]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoThis commit was generated by cvs2svn to compensate for changes in r146895,
Alexander Kabaev [Fri, 3 Jun 2005 03:28:44 +0000 (03:28 +0000)]
This commit was generated by cvs2svn to compensate for changes in r146895,
which included commits to RCS files with non-trunk default branches.

19 years agoGcc 3.4.4 release.
Alexander Kabaev [Fri, 3 Jun 2005 03:28:44 +0000 (03:28 +0000)]
Gcc 3.4.4 release.

19 years agoAdd support for IPv4 only rules to IPFW2 now that it supports IPv6 as well.
Max Laier [Fri, 3 Jun 2005 01:10:28 +0000 (01:10 +0000)]
Add support for IPv4 only rules to IPFW2 now that it supports IPv6 as well.
This is the last requirement before we can retire ip6fw.

Reviewed by: dwhite, brooks(earlier version)
Submitted by: dwhite (manpage)
Silence from: -ipfw

19 years agoCorrect typo in a comment describing vshiftl().
Hiten Pandya [Thu, 2 Jun 2005 23:56:10 +0000 (23:56 +0000)]
Correct typo in a comment describing vshiftl().

19 years agoIn aio_waitcomplete() correct two cases of using an aiocb after freeing it.
Alan Cox [Thu, 2 Jun 2005 23:14:38 +0000 (23:14 +0000)]
In aio_waitcomplete() correct two cases of using an aiocb after freeing it.

19 years agoDocument the previously existing RELENG_[45]_* security branch locks.
Peter Wemm [Thu, 2 Jun 2005 22:57:30 +0000 (22:57 +0000)]
Document the previously existing RELENG_[45]_* security branch locks.

Approved by:  core (quite a while ago)

19 years agoAdd a 'statement of intent' blurb to the top of MAINTAINERS. Split
Peter Wemm [Thu, 2 Jun 2005 21:28:33 +0000 (21:28 +0000)]
Add a 'statement of intent' blurb to the top of MAINTAINERS.  Split
the role of MAINTAINERS into advisory and strict parts. Introduce a
new LOCKS file to document enforced locked parts of the tree.

Strict locks are only added with core approval and will generally
have a renewal timeout.

Clarify that the source tree is a community effort, not a place to stake
out 'turf'.

This will be refined as needed.

With-core-hat-on:  Yes

19 years agoFix the load64 and store64 macros, created to handle 8-byte unaligned
Marcel Moolenaar [Thu, 2 Jun 2005 05:34:08 +0000 (05:34 +0000)]
Fix the load64 and store64 macros, created to handle 8-byte unaligned
loads and stores (resp.) The ldq_u and stq_u instruction mask off the
lower 3 bits of the final address before loading from or storing to
the address, so as to avoid unaligned loads and stores. They do not
themselves allow loads from or stores to unaligned addresses. Replace
the macro definitions by a packed struct dereference.

Submitted by: Richard Henderson (rth at twiddle dot net)

19 years agorestore led state on resume
Sam Leffler [Thu, 2 Jun 2005 03:18:55 +0000 (03:18 +0000)]
restore led state on resume

Submitted by: markus

19 years agoHonouring current authmode setting. With this patch, my wi card can now
Tai-hwa Liang [Thu, 2 Jun 2005 01:51:16 +0000 (01:51 +0000)]
Honouring current authmode setting.  With this patch, my wi card can now
associate to an AP which uses shared WEP key authentication.

Tested with: "ifconfig wi0 authmode shared"
Reviewed by: imp, sam

19 years agoUse IFF_LOCKGIANT/IFF_UNLOCKGIANT around calls to the interface
Ian Dowse [Thu, 2 Jun 2005 00:04:08 +0000 (00:04 +0000)]
Use IFF_LOCKGIANT/IFF_UNLOCKGIANT around calls to the interface
if_ioctl routine. This should fix a number of code paths through
soo_ioctl() that could call into Giant-locked network drivers without
first acquiring Giant.

19 years agoBack out revision 1.19 and 1.20 until I find mental clarity to deal with
Eivind Eklund [Wed, 1 Jun 2005 20:34:06 +0000 (20:34 +0000)]
Back out revision 1.19 and 1.20 until I find mental clarity to deal with
issues bde pointed out.

19 years agoAdd myself as a new committer.
Brad Davis [Wed, 1 Jun 2005 18:29:50 +0000 (18:29 +0000)]
Add myself as a new committer.

Approved by: ceri (mentor)

19 years agoProvide missing implementations for archive_entry_ctime() and
Scott Long [Wed, 1 Jun 2005 18:20:02 +0000 (18:20 +0000)]
Provide missing implementations for archive_entry_ctime() and
archive_entry_ctime_nano()

19 years agoGiant is no longer required in kern_setrlimit(); remove its acquisition and
Alan Cox [Wed, 1 Jun 2005 17:52:51 +0000 (17:52 +0000)]
Giant is no longer required in kern_setrlimit(); remove its acquisition and
release.

Reviewed by: jhb

19 years agoAllow the VESA code to handle devices that don't claim to be VGA devices.
John Baldwin [Wed, 1 Jun 2005 16:02:39 +0000 (16:02 +0000)]
Allow the VESA code to handle devices that don't claim to be VGA devices.
This fixes VESA support when running under vmware.

PR: i386/81445
Submitted by: Jia-Shiun Li jiashiun at gmail dot com
MFC after: 1 week

19 years agoRemove the C99-specific __func__ that is one of the few barrier to
Tim Kientzle [Wed, 1 Jun 2005 15:52:39 +0000 (15:52 +0000)]
Remove the C99-specific __func__ that is one of the few barrier to
compiling on IRIX and Solaris.  Remove the "archive_check_magic" macro
that existed only to provide __func__ to the underlying __archive_check_magic
function.

Thanks to: Darin Broady
MFC after: 14 days

19 years ago<Ooops> Two lines got dropped from the previous commit. Apologies.
Tim Kientzle [Wed, 1 Jun 2005 15:46:47 +0000 (15:46 +0000)]
<Ooops>  Two lines got dropped from the previous commit.  Apologies.

MFC after: 14 days

19 years agoA minor refinement to "pax" output: Remove suid/sgid/sticky bits
Tim Kientzle [Wed, 1 Jun 2005 15:44:23 +0000 (15:44 +0000)]
A minor refinement to "pax" output:  Remove suid/sgid/sticky bits
from mode before using mode for extended attributes entry, copy
mtime/atime/ctime to extended attributes entry so it's a little more
clear that it corresponds to the like-named regular entry.

MFC after: 14 days

19 years agoTemperarly disable building in the bzip2 support by default so we can fit
David E. O'Brien [Wed, 1 Jun 2005 15:32:57 +0000 (15:32 +0000)]
Temperarly disable building in the bzip2 support by default so we can fit
on the i386 floppies.  Sigh, I hate floppies.

19 years agoFix some warnings on 64-bit archs.
John Baldwin [Wed, 1 Jun 2005 14:55:45 +0000 (14:55 +0000)]
Fix some warnings on 64-bit archs.

Reviewed by: sam

19 years agoTypo.
John Baldwin [Wed, 1 Jun 2005 14:07:43 +0000 (14:07 +0000)]
Typo.

Submitted by: njl

19 years agoo Teach login(1) to respect "hushlogin" and "nocheckmail" attributes
Maxim Konovalov [Wed, 1 Jun 2005 12:23:06 +0000 (12:23 +0000)]
o Teach login(1) to respect "hushlogin" and "nocheckmail" attributes
defined in user's $HOME/.login_conf.

PR: bin/75001
Submitted by: Rostislav Krasny
MFC after: 2 weeks

19 years agoWhen aborting tcp_attach() due to a problem allocating or attaching the
Robert Watson [Wed, 1 Jun 2005 12:14:56 +0000 (12:14 +0000)]
When aborting tcp_attach() due to a problem allocating or attaching the
tcpcb, lock the inpcb before calling in_pcbdetach() or in6_pcbdetach(),
as they expect the inpcb to be passed locked.

MFC after: 7 days

19 years agoAssert tcbinfo lock, inpcb lock in tcp_disconnect().
Robert Watson [Wed, 1 Jun 2005 12:08:15 +0000 (12:08 +0000)]
Assert tcbinfo lock, inpcb lock in tcp_disconnect().
Assert tcbinfo lock, inpcb lock in in tcp_usrclosed().

MFC after: 7 days

19 years agoAssert tcbinfo lock in tcp_drop() due to its call of tcp_close()
Robert Watson [Wed, 1 Jun 2005 12:06:07 +0000 (12:06 +0000)]
Assert tcbinfo lock in tcp_drop() due to its call of tcp_close()
Assert tcbinfo lock in tcp_close() due to its call to in{,6}_detach()
Assert tcbinfo lock in tcp_drop_syn_sent() due to its call to tcp_drop()

MFC after: 7 days

19 years agoAssert that tcbinfo is locked in tcp_input() before calling into
Robert Watson [Wed, 1 Jun 2005 12:03:18 +0000 (12:03 +0000)]
Assert that tcbinfo is locked in tcp_input() before calling into
tcp_drop().

MFC after: 7 days

19 years agoAssert the tcbinfo lock whenever tcp_close() is to be called by
Robert Watson [Wed, 1 Jun 2005 11:49:14 +0000 (11:49 +0000)]
Assert the tcbinfo lock whenever tcp_close() is to be called by
tcp_input().

MFC after: 7 days

19 years agoAssert tcbinfo lock in tcp_attach(), as it is required; the caller
Robert Watson [Wed, 1 Jun 2005 11:44:43 +0000 (11:44 +0000)]
Assert tcbinfo lock in tcp_attach(), as it is required; the caller
(tcp_usr_attach()) currently grabs it.

MFC after: 7 days

19 years agoCommit correct version of previous commit (in_pcb.c:1.164). Use the
Robert Watson [Wed, 1 Jun 2005 11:43:39 +0000 (11:43 +0000)]
Commit correct version of previous commit (in_pcb.c:1.164).  Use the
local variables as currently named.

MFC after: 7 days

19 years agoAssert pcbinfo lock in in_pcbdisconnect() and in_pcbdetach(), as the
Robert Watson [Wed, 1 Jun 2005 11:39:42 +0000 (11:39 +0000)]
Assert pcbinfo lock in in_pcbdisconnect() and in_pcbdetach(), as the
global pcb lists are modified.

MFC after: 7 days

19 years agoSlight white space tweak.
Robert Watson [Wed, 1 Jun 2005 11:38:35 +0000 (11:38 +0000)]
Slight white space tweak.

MFC after: 7 days

19 years agoLock udbinfo and inp before calling in6_pcbdetach() from udp6_abort().
Robert Watson [Wed, 1 Jun 2005 11:38:19 +0000 (11:38 +0000)]
Lock udbinfo and inp before calling in6_pcbdetach() from udp6_abort().

MFC after: 1 week

19 years agoAdd a test for escaped newline handling.
Hartmut Brandt [Wed, 1 Jun 2005 11:26:47 +0000 (11:26 +0000)]
Add a test for escaped newline handling.

19 years agoAdd support for TODO tests. These are tests that are supposed to fail,
Hartmut Brandt [Wed, 1 Jun 2005 11:25:38 +0000 (11:25 +0000)]
Add support for TODO tests. These are tests that are supposed to fail,
because of a make bug that should be fixed.

19 years agoDe-spl UDP.
Robert Watson [Wed, 1 Jun 2005 11:24:00 +0000 (11:24 +0000)]
De-spl UDP.

MFC after: 3 days

19 years agoAdd a handy macro to represent null mount option, MOPT_NULL, and make
Xin LI [Wed, 1 Jun 2005 09:39:36 +0000 (09:39 +0000)]
Add a handy macro to represent null mount option, MOPT_NULL, and make
use of the macro in sbin/mount*'s, by replacing:

mopts[] = {
    MOPT_STDOPTS,
    { NULL }
}

With:
mopts[] = {
    MOPT_STDOPTS,
    MOPT_NULL
}

This change will help to reduce the situation that we don't explicitly
initialize "struct mntopt"'s.  It should not contribute to any
functional/logical changes as far as I can tell.

19 years agoAdding note about required firmware to save others time and frustration with
Tai-hwa Liang [Wed, 1 Jun 2005 07:13:21 +0000 (07:13 +0000)]
Adding note about required firmware to save others time and frustration with
this device.

Obtained from: imp (if_iwi)
Encouraged by: damien

19 years agoAdd a text description for the Intel IOP302/303 processors. Be slightly
Scott Long [Wed, 1 Jun 2005 07:11:17 +0000 (07:11 +0000)]
Add a text description for the Intel IOP302/303 processors.  Be slightly
more verbose about the allocation of RAM on the controller.

Sbumitted by: Jeremy Chadwick
PR: kern/81259
MFC-After: 3 days

19 years agoPrinting a warning once when trying to bring up interface before firmware load.
Tai-hwa Liang [Wed, 1 Jun 2005 01:54:00 +0000 (01:54 +0000)]
Printing a warning once when trying to bring up interface before firmware load.

Obtained from: imp (if_iwi)
Reviewed by: damien

19 years agoDon't use a patch w/in /usr/src. Programmatically change files when needed.
David E. O'Brien [Tue, 31 May 2005 22:16:49 +0000 (22:16 +0000)]
Don't use a patch w/in /usr/src.  Programmatically change files when needed.

19 years agoDon't enable I/O or memory mode in a device's command register if the BAR
John Baldwin [Tue, 31 May 2005 21:33:33 +0000 (21:33 +0000)]
Don't enable I/O or memory mode in a device's command register if the BAR
we are processing has a base address of zero.  Note that this will only
change behavior for devices where all the BARs of a given type have a base
address of 0 since we will enable the appropriate access when we encounter
the first BAR with a base that is not 0.  Specifically, this allows certain
Toshiba laptops to no longer require 'hw.pci.enable_io_modes=0' to avoid
hangs during boot.

PR: kern/20040
PR: i386/63776 (possibly)
PR: i386/68900 (possibly)
PR: i386/74532 (possibly)
MFC after: 1 week

19 years agoAdd a missing const to alpha_setcurrdev() to quiet a warning.
John Baldwin [Tue, 31 May 2005 21:23:43 +0000 (21:23 +0000)]
Add a missing const to alpha_setcurrdev() to quiet a warning.

19 years agoChange the type the buf arg to the strategy routines from void * to char *
John Baldwin [Tue, 31 May 2005 21:23:21 +0000 (21:23 +0000)]
Change the type the buf arg to the strategy routines from void * to char *
to quiet some warnings.

19 years agoWhitespace.
John Baldwin [Tue, 31 May 2005 21:22:01 +0000 (21:22 +0000)]
Whitespace.

19 years agoInitialise `hex' at the top of the file, ISO C requires that an array with
Stefan Farfeleder [Tue, 31 May 2005 21:19:14 +0000 (21:19 +0000)]
Initialise `hex' at the top of the file, ISO C requires that an array with
internal linkage has a complete type.

19 years agoPut bzip2 support on equal footing with gzip support.
David E. O'Brien [Tue, 31 May 2005 21:16:50 +0000 (21:16 +0000)]
Put bzip2 support on equal footing with gzip support.
Enable bzip2 support by default, set LOADER_NO_BZIP2_SUPPORT to disable it.

Pointy hat to: sobomax

19 years agoBack out revision 1.51, it is wrong. We don't litter -I's within lib
David E. O'Brien [Tue, 31 May 2005 20:39:53 +0000 (20:39 +0000)]
Back out revision 1.51, it is wrong.  We don't litter -I's within lib
Makefiles to get headers from /usr/src vs. the standard include paths.

19 years agoRemove #ident directives, they don't exist on the vendor branch anymore.
Stefan Farfeleder [Tue, 31 May 2005 20:28:49 +0000 (20:28 +0000)]
Remove #ident directives, they don't exist on the vendor branch anymore.

19 years agoReassert some maintainership. People aren't using good judgment.
David E. O'Brien [Tue, 31 May 2005 20:12:59 +0000 (20:12 +0000)]
Reassert some maintainership.  People aren't using good judgment.

19 years agoUse __FBSDID() over #ident.
Stefan Farfeleder [Tue, 31 May 2005 20:07:46 +0000 (20:07 +0000)]
Use __FBSDID() over #ident.

19 years agoUse %z to print size_t values.
John Baldwin [Tue, 31 May 2005 20:01:58 +0000 (20:01 +0000)]
Use %z to print size_t values.

19 years agoJust use uintptr_t and intptr_t rather than requiring each arch to provide
John Baldwin [Tue, 31 May 2005 20:01:18 +0000 (20:01 +0000)]
Just use uintptr_t and intptr_t rather than requiring each arch to provide
explicit int/long typedefs.

19 years agoMove #ident into #if 0.
Stefan Farfeleder [Tue, 31 May 2005 20:00:29 +0000 (20:00 +0000)]
Move #ident into #if 0.

19 years agoAdd a missing END() to quiet a warning.
John Baldwin [Tue, 31 May 2005 19:58:23 +0000 (19:58 +0000)]
Add a missing END() to quiet a warning.

19 years agoMark _thread_exit() and __sys_exit() as __dead2 to quiet some warnings.
John Baldwin [Tue, 31 May 2005 19:57:23 +0000 (19:57 +0000)]
Mark _thread_exit() and __sys_exit() as __dead2 to quiet some warnings.

19 years agoStyle tweak.
John Baldwin [Tue, 31 May 2005 19:56:52 +0000 (19:56 +0000)]
Style tweak.

19 years agoThis patch addresses a standards violation issue. The standards say a
Ken Smith [Tue, 31 May 2005 19:39:52 +0000 (19:39 +0000)]
This patch addresses a standards violation issue.  The standards say a
file's access time should be updated when it gets executed.  A while
ago the mechanism used to exec was changed to use a more mmap based
mechanism and this behavior was broken as a side-effect of that.

A new vnode flag is added that gets set when the file gets executed,
and the VOP_SETATTR() vnode operation gets called.  The underlying
filesystem is expected to handle it based on its own semantics, some
filesystems don't support access time at all.  Those that do should
handle it in a way that does not block, does not generate I/O if possible,
etc.  In particular vn_start_write() has not been called.  The UFS code
handles it the same way as it would normally handle the access time if
a file was read - the IN_ACCESS flag gets set in the inode but no other
action happens at this point.  The actual time update will happen later
during a sync (which handles all the necessary locking).

Got me into this: cperciva
Discussed with: a lot with bde, a little with kan
Showed patches to: phk, jeffr, standards@, arch@
Minor discussion on: arch@

19 years agoEnsure GCC does not use FP registers in integer code.
David E. O'Brien [Tue, 31 May 2005 18:13:29 +0000 (18:13 +0000)]
Ensure GCC does not use FP registers in integer code.
I really don't like cluttering up the compiler invocation,
but this bigger hammer will fix reported problems for now.

19 years agoo Missed colon in getopt(3) argument makes tftpd(8) crash. Fix that.
Maxim Konovalov [Tue, 31 May 2005 17:22:53 +0000 (17:22 +0000)]
o Missed colon in getopt(3) argument makes tftpd(8) crash.  Fix that.

PR: misc/81732
Submitted by: Denis Grudkin
MFC after: 2 weeks

19 years agoAccording to:
Craig Rodrigues [Tue, 31 May 2005 15:18:17 +0000 (15:18 +0000)]
According to:
http://www.opengroup.org/onlinepubs/009695399/basedefs/sys/types.h.html

#include <sys/types.h>
should include the definitions of pthread types.

PR: standards/78907
Reported by: Brooks Davis
Approved by: das (mentor)

19 years agoDo not declare a struct as extern, and then implement
Craig Rodrigues [Tue, 31 May 2005 14:50:49 +0000 (14:50 +0000)]
Do not declare a struct as extern, and then implement
it as static in the same file.  This is not legal C,
and GCC 4.0 will issue an error.

Reviewed by: phk
Approved by: das (mentor)

19 years agoChange the test infrastructure so that tests are easier to maintain and
Hartmut Brandt [Tue, 31 May 2005 14:13:07 +0000 (14:13 +0000)]
Change the test infrastructure so that tests are easier to maintain and
so that make(1) will run in an almost clean environment and enhance the
description of the test infrastructure.

Add the ability to have multiple tests carried out per test script.

Give some tests more meaningful names.

Fix the usage message from the test scripts.

Make it possible to pass several commands to the test scripts like:
'sh test.t setup run compare clean'.

19 years agoThe new name for the DIAGNOSTICS section is EXIT STATUS.
Tim J. Robbins [Tue, 31 May 2005 12:57:44 +0000 (12:57 +0000)]
The new name for the DIAGNOSTICS section is EXIT STATUS.

Noticed by: ru

19 years agoPass the RE_ICASE flag to re_set_syntax() as appropriate to fix
Tim J. Robbins [Tue, 31 May 2005 12:49:04 +0000 (12:49 +0000)]
Pass the RE_ICASE flag to re_set_syntax() as appropriate to fix
case-insensitive matching (-i option) in certain multibyte locales.

Obtained from: Fedora

19 years agoMFi386: revision 1.533.
Yoshihiro Takahashi [Tue, 31 May 2005 11:44:42 +0000 (11:44 +0000)]
MFi386: revision 1.533.

19 years agoAdd support for XMM registers in GDB for x86 processors that support
Doug Rabson [Tue, 31 May 2005 09:43:04 +0000 (09:43 +0000)]
Add support for XMM registers in GDB for x86 processors that support
SSE (or its successors).

Reviewed by: marcel, davidxu
MFC After: 2 weeks

19 years agoFor ${SUBDIR} targets, change the type of dependency operator from `::'
Ruslan Ermilov [Tue, 31 May 2005 07:14:51 +0000 (07:14 +0000)]
For ${SUBDIR} targets, change the type of dependency operator from `::'
to `:', so that it stays compatible with a stale dependency recorded in
.depend when the type of "foo" changes from file to directory or back.
Compensate for the loss of the "If no sources are specified, the target
is always re-created" feature by marking these targets with the .PHONY
attribute.  While here, fix a bug in the target's script (nobody uses
these targets apparently).

19 years agoAdd install-info to the list of install tools now that we don't
Ruslan Ermilov [Tue, 31 May 2005 04:45:47 +0000 (04:45 +0000)]
Add install-info to the list of install tools now that we don't
always bootstrap texinfo.

Reported by: scottl

19 years agoSynchronize access to aio_freeproc with a mutex. Eliminate related spl
Alan Cox [Mon, 30 May 2005 22:26:34 +0000 (22:26 +0000)]
Synchronize access to aio_freeproc with a mutex.  Eliminate related spl
calls.

Reduce the scope of Giant in aio_daemon().

19 years agoAdd /etc/security, into which the BSM audit configuration files will be
Robert Watson [Mon, 30 May 2005 20:51:13 +0000 (20:51 +0000)]
Add /etc/security, into which the BSM audit configuration files will be
installed.  This is the same directory as found on Solaris.

NB: In FreeBSD 4.x and earlier, a script (file) named /etc/security
exists.  Does mergemaster need to be taught how to replace a file with
a directory?

Submitted by: wsalamon
Obtained from: TrustedBSD Project

19 years agoUse the proc mtx to prevent simultaneous changes to p_aioinfo.
Alan Cox [Mon, 30 May 2005 19:33:33 +0000 (19:33 +0000)]
Use the proc mtx to prevent simultaneous changes to p_aioinfo.

19 years agoMake apm(8) understand AC Line state 2 as "backup power".
Will Andrews [Mon, 30 May 2005 18:44:43 +0000 (18:44 +0000)]
Make apm(8) understand AC Line state 2 as "backup power".

Submitted by: liamfoy@sepulcrum.org

19 years agoFix LINT by defining vga_pxlmouse_planar and vga_pxlmouse_direct.
Scott Long [Mon, 30 May 2005 18:37:25 +0000 (18:37 +0000)]
Fix LINT by defining vga_pxlmouse_planar and vga_pxlmouse_direct.

19 years agoEliminate unnecessary calls to wakeup(); no one sleeps on &aio_freeproc.
Alan Cox [Mon, 30 May 2005 18:02:00 +0000 (18:02 +0000)]
Eliminate unnecessary calls to wakeup(); no one sleeps on &aio_freeproc.

Eliminate an unused flag, AIOP_SCHED; it's cleared but never set.

19 years agoNew release notes:
Hiroki Sato [Mon, 30 May 2005 16:51:57 +0000 (16:51 +0000)]
New release notes:
syscons VESA support,
libpcap v0.8.3 --> v0.9.1 (alpha 096), and
TCPDUMP v3.8.3 --> v3.9.1 (alpha 096).

19 years agoRebuild generated system call definition files following the addition of
Robert Watson [Mon, 30 May 2005 15:20:21 +0000 (15:20 +0000)]
Rebuild generated system call definition files following the addition of
the audit event field to the syscalls.master file format.

Submitted by: wsalamon
Obtained from: TrustedBSD Project

19 years agoIntroduce a new field in the syscalls.master file format to hold the
Robert Watson [Mon, 30 May 2005 15:09:18 +0000 (15:09 +0000)]
Introduce a new field in the syscalls.master file format to hold the
audit event identifier associated with each system call, which will
be stored by makesyscalls.sh in the sy_auevent field of struct sysent.
For now, default the audit identifier on all system calls to AUE_NULL,
but in the near future, other BSM event identifiers will be used.  The
mapping of system calls to event identifiers is many:one due to
multiple system calls that map to the same end functionality across
compatibility wrappers, ABI wrappers, etc.

Submitted by: wsalamon
Obtained from: TrustedBSD Project

19 years agoMarkup nits.
Ruslan Ermilov [Mon, 30 May 2005 12:33:04 +0000 (12:33 +0000)]
Markup nits.

19 years agoAdd the Dell PERC 4/IM as supported.
Christian Brueffer [Mon, 30 May 2005 12:26:12 +0000 (12:26 +0000)]
Add the Dell PERC 4/IM as supported.

Submitted by: Muthu_T@Dell.com

19 years ago - Don't set our bio op to be a READ when we've just completed a write. There
Jeff Roberson [Mon, 30 May 2005 07:04:15 +0000 (07:04 +0000)]
 - Don't set our bio op to be a READ when we've just completed a write.  There
   are subtle differences in the read and write completion path.  Instead,
   grab an extra write ref so the write path can drop it when we recursively
   call bufdone().  I believe this may be the source of the wrong bufobj
   panics.

Reported by: pho, kkenn

19 years ago - Add bufobj_wrefl() to add a write ref to a bufobj that is already locked.
Jeff Roberson [Mon, 30 May 2005 07:01:18 +0000 (07:01 +0000)]
 - Add bufobj_wrefl() to add a write ref to a bufobj that is already locked.

19 years agoDeny to switch into banked video mode when it is not available. Some
Xin LI [Mon, 30 May 2005 06:45:40 +0000 (06:45 +0000)]
Deny to switch into banked video mode when it is not available.  Some
users has reported corrupted display with old video cards.

Submitted by: Michal Mertl <mime traveller cz>

19 years agoKernel hooks to support PMC sampling modes.
Joseph Koshy [Mon, 30 May 2005 06:29:29 +0000 (06:29 +0000)]
Kernel hooks to support PMC sampling modes.

Reviewed by: alc

19 years agoEliminate aio_activeproc; it's unused.
Alan Cox [Mon, 30 May 2005 05:25:10 +0000 (05:25 +0000)]
Eliminate aio_activeproc; it's unused.

19 years agomalloc.h relies on param.h for a definition of MAXCPU. I guess that there is
Scott Long [Mon, 30 May 2005 05:01:44 +0000 (05:01 +0000)]
malloc.h relies on param.h for a definition of MAXCPU.  I guess that there is
other header pollution that makes this work right now, but it falls over when
doing a RELENG_5 -> HEAD upgrade.

19 years agoCorrect mistake in previous commit: add 'bsm' to LDIRS not LSUBDIRS.
Robert Watson [Mon, 30 May 2005 01:06:29 +0000 (01:06 +0000)]
Correct mistake in previous commit: add 'bsm' to LDIRS not LSUBDIRS.

Pointy hat: over here, please

19 years agoCreate nexus in configure_first() instead of in configure(). This
Marcel Moolenaar [Sun, 29 May 2005 23:44:22 +0000 (23:44 +0000)]
Create nexus in configure_first() instead of in configure(). This
makes sure that sysinit tasks that run after configure_first(),
but before configure() have a nexus to hang devices off.

19 years agos/-1000/BUS_PROBE_LOW_PRIORITY/
Seigo Tanimura [Sun, 29 May 2005 23:22:23 +0000 (23:22 +0000)]
s/-1000/BUS_PROBE_LOW_PRIORITY/

Pointed out by: nyan

19 years agoCall cninit_finish() from configure_final().
Marcel Moolenaar [Sun, 29 May 2005 22:53:48 +0000 (22:53 +0000)]
Call cninit_finish() from configure_final().

19 years agoCall cninit_finish() in configure_final().
Marcel Moolenaar [Sun, 29 May 2005 22:48:41 +0000 (22:48 +0000)]
Call cninit_finish() in configure_final().

19 years agoCall cninit_finish() and set cold to 0 in configure_final() instead
Marcel Moolenaar [Sun, 29 May 2005 22:45:36 +0000 (22:45 +0000)]
Call cninit_finish() and set cold to 0 in configure_final() instead
of in configure(). Call cninit_finish() before setting cold to 0.
This is how it's done for other platforms. Be alike to avoid problems.

19 years agoo Call cninit_finish() in configure_final().
Marcel Moolenaar [Sun, 29 May 2005 22:42:27 +0000 (22:42 +0000)]
o  Call cninit_finish() in configure_final().
o  Remove unused and compiled-out code while here.

19 years agoMake bzip2 support working again after bzip2 upgrade. This time commit
Maxim Sobolev [Sun, 29 May 2005 21:56:38 +0000 (21:56 +0000)]
Make bzip2 support working again after bzip2 upgrade. This time commit
BZ_NO_COMPRESS support to the bzip2 sources directly (yes, this takes file
off the vendor branch, but looks like bzip2 maintainer doesn't care), so that
it will not be removed when the next upgrade is performed. Also, add a short
note on how to test bzip2 support.

Pointy hat to:  obrien

Correct comment (libz -> libbz2) and remove useless full path to zutil.h
while I am here.

19 years agoEliminate aio_bufjobs; it's unused.
Alan Cox [Sun, 29 May 2005 21:29:15 +0000 (21:29 +0000)]
Eliminate aio_bufjobs; it's unused.

19 years agoCatch up with s/dev_t/struct cdev */.
Simon L. B. Nielsen [Sun, 29 May 2005 21:23:50 +0000 (21:23 +0000)]
Catch up with s/dev_t/struct cdev */.

19 years agoWhite space normalization: use tabs instead of spaces before and after
Robert Watson [Sun, 29 May 2005 21:06:56 +0000 (21:06 +0000)]
White space normalization: use tabs instead of spaces before and after
the system call type field.

19 years agoAdd missed ${.CURDIR}/../../contrib/bzip2 into include search path, otherwise
Maxim Sobolev [Sun, 29 May 2005 21:05:58 +0000 (21:05 +0000)]
Add missed ${.CURDIR}/../../contrib/bzip2 into include search path, otherwise
old version of bzlib.h can be picked up from the /usr/include.

19 years agoNormalize white space in syscalls.master: try to use tabs before system
Robert Watson [Sun, 29 May 2005 20:20:16 +0000 (20:20 +0000)]
Normalize white space in syscalls.master: try to use tabs before system
call types.

19 years agoAdd a new field, sy_auevent, to the system call entry description
Robert Watson [Sun, 29 May 2005 20:08:39 +0000 (20:08 +0000)]
Add a new field, sy_auevent, to the system call entry description
structure, sysent.  This field will hold the default audit event
to generate when the system call is entered.  Currently, it will
default to 0 due to allocation in bss.

Submitted by: wsalamon
Obtained from: TrustedBSD Project