]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
21 years agoCompletely redo thread states.
Julian Elischer [Wed, 11 Sep 2002 08:13:56 +0000 (08:13 +0000)]
Completely redo thread states.

Reviewed by: davidxu@freebsd.org

21 years agoInclude <vm/uma.h> instead of depending on namespace pollution in
Bruce Evans [Wed, 11 Sep 2002 07:13:28 +0000 (07:13 +0000)]
Include <vm/uma.h> instead of depending on namespace pollution in
<sys/malloc.h>.

Sorted includes as much as possible.  Removed banal comment(s) attached to
includes.

21 years agoFix LINT build on alpha by completing move of cy and apm_saver to
Nate Lawson [Wed, 11 Sep 2002 05:33:15 +0000 (05:33 +0000)]
Fix LINT build on alpha by completing move of cy and apm_saver to
i386/conf/NOTES rather than the global conf/NOTES.

Suggested by: bde

21 years agoAdd `restrict' type-qualifier.
Mike Barcroft [Wed, 11 Sep 2002 05:05:48 +0000 (05:05 +0000)]
Add `restrict' type-qualifier.

21 years agoConditionalize some objects to match the functions that they are used
Mike Barcroft [Wed, 11 Sep 2002 05:04:06 +0000 (05:04 +0000)]
Conditionalize some objects to match the functions that they are used
with.  Enable `restrict' type-qualifier.

21 years agoAdd support for the AMD x86-64 Hammer platform.
David E. O'Brien [Wed, 11 Sep 2002 04:57:09 +0000 (04:57 +0000)]
Add support for the AMD x86-64 Hammer platform.

21 years agoSome uses of the variable n needed to be int, others needed to be
Mike Barcroft [Wed, 11 Sep 2002 04:26:09 +0000 (04:26 +0000)]
Some uses of the variable n needed to be int, others needed to be
size_t.  Add a new variable to cope.

21 years agoClarify the return value from child_present.
Warner Losh [Wed, 11 Sep 2002 04:22:10 +0000 (04:22 +0000)]
Clarify the return value from child_present.

21 years ago- Lock down the accounting code globals with a subsystem mutex.
Andrew R. Reiter [Wed, 11 Sep 2002 04:10:41 +0000 (04:10 +0000)]
- Lock down the accounting code globals with a subsystem mutex.

Reviewed by: jhb, mdodd

21 years agoQuiet warnings about non-existent scripts. My own fault for not testing my
Gordon Tetlow [Wed, 11 Sep 2002 01:00:57 +0000 (01:00 +0000)]
Quiet warnings about non-existent scripts. My own fault for not testing my
own patches as well as I should.

21 years agoUnbreak world by fixing 4 syntax errors and 1 wrong variable type.
Mike Barcroft [Wed, 11 Sep 2002 01:00:56 +0000 (01:00 +0000)]
Unbreak world by fixing 4 syntax errors and 1 wrong variable type.

21 years agoRemember who asked for a connect or accept operation so we can actually tell
Benno Rice [Wed, 11 Sep 2002 00:52:50 +0000 (00:52 +0000)]
Remember who asked for a connect or accept operation so we can actually tell
them when it's done.

Reviewed by: archie

21 years agoFix style bugs in macros.
Tony Finch [Tue, 10 Sep 2002 21:14:14 +0000 (21:14 +0000)]
Fix style bugs in macros.

21 years agoTidy up the base64 code and relax the error handling.
Tony Finch [Tue, 10 Sep 2002 21:10:33 +0000 (21:10 +0000)]
Tidy up the base64 code and relax the error handling.

21 years agoAdjust to reflect reality, which is that sigaltstack() takes stack_t *'s.
Archie Cobbs [Tue, 10 Sep 2002 21:06:51 +0000 (21:06 +0000)]
Adjust to reflect reality, which is that sigaltstack() takes stack_t *'s.

MFC after: 3 days

21 years agoStyle: Don't treat pointers as booleans.
Tony Finch [Tue, 10 Sep 2002 20:53:46 +0000 (20:53 +0000)]
Style: Don't treat pointers as booleans.

21 years agoFix tilde-expansion of the filename obtained from the "begin" line in -s
Tony Finch [Tue, 10 Sep 2002 20:48:44 +0000 (20:48 +0000)]
Fix tilde-expansion of the filename obtained from the "begin" line in -s
mode, and be more accurate about identifying begin lines so that spurious
ones don't have to be manually edited out, and be more forgiving about
errors so that -c mode is more useful.

21 years agoReplace a goto with continue.
Tony Finch [Tue, 10 Sep 2002 20:33:39 +0000 (20:33 +0000)]
Replace a goto with continue.

21 years agoRemove register keyword.
Tony Finch [Tue, 10 Sep 2002 19:23:44 +0000 (19:23 +0000)]
Remove register keyword.

21 years agoStyle: space after return.
Tony Finch [Tue, 10 Sep 2002 18:52:03 +0000 (18:52 +0000)]
Style: space after return.

21 years agoFix some style bugs:
Mike Barcroft [Tue, 10 Sep 2002 18:41:45 +0000 (18:41 +0000)]
Fix some style bugs:
o Space used instead of a tab after `#define' and `typedef'.
o Sentences not ended with a period.
o Unaligned function names and other spacing issues.

21 years agoMake the handling of -i neater.
Tony Finch [Tue, 10 Sep 2002 18:34:07 +0000 (18:34 +0000)]
Make the handling of -i neater.

21 years agoOnly try to chmod the output if it's a regular file, rather than
Tony Finch [Tue, 10 Sep 2002 18:28:19 +0000 (18:28 +0000)]
Only try to chmod the output if it's a regular file, rather than
hard-coding /dev/stdout as a special case.

Suggested by: jmallett

21 years agoNew release notes: gre(4), kern.cam.scsi_delay tunable.
Bruce A. Mah [Tue, 10 Sep 2002 18:22:40 +0000 (18:22 +0000)]
New release notes:  gre(4), kern.cam.scsi_delay tunable.

21 years agoFix namespace issues by using the relatively new visibility
Mike Barcroft [Tue, 10 Sep 2002 18:12:16 +0000 (18:12 +0000)]
Fix namespace issues by using the relatively new visibility
primitives.

21 years ago- Forgot to remove `cardattached` declaration in revision 1.22.
Andrew R. Reiter [Tue, 10 Sep 2002 16:50:06 +0000 (16:50 +0000)]
- Forgot to remove `cardattached` declaration in revision 1.22.

21 years agoAdd missing calls to mtx_init().
Poul-Henning Kamp [Tue, 10 Sep 2002 16:41:08 +0000 (16:41 +0000)]
Add missing calls to mtx_init().

It seems counter-intuitive that all drivers have to do this.

Pointed in right direction by: gj

21 years agoAdd a paragraph which should clarify the separation of asterisks
Robert Drehmel [Tue, 10 Sep 2002 14:48:38 +0000 (14:48 +0000)]
Add a paragraph which should clarify the separation of asterisks
and adjacent tokens in declarations.
The added text was originally a single sentence I wrote and which
was heavily modified and extended by Bruce Evans.

This clarification attempt originates from differing usage of the
'restrict' type-qualifier.
Although various documents documents dicussing the C Programming
Language put a space between an asterisk and the 'restrict' keyword,
including the C99 standard (at least the n869.txt draft) and other
ISO/IEC JTC1/SC22/WG14 documents, the IEEE Std 1003.1-2001 document
does not separate them.

Discussed with: bde
Requested by: tjr
Separation using a single space also liked by: mike

21 years agoPrevent the wrapper from looping on itself forever, when there
Thomas Quinot [Tue, 10 Sep 2002 13:56:30 +0000 (13:56 +0000)]
Prevent the wrapper from looping on itself forever, when there
is a symbolic link in the PATH pointing back to /usr/bin/perl.

Change WARNS from 6 to 5 to account for the fact that sys/time.h,
included from sys/stat.h, produces a warning when compiled with
-pedantic.

PR: bin/42418
Reviewed by: roberto

21 years agoFixed namespace pollution in uma changes:
Bruce Evans [Tue, 10 Sep 2002 13:48:21 +0000 (13:48 +0000)]
Fixed namespace pollution in uma changes:
- use `struct uma_zone *' instead of uma_zone_t, so that <sys/uma.h> isn't
  a prerequisite.
- don't include <sys/uma.h>.
Namespace pollution makes "opaque" types like uma_zone_t perfectly
non-opaque.  Such types should never be used (see style(9)).

"Fixed" subsequently grown dependencies of this header on its own
pollution by polluting explicitly:
- include <sys/mutex.h> and its prerequisite <sys/lock.h> instead of
  depending on namespace pollution 2 layers deep in <sys/uma.h>.

21 years agoThe boot floppy requires the lnc driver.
Yoshihiro Takahashi [Tue, 10 Sep 2002 13:11:03 +0000 (13:11 +0000)]
The boot floppy requires the lnc driver.

21 years agoRemove unused code.
Yoshihiro Takahashi [Tue, 10 Sep 2002 13:08:23 +0000 (13:08 +0000)]
Remove unused code.

21 years agoUse ${TARGET}/dokern.sh instead of ${TARGET_ARCH}/dokern.sh.
Yoshihiro Takahashi [Tue, 10 Sep 2002 13:06:53 +0000 (13:06 +0000)]
Use ${TARGET}/dokern.sh instead of ${TARGET_ARCH}/dokern.sh.
(pc98/dokern.sh is (repository) copied from i386/dokern.sh.)

21 years agoInclude <sys/malloc.h> instead of depending on namespace pollution 2
Bruce Evans [Tue, 10 Sep 2002 12:45:02 +0000 (12:45 +0000)]
Include <sys/malloc.h> instead of depending on namespace pollution 2
layers deep in <sys/proc.h> or <sys/vnode.h>.

21 years agoFixed namespace pollution in uma changes:
Bruce Evans [Tue, 10 Sep 2002 12:19:16 +0000 (12:19 +0000)]
Fixed namespace pollution in uma changes:
- use `struct uma_zone *' instead of uma_zone_t, so that <sys/uma.h> isn't
  a prerequisite.
- don't include <sys/uma.h>.
Namespace pollution makes "opaque" types like uma_zone_t perfectly
non-opaque.  Such types should never be used (see style(9)).

21 years agoInclude <sys/malloc.h> instead of depending on namespace pollution 2
Bruce Evans [Tue, 10 Sep 2002 11:57:02 +0000 (11:57 +0000)]
Include <sys/malloc.h> instead of depending on namespace pollution 2
layers deep in <sys/proc.h> or <sys/vnode.h>.

Removed unused includes.  Sorted includes.

21 years agovfs_syscalls.c:
Bruce Evans [Tue, 10 Sep 2002 11:09:13 +0000 (11:09 +0000)]
vfs_syscalls.c:
Changed rename(2) to follow the letter of the POSIX spec.  POSIX
requires rename() to have no effect if its args "resolve to the same
existing file".  I think "file" can only reasonably be read as referring
to the inode, although the rationale and "resolve" seem to say that
sameness is at the level of (resolved) directory entries.

ext2fs_vnops.c, ufs_vnops.c:
Replaced code that gave the historical BSD behaviour of removing one
link name by checks that this code is now unreachable.  This fixes
some races.  All vnodes needed to be unlocked for the removal, and
locking at another level using something like IN_RENAME was not even
attempted, so it was possible for rename(x, y) to return with both x
and y removed even without any unlink(2) syscalls (one process can
remove x using rename(x, y) and another process can remove y using
rename(y, x)).

Prodded by: alfred
MFC after: 8 weeks
PR: 42617

21 years agoRemove a sentence about wrapping macro definitions in bare braces,
Tony Finch [Tue, 10 Sep 2002 10:54:39 +0000 (10:54 +0000)]
Remove a sentence about wrapping macro definitions in bare braces,
which became wrong after using do { } while (0) became recommended.
Move the definition of what braces are to their new first occurrence.

Reviewed by: bde

21 years agoCosmetics: #define NETATALK --> #define NETATALK 1, so that it is in line with
Maxim Sobolev [Tue, 10 Sep 2002 10:19:03 +0000 (10:19 +0000)]
Cosmetics: #define NETATALK --> #define NETATALK 1, so that it is in line with
other #define FOO.

21 years agoReplace FILE with struct __sFILE in the prototypes for the wide character
Tim J. Robbins [Tue, 10 Sep 2002 09:43:28 +0000 (09:43 +0000)]
Replace FILE with struct __sFILE in the prototypes for the wide character
I/O functions to avoid having to bring in pollution from <stdio.h>.

Suggested by: bde

21 years agoRemove superfluous break.
Maxim Sobolev [Tue, 10 Sep 2002 09:18:33 +0000 (09:18 +0000)]
Remove superfluous break.

21 years agoRemove extraneous extern from function prototype introduced in rev.1.14.
Maxim Sobolev [Tue, 10 Sep 2002 08:20:30 +0000 (08:20 +0000)]
Remove extraneous extern from function prototype introduced in rev.1.14.

21 years agoreprobe not needed here either
Warner Losh [Tue, 10 Sep 2002 06:37:56 +0000 (06:37 +0000)]
reprobe not needed here either

21 years agoRemove more bogus reprobe code. I don't think it is needed here either.
Warner Losh [Tue, 10 Sep 2002 06:37:02 +0000 (06:37 +0000)]
Remove more bogus reprobe code.  I don't think it is needed here either.

21 years agoo Dike out the bogus reprobe stuff. It was getting in the way of newbus' own
Warner Losh [Tue, 10 Sep 2002 06:32:18 +0000 (06:32 +0000)]
o Dike out the bogus reprobe stuff.  It was getting in the way of newbus' own
  reprobe code working.  This bogusness was thrust upon me.
o Don't delete the device if no drivers attach.  I had thrust this bogusness
  upon others.

These two changes make kldload of a pccard driver work again.

21 years agoregen for 1.34
Warner Losh [Tue, 10 Sep 2002 06:29:50 +0000 (06:29 +0000)]
regen for 1.34

21 years agoAdd ARGOSY SP320
Warner Losh [Tue, 10 Sep 2002 06:29:09 +0000 (06:29 +0000)]
Add ARGOSY SP320

cis submitted by: phk

21 years agomove some printfs under bootverbose
Sam Leffler [Tue, 10 Sep 2002 05:06:48 +0000 (05:06 +0000)]
move some printfs under bootverbose

Reviewed by: phk

21 years ago- Spell "these" properly.
Andrew R. Reiter [Tue, 10 Sep 2002 03:58:44 +0000 (03:58 +0000)]
- Spell "these" properly.

21 years agoUpdate to use *.tbz suffix, also update example URL.
Bruce A. Mah [Tue, 10 Sep 2002 02:42:59 +0000 (02:42 +0000)]
Update to use *.tbz suffix, also update example URL.

While I'm here, fix a gramm-o.

PR: 42576
Submitted by: Jeff Ito <jeffi@rcn.com>
MFC after: 1 day

21 years agoImplement C99's _Exit() interface.
Garrett Wollman [Tue, 10 Sep 2002 02:04:49 +0000 (02:04 +0000)]
Implement C99's _Exit() interface.
Implement a version of qsort that provides a thunk to the comparison function.

Update manual pages.

21 years agoWithout fixing the namespace issues, add prototypes for the new _Exit()
Garrett Wollman [Tue, 10 Sep 2002 02:02:49 +0000 (02:02 +0000)]
Without fixing the namespace issues, add prototypes for the new _Exit()
and qsort_r() functions.  Fix one other missorted declaration.

21 years agoAdd the 1109 incident.
Greg Lehey [Tue, 10 Sep 2002 00:07:19 +0000 (00:07 +0000)]
Add the 1109 incident.

21 years agoSigh, finally activate pkgwrap.c.
Maxim Sobolev [Mon, 9 Sep 2002 22:59:06 +0000 (22:59 +0000)]
Sigh, finally activate pkgwrap.c.

Big pointy hat to: sobomax (who apparently needs some sleep)
Submitted by: walt <wa1ter@hotmail.com>

21 years agoFix syscons so it actually does start.
Gordon Tetlow [Mon, 9 Sep 2002 22:40:34 +0000 (22:40 +0000)]
Fix syscons so it actually does start.

Reported by: keramida, ume
Submitted by: keramida

21 years agoNew Translation
Sebastien Gioria [Mon, 9 Sep 2002 22:26:07 +0000 (22:26 +0000)]
New Translation

21 years agoAdding infrastructure for harware notes
Sebastien Gioria [Mon, 9 Sep 2002 22:09:48 +0000 (22:09 +0000)]
Adding infrastructure for harware notes

21 years agoContinuing translation
Sebastien Gioria [Mon, 9 Sep 2002 21:21:48 +0000 (21:21 +0000)]
Continuing translation

21 years agoFix minor off-by-one error.
Bruce A. Mah [Mon, 9 Sep 2002 21:21:34 +0000 (21:21 +0000)]
Fix minor off-by-one error.

Submitted by: jhb

21 years agoMake pkg_install go to the right place for 4.6-RELEASE and 4.6.2-RELEASE
Bruce A. Mah [Mon, 9 Sep 2002 19:44:17 +0000 (19:44 +0000)]
Make pkg_install go to the right place for 4.6-RELEASE and 4.6.2-RELEASE
packages.

Pointed out by: obrien
Reviewed by: jhb

21 years agoIntroduce notion of the package tools revision and allow to wrap all tools
Maxim Sobolev [Mon, 9 Sep 2002 19:43:30 +0000 (19:43 +0000)]
Introduce notion of the package tools revision and allow to wrap all tools
included into pkg_install according to the content of /var/db/pkg_install.conf
file, which specifies version and alternative location of the tools. Format
of the said file is very simple: one line which specifies revision of the
alternative version of the tools and their location separated by space,
i.e.:

20030102 /usr/local/sbin

This would allow bsd.port.mk to install and use up to date version of tools
on older system from ports.

Also add new `-P' flag to pkg_info, which causes it to report currently
installed version of package tools.

Discussed with: will

21 years agoIn reference to previous commit: use err(3).
Jacques Vidrine [Mon, 9 Sep 2002 18:51:59 +0000 (18:51 +0000)]
In reference to previous commit: use err(3).

Noticed by: bde

21 years agoMake sure a $PIR table header has a valid length before accepting the table
John Baldwin [Mon, 9 Sep 2002 18:24:35 +0000 (18:24 +0000)]
Make sure a $PIR table header has a valid length before accepting the table
as valid.

Submitted by: Michal Mertl <mime@traveller.cz>

21 years agoContinuing translation
Sebastien Gioria [Mon, 9 Sep 2002 17:59:39 +0000 (17:59 +0000)]
Continuing translation

21 years agoAdd tcl81 that was missing.
Akinori MUSHA [Mon, 9 Sep 2002 17:53:41 +0000 (17:53 +0000)]
Add tcl81 that was missing.

21 years agoQuiet another annoying warning.
Garrett Wollman [Mon, 9 Sep 2002 17:44:08 +0000 (17:44 +0000)]
Quiet another annoying warning.

21 years agoadd missing \n to printf
Bernd Walter [Mon, 9 Sep 2002 17:39:48 +0000 (17:39 +0000)]
add missing \n to printf

Approved by: gallatin (mentor)

21 years agoAdd security.mac.mmap_revocation, a flag indicating whether we
Robert Watson [Mon, 9 Sep 2002 17:12:24 +0000 (17:12 +0000)]
Add security.mac.mmap_revocation, a flag indicating whether we
should revoke access to memory maps on a process label change.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agoMinor code sync to MAC tree: push Giant locking up from
Robert Watson [Mon, 9 Sep 2002 17:10:16 +0000 (17:10 +0000)]
Minor code sync to MAC tree: push Giant locking up from
mac_cred_mmapped_drop_perms() to the caller.

Obtained from: TrustedBSD Project
Sponsored by: DARPA, NAI Labs

21 years agosshd didn't handle actual size of struct sockaddr correctly,
Hajimu UMEMOTO [Mon, 9 Sep 2002 16:49:11 +0000 (16:49 +0000)]
sshd didn't handle actual size of struct sockaddr correctly,
and did copy it as long as just size of struct sockaddr.  So,
If connection is via IPv6, sshd didn't log hostname into utmp
correctly.
This problem occured only under FreeBSD because of our hack.
However, this is potential problem of OpenSSH-portable, and
they agreed to fix this.
Though, there is no fixed version of OpenSSH-portable available
yet, since this problem is serious for IPv6 users, I commit the
fix.

Reported by: many people
Reviewed by: current@ and stable@ (no objection)
MFC after: 3 days

21 years agoBack out previous commit. I was a bit overzealous: the fd_set size is
Jacques Vidrine [Mon, 9 Sep 2002 16:43:18 +0000 (16:43 +0000)]
Back out previous commit.  I was a bit overzealous: the fd_set size is
calculated dynamically here.

21 years agoCheck for FD_SET overrun.
Jacques Vidrine [Mon, 9 Sep 2002 16:21:10 +0000 (16:21 +0000)]
Check for FD_SET overrun.

21 years agoCheck for FD_SET overrun.
Jacques Vidrine [Mon, 9 Sep 2002 16:14:03 +0000 (16:14 +0000)]
Check for FD_SET overrun.

21 years agoCheck for FD_SET overrun.
Jacques Vidrine [Mon, 9 Sep 2002 16:06:41 +0000 (16:06 +0000)]
Check for FD_SET overrun.

21 years agoCheck for FD_SET overrun.
Jacques Vidrine [Mon, 9 Sep 2002 15:58:08 +0000 (15:58 +0000)]
Check for FD_SET overrun.

21 years ago - Fix iBCS2 emulation of COFF files that have multiple libraries
Robert Drehmel [Mon, 9 Sep 2002 15:51:10 +0000 (15:51 +0000)]
 - Fix iBCS2 emulation of COFF files that have multiple libraries
   in their library (STYP_LIB) section.
 - Attempt to make the code which calculates the next entry and
   string offsets look clearer.

PR: kern/42580
Tested by: Olaf Klein <ok@adimus.de> (on 4.7-PRERELEASE)

21 years agoCheck for FD_SET overruns.
Jacques Vidrine [Mon, 9 Sep 2002 15:45:28 +0000 (15:45 +0000)]
Check for FD_SET overruns.

Obtained from: NetBSD

21 years agoKill debugging line occasionally slipped into rev.1.3.
Maxim Sobolev [Mon, 9 Sep 2002 11:13:00 +0000 (11:13 +0000)]
Kill debugging line occasionally slipped into rev.1.3.

Submitted by: reg

21 years agoRemove the second lnc line. Move CD9660 to the third floppy to make
John Hay [Mon, 9 Sep 2002 11:01:32 +0000 (11:01 +0000)]
Remove the second lnc line. Move CD9660 to the third floppy to make
mfsroot.flp fit again.

21 years agoSince from now on encap_input() also catches IPPROTO_MOBILE and IPPROTO_GRE
Maxim Sobolev [Mon, 9 Sep 2002 09:36:47 +0000 (09:36 +0000)]
Since from now on encap_input() also catches IPPROTO_MOBILE and IPPROTO_GRE
packets in addition to IPPROTO_IPV4 and IPPROTO_IPV6, explicitly specify
IPPROTO_IPV4 or IPPROTO_IPV6 instead of -1 when calling encap_attach().

MFC after: 28 days
(along with other if_gre changes)

21 years ago- Do not swap out a process if it is in creation. The process may have no
Seigo Tanimura [Mon, 9 Sep 2002 09:05:06 +0000 (09:05 +0000)]
- Do not swap out a process if it is in creation.  The process may have no
  address space yet.

- Check whether a process is a system process prior to dereferencing
  its p_vmspace.  Aio assumes that only the curthread switches the address
  space of a system process.

21 years ago#include "opt_bla.h" goes first says Bruce.
Poul-Henning Kamp [Mon, 9 Sep 2002 08:44:52 +0000 (08:44 +0000)]
#include "opt_bla.h" goes first says Bruce.

21 years agoAdd `device gre'.
Maxim Sobolev [Mon, 9 Sep 2002 08:31:04 +0000 (08:31 +0000)]
Add `device gre'.

Reminded by:    bde
MFC after:      28 days
(along with other if_gre stuff)

21 years agoPrevent namespace pollution in use-land by putting everything used only in
Maxim Sobolev [Mon, 9 Sep 2002 08:25:00 +0000 (08:25 +0000)]
Prevent namespace pollution in use-land by putting everything used only in
kernel (softc and such) under #ifdef _KERNEL.

Submitted by:   bde

21 years agoVastly decrease amount of memory comsumed in the case when we have to read
Maxim Sobolev [Mon, 9 Sep 2002 08:00:32 +0000 (08:00 +0000)]
Vastly decrease amount of memory comsumed in the case when we have to read
ports/INDEX, by allocating eactly amount of memory necessary for storing
each particular entry, insdead of 4K per entry (more than 7000 entries -
go figure). Memory consumption went down to some 500K from some 30M.

21 years agoRemove the typedef of wctype_t from this header; we don't support the
Tim J. Robbins [Mon, 9 Sep 2002 07:17:55 +0000 (07:17 +0000)]
Remove the typedef of wctype_t from this header; we don't support the
XSI mistake of making everything from <wctype.h> visible here.
If we did choose to support it, we could just #include <wctype.h>.

21 years agoBump __FreeBSD_cc_version for gcc 3.1-prerelease -> 3.2.1-snap upgrade.
David E. O'Brien [Mon, 9 Sep 2002 06:56:29 +0000 (06:56 +0000)]
Bump __FreeBSD_cc_version for gcc 3.1-prerelease -> 3.2.1-snap upgrade.

21 years agoUse --fast-read mode to extract metadata from package file. Since metadata
Maxim Sobolev [Mon, 9 Sep 2002 06:06:09 +0000 (06:06 +0000)]
Use --fast-read mode to extract metadata from package file. Since metadata
files are located at the very beginning of the package, this patch in
conjuction with latest tar(1) --fast-mode fix greatly speeds up pkg_info(1)
operation on package files.

MFC after: 1 week

21 years agoIn the --fast-read mode send SIGTERM to decompressor once we have extracted
Maxim Sobolev [Mon, 9 Sep 2002 06:00:16 +0000 (06:00 +0000)]
In the --fast-read mode send SIGTERM to decompressor once we have extracted
all files we have been asked for. This vastly improves performance of
pkg_add(1), because usually it only extract +CONTENTS file in --fast-read
mode, checks it and only then extracts full content of tarball in normal
mode. Without this patch, it have had to decompress the tarball twice,
effectively doubling consumed CPU time.

MFC after: 1 week

21 years agoSolve the <runetype.h> pollution problem by disabling inline
Mike Barcroft [Mon, 9 Sep 2002 05:38:05 +0000 (05:38 +0000)]
Solve the <runetype.h> pollution problem by disabling inline
optimizations when a standard has been requested, except when the
inline optimizations are also specifically requested.

21 years agoRemove duplicate entry. Since we call things cbb in the source, pccbb
Warner Losh [Mon, 9 Sep 2002 05:35:46 +0000 (05:35 +0000)]
Remove duplicate entry.  Since we call things cbb in the source, pccbb
is obsolete.

21 years agostyle(9): line up function names.
Mike Barcroft [Mon, 9 Sep 2002 03:52:43 +0000 (03:52 +0000)]
style(9): line up function names.

21 years agoCatch up with the ISO-C++ standard.
David E. O'Brien [Mon, 9 Sep 2002 02:58:42 +0000 (02:58 +0000)]
Catch up with the ISO-C++ standard.

Submitted by: Craig Rodrigues <rodrigc@attbi.com>

21 years agoUse "options " rather than "options<tab>".
Jun Kuriyama [Mon, 9 Sep 2002 02:40:59 +0000 (02:40 +0000)]
Use "options " rather than "options<tab>".

21 years agoFix typo (s@src/crypto/openssh-portable@src/crypto/openssh@).
Jun Kuriyama [Mon, 9 Sep 2002 02:00:28 +0000 (02:00 +0000)]
Fix typo (s@src/crypto/openssh-portable@src/crypto/openssh@).

21 years agoBack out last commit where I moved pcn(4) off of the modules floppy back
David E. O'Brien [Sun, 8 Sep 2002 23:59:38 +0000 (23:59 +0000)]
Back out last commit where I moved pcn(4) off of the modules floppy back
into the install kernel.  Unfortunately pcn(4) also needs mii(4) so that
would also have to added to install kernel, which will bloat it up so that it
doesn't fit on the floppy any more.  Turns out we grew a lnc(4) module since
I last looked.  So handle it as a kld loadable module during install rather
than have it statically compiled into the kernel.

21 years agoChange the order that we look for free memory windows from 0 -> MAX-1
Warner Losh [Sun, 8 Sep 2002 22:18:06 +0000 (22:18 +0000)]
Change the order that we look for free memory windows from 0 -> MAX-1
to MAX-1 -> 0.  This should allow for less potential for conflict
between pccardd/pccardc and drivers.  As far as I can tell no drivers
try to use window 4, so this should be a no-op for them.

21 years agoUpdate the hardware list for the bge driver.
John Polstra [Sun, 8 Sep 2002 19:32:08 +0000 (19:32 +0000)]
Update the hardware list for the bge driver.

21 years agoMention support for the on-board NICs on IBM x235 servers.
John Polstra [Sun, 8 Sep 2002 19:20:23 +0000 (19:20 +0000)]
Mention support for the on-board NICs on IBM x235 servers.

Be more precise when referring to the chip family (BCM570x) vs.
specific models (e.g., BCM5700).

MFC after: 1 month

21 years agoAdd support for the BCM5703x chips. I do not have one of these
John Polstra [Sun, 8 Sep 2002 19:12:02 +0000 (19:12 +0000)]
Add support for the BCM5703x chips.  I do not have one of these
cards to test; however the submitter reports that this patch works
with the on-board interface on the IBM x235 server.

Submitted by: Jung-uk Kim <jkim@niksun.com>
MFC after: 1 month