23:55:29 +0000 (23:55 +0000)]
Fix PAP, CHAP & LQR req (I broke the byte ordering when
I did the deflate re-org).
Make PAP & CHAP negotiation prettier in the log file.
If both PAP & CHAP are `enabled' and the peer NAKs CHAP
suggesting PAP, be friendly and REQ PAP the next time.
This is in line with the rfc.

26 years agoUse `cp -R' instead `cp -r' for local to local coping
Wolfram Schneider [Sun, 7 Dec 1997 20:49:39 +0000 (20:49 +0000)]
Use `cp -R' instead `cp -r' for local to local coping
(e.g.: rcp -r /tmp/1 /tmp/2). See the cp(1) manpage for more details.

26 years agooops, remove a dangling predicate left over after a sentence was rewritten.
Garrett Wollman [Sun, 7 Dec 1997 20:25:45 +0000 (20:25 +0000)]
oops, remove a dangling predicate left over after a sentence was rewritten.

26 years agoAdd some more macro advice and correct spelling of ``parentheses''.
Garrett Wollman [Sun, 7 Dec 1997 20:19:20 +0000 (20:19 +0000)]
Add some more macro advice and correct spelling of ``parentheses''.

26 years agoAdded some advice to avoid typedef'ing structures, as this breaks
Garrett Wollman [Sun, 7 Dec 1997 19:53:44 +0000 (19:53 +0000)]
Added some advice to avoid typedef'ing structures, as this breaks
information-hiding.  Also recommended against naming typedefs to end
in _t unless POSIX or ANSI requires it, and in favor of using queue(3)
macros to generate lists rather than rolling one's own.

26 years agoAdded easterog() and easteroj() which compute orthodox easter for
Wolfgang Helbig [Sun, 7 Dec 1997 19:04:14 +0000 (19:04 +0000)]
Added easterog() and easteroj() which compute orthodox easter for
Gregorian and Julian Calendar.
Suggested by: Andrey