]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
22 years agoRun find.1 through ispell.
Giorgos Keramidas [Mon, 1 Apr 2002 12:41:14 +0000 (12:41 +0000)]
Run find.1 through ispell.

PR: docs/36601
Submitted by: Joshua Goodall <joshua@roughtrade.net>
MFC after: 3 days

22 years agoThis commit was generated by cvs2svn to compensate for changes in r93526,
Dag-Erling Smørgrav [Mon, 1 Apr 2002 12:20:48 +0000 (12:20 +0000)]
This commit was generated by cvs2svn to compensate for changes in r93526,
which included commits to RCS files with non-trunk default branches.

22 years agoFix braino in code path for new items (perforce change 8760).
Dag-Erling Smørgrav [Mon, 1 Apr 2002 12:20:48 +0000 (12:20 +0000)]
Fix braino in code path for new items (perforce change 8760).
Based on a patch submitted by Joe Marcus Clarke <marcus@marcuscom.com>.

Sponsored by: DARPA, NAI LabS
cvs: ----------------------------------------------------------------------

22 years ago1) Use FBSDID for vendor string.
David Malone [Mon, 1 Apr 2002 11:06:00 +0000 (11:06 +0000)]
1) Use FBSDID for vendor string.
2) Change an int for a size_t, as that's what it is mainly used for.
3) Add some braces to keep gcc happy.

22 years agoDon't use gcc specific flags.
David Malone [Mon, 1 Apr 2002 10:54:11 +0000 (10:54 +0000)]
Don't use gcc specific flags.

22 years ago1) Clean up vendor and ID strings.
David Malone [Mon, 1 Apr 2002 10:49:11 +0000 (10:49 +0000)]
1) Clean up vendor and ID strings.
2) include stdlib.h for atoi.
3) staticise and constify.
4) add some missing prototypes.
5) add some parens to keep gcc happy.

22 years agoConst the code.
David Malone [Mon, 1 Apr 2002 10:45:09 +0000 (10:45 +0000)]
Const the code.

22 years agoFix some whitespace style bugs.
Mike Barcroft [Mon, 1 Apr 2002 09:45:05 +0000 (09:45 +0000)]
Fix some whitespace style bugs.

22 years agoFix SCM ID's.
David E. O'Brien [Mon, 1 Apr 2002 09:39:07 +0000 (09:39 +0000)]
Fix SCM ID's.

22 years agoRemove 'tech'. Repo-copied to 'freebsd'
Murray Stokely [Mon, 1 Apr 2002 09:08:24 +0000 (09:08 +0000)]
Remove 'tech'.  Repo-copied to 'freebsd'

Reminded by: joe

22 years agoAdd a few more tips, and tweak a couple.
Josef Karthauser [Mon, 1 Apr 2002 08:53:21 +0000 (08:53 +0000)]
Add a few more tips, and tweak a couple.

PR: misc/34039

22 years agoDocument recent DHCP upgrade.
Murray Stokely [Mon, 1 Apr 2002 08:44:51 +0000 (08:44 +0000)]
Document recent DHCP upgrade.

22 years agoAugment the BLOCKSIZE tip. (Based on the PR with slight changes by myself).
Josef Karthauser [Mon, 1 Apr 2002 08:43:58 +0000 (08:43 +0000)]
Augment the BLOCKSIZE tip.  (Based on the PR with slight changes by myself).

PR: misc/34039

22 years agoDuh, savecore is called savecore.
Poul-Henning Kamp [Mon, 1 Apr 2002 08:27:19 +0000 (08:27 +0000)]
Duh, savecore is called savecore.

22 years agoo Implement <sys/_types.h>, a new header for storing types that are
Mike Barcroft [Mon, 1 Apr 2002 08:12:25 +0000 (08:12 +0000)]
o Implement <sys/_types.h>, a new header for storing types that are
  MI, not required to be a fixed size, and used in multiple headers.
  This will grow in time, as more things move here from <sys/types.h>
  and <machine/ansi.h>.
o Add missing type definitions (uint16_t and uint32_t) to
  <arpa/inet.h> and <netinet/in.h>.
o Reduce pollution in <sys/types.h> by using `#if _FOO_T_DECLARED'
  widgets to avoid including <sys/stdint.h>.
o Add some missing type definitions to <unistd.h> and note the ones
  that still need to be added.
o Make use of <sys/_types.h> primitives in <grp.h> and <sys/types.h>.

Reviewed by: bde

22 years agoResolve conflicts.
Murray Stokely [Mon, 1 Apr 2002 08:09:50 +0000 (08:09 +0000)]
Resolve conflicts.

22 years agoResolve conflicts:
Murray Stokely [Mon, 1 Apr 2002 08:02:18 +0000 (08:02 +0000)]
Resolve conflicts:
  * $FreeBSD$
  * connect std{in,out,err} (r1.19).
  * prefer strlcpy to strpy (r1.13).
  * quiet by default (r1.6).
  * document -D option.

22 years agoResolve conflicts:
Murray Stokely [Mon, 1 Apr 2002 07:59:50 +0000 (07:59 +0000)]
Resolve conflicts:
  * $FreeBSD$ line.
  * Prefer strlcpy to strcpy (r1.6).
  * dhclient-script pathname.

22 years agoImplement a fine-grain control system which allows header developers
Mike Barcroft [Mon, 1 Apr 2002 07:58:26 +0000 (07:58 +0000)]
Implement a fine-grain control system which allows header developers
to control the exposure of macros and prototypes depending upon the
POSIX, X/Open, or ISO C version an application has requested.

Submitted by: wollman
Reviewed by: bde, imp

22 years agoThis commit was generated by cvs2svn to compensate for changes in r93507,
Murray Stokely [Mon, 1 Apr 2002 07:54:44 +0000 (07:54 +0000)]
This commit was generated by cvs2svn to compensate for changes in r93507,
which included commits to RCS files with non-trunk default branches.

22 years agoImport ISC DHCP 3.0.1 RC8 client.
Murray Stokely [Mon, 1 Apr 2002 07:54:44 +0000 (07:54 +0000)]
Import ISC DHCP 3.0.1 RC8 client.

22 years agoRemove erronious destroy_dev. Not sure what the author's intent was here
Alfred Perlstein [Mon, 1 Apr 2002 06:09:17 +0000 (06:09 +0000)]
Remove erronious destroy_dev.  Not sure what the author's intent was here
but this makes the driver not panic my -current box.

Approved by: scottl

22 years agoFix 3 of the four problems with my last indentation fix. ("fixing" the
Kenneth D. Merry [Mon, 1 Apr 2002 05:41:33 +0000 (05:41 +0000)]
Fix 3 of the four problems with my last indentation fix.  ("fixing" the
fourth would be a divergence from the prevailing style.)

Thanks to bde for catching this.

Pointed out by: bde

22 years agoktrdump is a utility to dump the ktr trace buffer from userland. It can
Jake Burkholder [Mon, 1 Apr 2002 05:41:13 +0000 (05:41 +0000)]
ktrdump is a utility to dump the ktr trace buffer from userland.  It can
also be run on a core dump.

22 years agoktr changes to improve performance and make writing a userland utility to
Jake Burkholder [Mon, 1 Apr 2002 05:35:26 +0000 (05:35 +0000)]
ktr changes to improve performance and make writing a userland utility to
dump the trace buffer feasible.
- Remove KTR_EXTEND.  This changes the format of the trace entries when
  activated, making writing a userland tool which is not tied to a specific
  kernel configuration difficult.
- Use get_cyclecount() for timestamps.  nanotime() is much too heavy weight
  and requires recursion protection due to ktr traces occuring as a result
  of ktr traces.  KTR_VERBOSE may still require recursion protection, which
  is now conditional on it.
- Allow KTR_CPU to be overridden by MD code.  This is so that it is possible
  to trace early in startup before pcpu and/or curthread are setup.
- Add a version number for the ktr interface.  A userland tool can check this
  to detect mismatches.
- Use an array for the parameters to make decoding in userland easier.
- Add file and line recording to the non-extended traces now that the extended
  version is no more.

These changes will break gdb macros to decode the extended version of the
trace buffer which are floating around.  Users of these macros should either
use the show ktr command in ddb, or use the userland utility which can be run
on a core dump.

Approved by: jhb
Tested on: i386, sparc64

22 years agoFix an indentation problem.
Kenneth D. Merry [Mon, 1 Apr 2002 03:59:47 +0000 (03:59 +0000)]
Fix an indentation problem.

22 years agoFix a bzero length calculation. sizeof(struct ccb_getdev) should have been
Kenneth D. Merry [Mon, 1 Apr 2002 03:58:51 +0000 (03:58 +0000)]
Fix a bzero length calculation.  sizeof(struct ccb_getdev) should have been
sizeof(struct ccb_getdevstats).

MFC after: 3 days

22 years agoRemove reference to share/dict/special/. That directory does not
Murray Stokely [Mon, 1 Apr 2002 01:43:34 +0000 (01:43 +0000)]
Remove reference to share/dict/special/.  That directory does not
exist, nor does the referenced program.

MFC after: 3 days

22 years agoAdd a note about the FreeBSD technical word list.
Murray Stokely [Mon, 1 Apr 2002 00:57:34 +0000 (00:57 +0000)]
Add a note about the FreeBSD technical word list.

22 years agoThe freebsd jargon dictionary was repo-copied from 'tech' to 'freebsd'
Murray Stokely [Mon, 1 Apr 2002 00:51:34 +0000 (00:51 +0000)]
The freebsd jargon dictionary was repo-copied from 'tech' to 'freebsd'
to indicate that this is not a general dictionary of all technical
terms.

Requested by: sheldon

22 years agoWork around a gcc bug which causes zlib to dump core in certain
Kris Kennaway [Sun, 31 Mar 2002 23:47:52 +0000 (23:47 +0000)]
Work around a gcc bug which causes zlib to dump core in certain
situations.  This can be removed once gcc is fixed.

Submitted by: bde

22 years agoHere follows the new kernel dumping infrastructure.
Poul-Henning Kamp [Sun, 31 Mar 2002 22:37:00 +0000 (22:37 +0000)]
Here follows the new kernel dumping infrastructure.

Caveats:

The new savecore program is not complete in the sense that it emulates
enough of the old savecores features to do the job, but implements none
of the options yet.

I would appreciate if a userland hacker could help me out getting savecore
to do what we want it to do from a users point of view, compression,
email-notification, space reservation etc etc.  (send me email if
you are interested).

Currently, savecore will scan all devices marked as "swap" or "dump" in
/etc/fstab _or_ any devices specified on the command-line.

All architectures but i386 lack an implementation of dumpsys(), but
looking at the i386 version it should be trivial for anybody familiar
with the platform(s) to provide this function.

Documentation is quite sparse at this time, more to come.

Details:

ATA and SCSI drivers should work as the dump formatting code has been
removed.  The IDA, TWE and AAC have not yet been converted.

Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set
the device as dumpdev.  To implement the "off" argument, /dev/null
is used as the device.

Savecore will fail if handed any options since they are not (yet)
implemented.  All devices marked "dump" or "swap" in /etc/fstab
will be scanned and dumps found will be saved to diskfiles
named from the MD5 hash of the header record.  The header record
is dumped in readable format in the .info file.  The kernel
is not saved.  Only complete dumps will be saved.

All maintainer rights for this code are disclaimed: feel free to
improve and extend.

Sponsored by:   DARPA, NAI Labs

22 years agoThe AAC, TWE and IDA diskdrivers cannot dump until I and msmith
Poul-Henning Kamp [Sun, 31 Mar 2002 22:29:52 +0000 (22:29 +0000)]
The AAC, TWE and IDA diskdrivers cannot dump until I and msmith
have ripped all the i386 specific formatting code from their
dump routines.  Due to the potential for trashing disks, I did
not want to do this "blind".

22 years agoAdd the i386dump.c dumpsys() source file.
Poul-Henning Kamp [Sun, 31 Mar 2002 22:28:28 +0000 (22:28 +0000)]
Add the i386dump.c dumpsys() source file.

22 years agoDA (scsi) and AD (ata) diskdrivers:
Poul-Henning Kamp [Sun, 31 Mar 2002 22:28:03 +0000 (22:28 +0000)]
DA (scsi) and AD (ata) diskdrivers:
Make the dump routine do just writing of data.

22 years agoSavecore:
Poul-Henning Kamp [Sun, 31 Mar 2002 22:26:56 +0000 (22:26 +0000)]
Savecore:

Commandline compatible with the previous savecore unless you specify
any options, none of them are implemented (yet).

Scans all devices marked "dump" or "swap" for dump header signatures
and saves dumps off under a name which is a MD5 hash of the header
information.  This should give unique filenames.  A *.info file contains
ascii version of the header information.

22 years agoHere follows the new kernel dumping infrastructure.
Poul-Henning Kamp [Sun, 31 Mar 2002 22:24:24 +0000 (22:24 +0000)]
Here follows the new kernel dumping infrastructure.

Caveats:

The new savecore program is not complete in the sense that it emulates
enough of the old savecores features to do the job, but implements none
of the options yet.

I would appreciate if a userland hacker could help me out getting savecore
to do what we want it to do from a users point of view, compression,
email-notification, space reservation etc etc.  (send me email if
you are interested).

Currently, savecore will scan all devices marked as "swap" or "dump" in
/etc/fstab _or_ any devices specified on the command-line.

All architectures but i386 lack an implementation of dumpsys(), but
looking at the i386 version it should be trivial for anybody familiar
with the platform(s) to provide this function.

Documentation is quite sparse at this time, more to come.

Sponsored by:   DARPA, NAI Labs

Details:

Dumpon now opens the device and uses ioctl(DIOCGKERNELDUMP) to set it
to be the dumpdevice.  When "off" is set, /dev/null is used.

22 years agoAdd a new #include which describes the common header format for kerneldumps.
Poul-Henning Kamp [Sun, 31 Mar 2002 21:48:08 +0000 (21:48 +0000)]
Add a new #include which describes the common header format for kerneldumps.

This design is my best effort and it is quite likely that people more used
to kernel dumps may want to change this subsequently so two levels of
version numbers are provided: one for the common header and one per
architecture.

Sponsored by: DARPA & NAI Labs.

22 years agoRegen.
Josef Karthauser [Sun, 31 Mar 2002 21:38:38 +0000 (21:38 +0000)]
Regen.

22 years agoAdd more usb adapters, from NetBSD.
Josef Karthauser [Sun, 31 Mar 2002 21:38:05 +0000 (21:38 +0000)]
Add more usb adapters, from NetBSD.

22 years agoImplement the two "GEOM" ioctls DIOCGSECTORSIZE and DIOCGMEDIASIZE for
Poul-Henning Kamp [Sun, 31 Mar 2002 21:17:12 +0000 (21:17 +0000)]
Implement the two "GEOM" ioctls DIOCGSECTORSIZE and DIOCGMEDIASIZE for
the non-GEOM code as well.  This simplifies the the kernel-dumping
and disk-management tools as less compatibility cruft will be needed.

Sponsored by: DARPA and NAI Labs.

22 years agoVarious mdoc fixes, including a change that corrects spacing as
Giorgos Keramidas [Sun, 31 Mar 2002 20:49:37 +0000 (20:49 +0000)]
Various mdoc fixes, including a change that corrects spacing as
described in PR docs/36461.

PR: docs/36461
Noticed by: Gary W. Swearingen <swear@blarg.net>
Submitted by: ru

22 years agoOnly install the help file if we can find it. Use ${BASE}.help
Marcel Moolenaar [Sun, 31 Mar 2002 20:48:13 +0000 (20:48 +0000)]
Only install the help file if we can find it. Use ${BASE}.help
in both the condition and for the install. We expect to find
the help file in ${.OBJDIR}.

22 years agoUncomment now implemented part of -d description
Andrey A. Chernov [Sun, 31 Mar 2002 20:18:35 +0000 (20:18 +0000)]
Uncomment now implemented part of -d description

22 years agoKeep the reference to the file acquired in _aio_aqueue() until the operation
Alan Cox [Sun, 31 Mar 2002 20:17:56 +0000 (20:17 +0000)]
Keep the reference to the file acquired in _aio_aqueue() until the operation
completes.  The reference is released in aio_free_entry().

Submitted by: tegge

22 years agoRemove \n from __COPYRIGHT to shut assembler warning
Andrey A. Chernov [Sun, 31 Mar 2002 20:11:10 +0000 (20:11 +0000)]
Remove \n from __COPYRIGHT to shut assembler warning

22 years agoAdd P1003.1-2001 -f and -p options.
Juli Mallett [Sun, 31 Mar 2002 18:44:36 +0000 (18:44 +0000)]
Add P1003.1-2001 -f and -p options.

PR: standards/36243
Submitted by: Tim J. Robbins <tim@robbins.dropbear.id.au>
Reviewed by: mike
MFC after: 2 weeks

22 years agoMake it easier to specify a custom kernel configuration for the boot
Murray Stokely [Sun, 31 Mar 2002 17:21:29 +0000 (17:21 +0000)]
Make it easier to specify a custom kernel configuration for the boot
floppies.

PR: misc/36032
Submitted by: David Yeske <dyeske@yahoo.com>
MFC after: 1 week

22 years agoATA100 is allowed on the HPT chips rev >= 3
Søren Schmidt [Sun, 31 Mar 2002 13:33:55 +0000 (13:33 +0000)]
ATA100 is allowed on the HPT chips rev >= 3

22 years agoClose some holes with p->p_args by NULL'ing out the p->p_args pointer
Alfred Perlstein [Sun, 31 Mar 2002 10:33:12 +0000 (10:33 +0000)]
Close some holes with p->p_args by NULL'ing out the p->p_args pointer
while holding the proc lock, and by holding the pargs structure when
accessing it from outside of the owner.

Submitted by: Jonathan Mini <mini@haikugeek.com>

22 years agoSupport more than 32 sio unit numbers. The maximum unit number is now
Bruce Evans [Sun, 31 Mar 2002 09:15:43 +0000 (09:15 +0000)]
Support more than 32 sio unit numbers.  The maximum unit number is now
(65536 * 32 - 1), but MAKEDEV only supports up to (32 * 32 -1).  Device
names use the unit number in base 32 for all "digits".

This required fixing an old bug in MAKEDEV:ttyminor().  Its arg was the
global $unit instead of $1.

Reminded by: Valentin K. Ponomarenko <valka@krog.ukrtel.net>
MFC-after: 1 week

22 years agoMFi386: revision 1.508.
KATO Takenori [Sun, 31 Mar 2002 08:00:35 +0000 (08:00 +0000)]
MFi386: revision 1.508.

22 years agoA couple of bits survived Dans nukage of CV_DEBUG in favour of INVARIANTS,
Poul-Henning Kamp [Sun, 31 Mar 2002 07:23:31 +0000 (07:23 +0000)]
A couple of bits survived Dans nukage of CV_DEBUG in favour of INVARIANTS,
take them out with tacticals.

22 years agoCentralize the "bootdev" and "dumpdev" variables. They are still pretty
Poul-Henning Kamp [Sun, 31 Mar 2002 07:15:28 +0000 (07:15 +0000)]
Centralize the "bootdev" and "dumpdev" variables.  They are still pretty
bogus all things considered, but at least now they don't camouflage as
being MD variables.

22 years agoHacks for measuring interrupt latency. Interrupt latency can be
Bruce Evans [Sun, 31 Mar 2002 06:49:38 +0000 (06:49 +0000)]
Hacks for measuring interrupt latency.  Interrupt latency can be
measured accurately for periodic interrupts provided the interrupts
don't need to be serviced very quickly to keep their period almost
constant.  sio output interrupts have this property (interrupt service
can be delayed for up to 1 character time without the period changing).

This is non-optional and undocumented so that it can be added and
removed easily.  It has no significant effect unless it is enabled by
hacking on a variable using a debugger.  Hardclock and statclock interrupts
would work even better for this, at least on i386's, provided their
interrupt handlers are fast (as they are in -current but not in -stable
or in my version of -current).

22 years agoFix a typo
Brian Somers [Sun, 31 Mar 2002 01:57:06 +0000 (01:57 +0000)]
Fix a typo

22 years agoAdd a missing ``#ifndef NONETGRAPH''
Brian Somers [Sun, 31 Mar 2002 01:56:34 +0000 (01:56 +0000)]
Add a missing ``#ifndef NONETGRAPH''

22 years agoRemove some extraneous whitespace
Brian Somers [Sun, 31 Mar 2002 01:51:15 +0000 (01:51 +0000)]
Remove some extraneous whitespace

22 years agoInclude arpa/inet.h
Brian Somers [Sun, 31 Mar 2002 01:36:08 +0000 (01:36 +0000)]
Include arpa/inet.h

22 years agoImplement i386's (o)sigreturn() like the alpha's: Use copyin() to read
Alan Cox [Sun, 31 Mar 2002 01:13:21 +0000 (01:13 +0000)]
Implement i386's (o)sigreturn() like the alpha's: Use copyin() to read
the osigcontext or ucontext_t rather than useracc() followed by direct user-
space memory accesses.  This reduces (o)sigreturn()'s execution time by 5-
50%.

Submitted by: bde

22 years agoAdd a local proc *p in exec_new_vmspace() to avoid repeated dereferencing
Alan Cox [Sun, 31 Mar 2002 00:05:30 +0000 (00:05 +0000)]
Add a local proc *p in exec_new_vmspace() to avoid repeated dereferencing
to obtain it.

22 years agoPass the address of the bootinfo block to the kernel in register
Marcel Moolenaar [Sat, 30 Mar 2002 23:52:34 +0000 (23:52 +0000)]
Pass the address of the bootinfo block to the kernel in register
r8. Keep it at the hardwired address for now. Bump the version.

22 years agoTransition to a model where the loader passes the address of the
Marcel Moolenaar [Sat, 30 Mar 2002 23:25:22 +0000 (23:25 +0000)]
Transition to a model where the loader passes the address of the
bootinfo block in register r8. In locore.s we save the address
in the global variable 'pa_bootinfo'. In machdep.c we compare
this value against the hardwired address, but don't depend on its
validity yet (ie: we still expect the bootinfo block to be at the
hardwired address). After a small amount of time, we'll flip the
switch and depend on the loader to pass us the address. From that
moment on the loader is free to put it anywhere it likes, provided
the machine itself likes it as well.

Add some verbosity to aid in the transition. We emit a message if
the loader didn't pass the address and we also emit a message if
there's no bootinfo block at the hardwired address.

While in locore.s, reduce the number of redundant serialization
instructions. A srlz.i is a proper superset of a srlz.d and thus
is a valid replacement. Also slightly reorder the movl instructions
to improve bundle density.

22 years agoFix braino: the example interface is fxp0, not ed0.
Dima Dorfman [Sat, 30 Mar 2002 23:23:41 +0000 (23:23 +0000)]
Fix braino: the example interface is fxp0, not ed0.

PR: 36471
Submitted by: John Nielsen <zoobie@bsdconspiracy.net>

22 years agoCorrect typo.
Dima Dorfman [Sat, 30 Mar 2002 23:19:18 +0000 (23:19 +0000)]
Correct typo.

PR: 36544
Submitted by: dak@klemm.delta6.net

22 years agoPass the physical address of the bootinfo block to the kernel in
Marcel Moolenaar [Sat, 30 Mar 2002 23:00:05 +0000 (23:00 +0000)]
Pass the physical address of the bootinfo block to the kernel in
register r8. We continue to write the bootinfo block at the same
hardwired address, because the kernel still expects it there.
It is expected that future kernels use register r8 to get to the
bootinfo block and don't depend on the hardwired address anymore.

Bump the loader version once again due to the interface change.

22 years agoCorrect a comment: sendsig() calls the MI vm_map_growstack() but
Alan Cox [Sat, 30 Mar 2002 20:58:08 +0000 (20:58 +0000)]
Correct a comment: sendsig() calls the MI vm_map_growstack() but
 the corresponding comment refers to a MD grow_stack() that doesn't exist.

22 years agoUse the MI vm_map_growstack() instead of the MD grow_stack() in trap(). Remove
Alan Cox [Sat, 30 Mar 2002 20:44:31 +0000 (20:44 +0000)]
Use the MI vm_map_growstack() instead of the MD grow_stack() in trap().  Remove
the MD grow_stack().

22 years agoMove the "dumping" variable from systm.h to conf.h.
Poul-Henning Kamp [Sat, 30 Mar 2002 19:58:31 +0000 (19:58 +0000)]
Move the "dumping" variable from systm.h to conf.h.

22 years agosendmail 8.12 doesn't require _FFR_UNSAFE_SASL anymore.
Hajimu UMEMOTO [Sat, 30 Mar 2002 18:14:37 +0000 (18:14 +0000)]
sendmail 8.12 doesn't require _FFR_UNSAFE_SASL anymore.
This is standard feature.

Submitted by: ru

22 years agoFix CLEANFILES.
Ruslan Ermilov [Sat, 30 Mar 2002 17:57:51 +0000 (17:57 +0000)]
Fix CLEANFILES.

22 years agosendmail 8.12 claims against GroupReadableSASLFile.
Hajimu UMEMOTO [Sat, 30 Mar 2002 17:53:10 +0000 (17:53 +0000)]
sendmail 8.12 claims against GroupReadableSASLFile.

Submitted by: mistral@imasy.or.jp

22 years agoDon't nowerror for sys/dev/pdq/pdq.c
Matthew N. Dodd [Sat, 30 Mar 2002 17:41:23 +0000 (17:41 +0000)]
Don't nowerror for sys/dev/pdq/pdq.c

22 years agoRestore the correct CLEANFILES that got damaged in rev. 1.15.
Ruslan Ermilov [Sat, 30 Mar 2002 17:37:43 +0000 (17:37 +0000)]
Restore the correct CLEANFILES that got damaged in rev. 1.15.

22 years agoMention the UltraSparc as supported in the README, also deal with the
Bruce A. Mah [Sat, 30 Mar 2002 17:34:31 +0000 (17:34 +0000)]
Mention the UltraSparc as supported in the README, also deal with the
fact that (now) not every supported platform has sysinstall(8).

22 years agoDon't use gcc specific flags.
David Malone [Sat, 30 Mar 2002 17:19:40 +0000 (17:19 +0000)]
Don't use gcc specific flags.

22 years ago1) Staticise.
David Malone [Sat, 30 Mar 2002 17:17:26 +0000 (17:17 +0000)]
1) Staticise.
2) Remove registers.
3) Change some ints to size_t which are used with fread/fwrite.

22 years ago1) Const enough things to avoid warnings.
David Malone [Sat, 30 Mar 2002 17:10:20 +0000 (17:10 +0000)]
1) Const enough things to avoid warnings.
2) Cast ifdef_level to a size_t before comparing it to a ratio of size_ts.
   Ifdef_level should always be positive.
3) Complete prototype for chfont.
4) Cast some ptrdiff_ts to ints before using as a field width.
5) Avoid shadowing a local variable p with another local variable p.

22 years agoFix /dev/acdNc minor device number. This bug prevents installing FreeBSD
Makoto Matsushita [Sat, 30 Mar 2002 16:59:06 +0000 (16:59 +0000)]
Fix /dev/acdNc minor device number.  This bug prevents installing FreeBSD
from CD-ROM in 4-stable.  Note that in 5-current, we use devfs so this
change (hopefully) shouldn't change anything.

I'll MFC to 4-stable later.

Tested with: FreeBSD/i386, 4.5-STABLE-20020330-JPSNAP

22 years agoRemove debug output in last commit.
Søren Schmidt [Sat, 30 Mar 2002 16:51:47 +0000 (16:51 +0000)]
Remove debug output in last commit.

22 years agoAdd a "special progname lib xxx ..." command to crunchgen
Luigi Rizzo [Sat, 30 Mar 2002 16:48:30 +0000 (16:48 +0000)]
Add a "special progname lib xxx ..." command to crunchgen
so the .lo files can be partially linked against libraries
which redefine symbols in the standard libs, or which reference
symbols in the objects.

Submitted by: Sam Leffler
MFC After: 3 days

22 years ago1) Add missing prototypes for fmt_perm and cvt_time.
David Malone [Sat, 30 Mar 2002 16:46:04 +0000 (16:46 +0000)]
1) Add missing prototypes for fmt_perm and cvt_time.
2) Move a break outside a #if block to keep gcc3 from seeing a "default:"
   at the end of a block.
3) Fix some format warnings. Some remain which can be fixed more easily
   when we have a full C99 printf.

22 years agoAdd support for getting status (fan, temp, 5V and 12V levels) from
Søren Schmidt [Sat, 30 Mar 2002 16:36:41 +0000 (16:36 +0000)]
Add support for getting status (fan, temp, 5V and 12V levels) from
Promise Superswap enclosures.

Sponsored by: Advanis

22 years agoUse the method described in the strtol man page to check if it parsed
David Malone [Sat, 30 Mar 2002 16:24:03 +0000 (16:24 +0000)]
Use the method described in the strtol man page to check if it parsed
the entire string.  This avoids signed/unsigned comparison.

22 years agoFixed handling of short reads in readdisklabel() and writedisklabel().
Bruce Evans [Sat, 30 Mar 2002 16:02:43 +0000 (16:02 +0000)]
Fixed handling of short reads in readdisklabel() and writedisklabel().
These functions use DEV_STRATEGY() which can easily return a short
count (with no error) for reads near EOF.  EOF happens for "disks" too
small to contain a label sector (mainly for empty slices).  The functions
didn't understand this at all, and looked for labels in the garbage
in the buffer beyond what DEV_STRATEGY() returned.  The recent UMA
changes combined with my local changes and configuration resulted in
the garbage often containing a valid but garbage label left over from
a previous call.

Bugs in EOF handling in -current limited the problem to "disks" with
size precisely LABELSECTOR sectors.  LABELSECTOR happens to be a very
unusual "disk" size since it is only 0 for non-i386 arches that don't
usually have disks with DOS MBRs.

22 years agoIn ffs_mountffs(), set mnt_iosize_max to si_iosize_max unconditionally
Bruce Evans [Sat, 30 Mar 2002 15:12:57 +0000 (15:12 +0000)]
In ffs_mountffs(), set mnt_iosize_max to si_iosize_max unconditionally
provided the latter is nonzero.  At this point, the former is a fairly
arbitrary default value (DFTPHYS), so changing it to any reasonable
value specified by the device driver is safe.  Using the maximum of
these limits broke ffs clustered i/o for devices whose si_iosize_max
is < DFLTPHYS.  Using the minimum would break device drivers' ability
to increase the active limit from DFTLPHYS up to MAXPHYS.

Copied the code for this and the associated (unnecessary?) fixup of
mp_iosize_max to all other filesystems that use clustering (ext2fs and
msdosfs).  It was completely missing.

PR: 36309
MFC-after: 1 week

22 years agoFix vendor ID.
David Malone [Sat, 30 Mar 2002 14:32:07 +0000 (14:32 +0000)]
Fix vendor ID.
Make usage message match the man page. (Missing -o, -h doesn't depend on -d).

PR: 36470
Submitted by: Gary W. Swearingen <swear@blarg.net>
MFC after: 1 week

22 years agoClean up vendor ID and FBSDID.
David Malone [Sat, 30 Mar 2002 14:18:15 +0000 (14:18 +0000)]
Clean up vendor ID and FBSDID.
We now seem to include <arpa/inet.h> to get ntoh*.

22 years agoFix constness warnings.
David Malone [Sat, 30 Mar 2002 14:03:53 +0000 (14:03 +0000)]
Fix constness warnings.
Remove register keyword.
Don't initialise "badtype" in declaration - it was initialised below anyway.
Remove prototype for strcpy.

22 years agoDon't use gcc specific flags.
David Malone [Sat, 30 Mar 2002 14:01:55 +0000 (14:01 +0000)]
Don't use gcc specific flags.

22 years agoChange a "/*" within a comment to a "**".
David Malone [Sat, 30 Mar 2002 13:54:53 +0000 (13:54 +0000)]
Change a "/*" within a comment to a "**".
Add a missing include spotted by gcc30.

22 years agoMark some parameters as unused.
David Malone [Sat, 30 Mar 2002 13:51:40 +0000 (13:51 +0000)]
Mark some parameters as unused.
Make more code depend on '#ifndef NO_X' so we don't get unused warnings
when compiling without X.

22 years agoMinor post warns stuff.
David Malone [Sat, 30 Mar 2002 13:44:45 +0000 (13:44 +0000)]
Minor post warns stuff.
1) Fix up vendor IDs for files I'm touching and put FBSDID in the right place.
2) Cast return value of snprintf to size_t rather than casting sizeof() to
   an int, 'cos we know snprintf returns a non-genative number.
3) Avoid 'char *blank = "";' by just strduping. This could be bad sometimes,
   but all these variables usually point at malloced memory so this makes
   them always point at malloced memory.

22 years agoAdd a sprinkling of mp_Down() calls to ensure that we don't think that
Brian Somers [Sat, 30 Mar 2002 12:52:55 +0000 (12:52 +0000)]
Add a sprinkling of mp_Down() calls to ensure that we don't think that
the multi-link NCP is up when it isn't.

22 years agoDon't use the RTF_GATEWAY flag and pass the RTA_GATEWAY address when updating
Brian Somers [Sat, 30 Mar 2002 12:50:49 +0000 (12:50 +0000)]
Don't use the RTF_GATEWAY flag and pass the RTA_GATEWAY address when updating
routes.

22 years agoAdd a missing break
Brian Somers [Sat, 30 Mar 2002 12:47:29 +0000 (12:47 +0000)]
Add a missing break

22 years agoMerge the NETGRAPH branch into HEAD. tty devices now use netgraph's line
Brian Somers [Sat, 30 Mar 2002 12:30:09 +0000 (12:30 +0000)]
Merge the NETGRAPH branch into HEAD.  tty devices now use netgraph's line
discipline to do the async escaping, but no other benefits are available yet.

Change ``ifdef HAVE_DES'' to ``ifndef NODES'' for consistency.

Make the Makefile a little more sane WRT RELEASE_CRUNCH.

22 years agoFix a few typos.
Giorgos Keramidas [Sat, 30 Mar 2002 11:47:18 +0000 (11:47 +0000)]
Fix a few typos.
as a user ID -> has a user ID
command constitutes of -> command consists of

PR: misc/36523
Submitted by: Chris Pepper <pepper@mail.rockefeller.edu>
MFC after: 3 days

22 years agoMFi386: revision 1.55
Yoshihiro Takahashi [Sat, 30 Mar 2002 11:18:30 +0000 (11:18 +0000)]
MFi386: revision 1.55

22 years agoUse the raid lun not the magic when writing Promise config.
Søren Schmidt [Sat, 30 Mar 2002 11:15:46 +0000 (11:15 +0000)]
Use the raid lun not the magic when writing Promise config.