]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoThis commit was generated by cvs2svn to compensate for changes in r140229,
Ruslan Ermilov [Fri, 14 Jan 2005 09:50:45 +0000 (09:50 +0000)]
This commit was generated by cvs2svn to compensate for changes in r140229,
which included commits to RCS files with non-trunk default branches.

19 years agoPull up a vendor fix for a problem exposed by tr/tr.1,v 1.30.
Ruslan Ermilov [Fri, 14 Jan 2005 09:50:45 +0000 (09:50 +0000)]
Pull up a vendor fix for a problem exposed by tr/tr.1,v 1.30.

19 years agoMarkup fixes.
Ruslan Ermilov [Fri, 14 Jan 2005 09:40:13 +0000 (09:40 +0000)]
Markup fixes.

19 years agoMarkup fixes.
Ruslan Ermilov [Fri, 14 Jan 2005 09:28:17 +0000 (09:28 +0000)]
Markup fixes.

19 years agoFix off-by-one error.
Stefan Farfeleder [Fri, 14 Jan 2005 09:21:54 +0000 (09:21 +0000)]
Fix off-by-one error.

19 years agoTiny markup nits.
Ruslan Ermilov [Fri, 14 Jan 2005 09:12:05 +0000 (09:12 +0000)]
Tiny markup nits.

19 years agoo Clean up interface between ip_fw_chk() and its callers:
Gleb Smirnoff [Fri, 14 Jan 2005 09:00:46 +0000 (09:00 +0000)]
o Clean up interface between ip_fw_chk() and its callers:

- ip_fw_chk() returns action as function return value. Field retval is
  removed from args structure. Action is not flag any more. It is one
  of integer constants.
- Any action-specific cookies are returned either in new "cookie" field
  in args structure (dummynet, future netgraph glue), or in mbuf tag
  attached to packet (divert, tee, some future action).

o Convert parsing of return value from ip_fw_chk() in ipfw_check_{in,out}()
  to a switch structure, so that the functions are more readable, and a future
  actions can be added with less modifications.

Approved by: andre
MFC after: 2 months

19 years agoEliminate unused and constant arguments to smbfs_vinvalbuf()
Poul-Henning Kamp [Fri, 14 Jan 2005 08:52:55 +0000 (08:52 +0000)]
Eliminate unused and constant arguments to smbfs_vinvalbuf()

19 years agoEliminate constant and unused arguments to nwfs_vinvalbuf()
Poul-Henning Kamp [Fri, 14 Jan 2005 08:09:42 +0000 (08:09 +0000)]
Eliminate constant and unused arguments to nwfs_vinvalbuf()

19 years agoEliminate unused and unnecessary "cred" argument from vinvalbuf()
Poul-Henning Kamp [Fri, 14 Jan 2005 07:33:51 +0000 (07:33 +0000)]
Eliminate unused and unnecessary "cred" argument from vinvalbuf()

19 years agoMark all inline asms that read the floating-point control or status
David Schultz [Fri, 14 Jan 2005 07:09:23 +0000 (07:09 +0000)]
Mark all inline asms that read the floating-point control or status
registers as volatile.  Instructions that *wrote* to FP state were
already marked volatile, but apparently gcc has license to move
non-volatile asms past volatile asms.  This broke amd64's feupdateenv
at -O2 due to a WAR conflict between fnstsw and fldenv there.

19 years agopcic is no more on i386 port, so remove it from the hints.
Warner Losh [Fri, 14 Jan 2005 06:54:59 +0000 (06:54 +0000)]
pcic is no more on i386 port, so remove it from the hints.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r140216,
Nate Lawson [Fri, 14 Jan 2005 06:37:43 +0000 (06:37 +0000)]
This commit was generated by cvs2svn to compensate for changes in r140216,
which included commits to RCS files with non-trunk default branches.

19 years agoClean up a small memory leak introduced with the previous commit to
Nate Lawson [Fri, 14 Jan 2005 06:37:43 +0000 (06:37 +0000)]
Clean up a small memory leak introduced with the previous commit to
dsutils.c.  We need to GC the implicitly-returned object.

Submitted by: Robert.Moore at Intel
MFC after: 1 day

19 years agoMatch the LINUX32's style with existing style
David E. O'Brien [Fri, 14 Jan 2005 04:44:56 +0000 (04:44 +0000)]
Match the LINUX32's style with existing style
Submitted by: Jung-uk Kim <jkim@niksun.com>

Use positive, not negative logic.

19 years agoFix Linux compat 'uname -m' on AMD64.
David E. O'Brien [Fri, 14 Jan 2005 03:45:26 +0000 (03:45 +0000)]
Fix Linux compat 'uname -m' on AMD64.

Submitted by: Jung-uk Kim <jkim@niksun.com>
(patch reworked by me)

19 years agoFixed too many of "the", and enclose multi-word argument in double quotes.
Stefan Farfeleder [Thu, 13 Jan 2005 20:33:42 +0000 (20:33 +0000)]
Fixed too many of "the", and enclose multi-word argument in double quotes.

Obtained from: ru

19 years agoRemove duplicate code.
Poul-Henning Kamp [Thu, 13 Jan 2005 19:27:28 +0000 (19:27 +0000)]
Remove duplicate code.

19 years agoUse the standard FreeBSD license
Warner Losh [Thu, 13 Jan 2005 19:12:10 +0000 (19:12 +0000)]
Use the standard FreeBSD license

Approved by: imp, jon

19 years agoUse the standard FreeBSD license for these files.
Warner Losh [Thu, 13 Jan 2005 19:05:25 +0000 (19:05 +0000)]
Use the standard FreeBSD license for these files.

Approved by: imp, jon

19 years agoWhitespace in vop_vector{} initializations.
Poul-Henning Kamp [Thu, 13 Jan 2005 18:59:48 +0000 (18:59 +0000)]
Whitespace in vop_vector{} initializations.

19 years agoImport the subset of J.T. Conklin's single-precision x86-optimized
David Schultz [Thu, 13 Jan 2005 18:58:25 +0000 (18:58 +0000)]
Import the subset of J.T. Conklin's single-precision x86-optimized
math routines that appear to be (a) correct and (b) faster than their
MI counterparts on my Pentium 4.

Obtained from: NetBSD

19 years agoremember to reset nextgrp in getnetgrent(), so that subsequent calls to getnetgrent...
Jonathan Chen [Thu, 13 Jan 2005 18:33:01 +0000 (18:33 +0000)]
remember to reset nextgrp in getnetgrent(), so that subsequent calls to getnetgrent() doesn't return bogus information.

MFC after: 3 days

19 years ago"The ... function" and minor markup nits.
Ruslan Ermilov [Thu, 13 Jan 2005 16:46:49 +0000 (16:46 +0000)]
"The ... function" and minor markup nits.

19 years agoMarkup nits.
Ruslan Ermilov [Thu, 13 Jan 2005 16:37:26 +0000 (16:37 +0000)]
Markup nits.

19 years agoRemoved stray quotes.
Ruslan Ermilov [Thu, 13 Jan 2005 16:30:53 +0000 (16:30 +0000)]
Removed stray quotes.

19 years agoAllow the dragon and snake screen savers to be statically compiled into a
John Baldwin [Thu, 13 Jan 2005 15:55:34 +0000 (15:55 +0000)]
Allow the dragon and snake screen savers to be statically compiled into a
kernel and add them to NOTES.

MFC after: 2 weeks

19 years agoThe isnormal() in rev 1.2 should have been isfinite() so subnormals
David Schultz [Thu, 13 Jan 2005 15:43:41 +0000 (15:43 +0000)]
The isnormal() in rev 1.2 should have been isfinite() so subnormals
round correctly.

Noticed by: stefanf

19 years agoThings that are broken, unneeded, and unused since 1997 belong in the attic.
David Schultz [Thu, 13 Jan 2005 15:43:22 +0000 (15:43 +0000)]
Things that are broken, unneeded, and unused since 1997 belong in the attic.

19 years agoDon't do setuid checks on file systems mounted with noexec option.
Gleb Smirnoff [Thu, 13 Jan 2005 15:07:35 +0000 (15:07 +0000)]
Don't do setuid checks on file systems mounted with noexec option.

Reviewed by: brian, ru
MFC after: 1 week

19 years agoRemoved bitrot.
Ruslan Ermilov [Thu, 13 Jan 2005 13:34:00 +0000 (13:34 +0000)]
Removed bitrot.

19 years agoDitch vfs_object_create() and make the callers call VOP_CREATEVOBJECT()
Poul-Henning Kamp [Thu, 13 Jan 2005 12:25:19 +0000 (12:25 +0000)]
Ditch vfs_object_create() and make the callers call VOP_CREATEVOBJECT()
directly.

19 years agoImproved markup.
Ruslan Ermilov [Thu, 13 Jan 2005 11:22:29 +0000 (11:22 +0000)]
Improved markup.

19 years agoMarkup nits, and fix sentences to start from a capital letter.
Ruslan Ermilov [Thu, 13 Jan 2005 11:09:58 +0000 (11:09 +0000)]
Markup nits, and fix sentences to start from a capital letter.

19 years agoMarkup nits.
Ruslan Ermilov [Thu, 13 Jan 2005 10:43:01 +0000 (10:43 +0000)]
Markup nits.

19 years agoNits.
Ruslan Ermilov [Thu, 13 Jan 2005 10:10:34 +0000 (10:10 +0000)]
Nits.

19 years agoMarkup nits.
Ruslan Ermilov [Thu, 13 Jan 2005 10:07:51 +0000 (10:07 +0000)]
Markup nits.

19 years agoFixed too many of "the", and enclose multi-word argument in double quotes.
Ruslan Ermilov [Thu, 13 Jan 2005 09:35:47 +0000 (09:35 +0000)]
Fixed too many of "the", and enclose multi-word argument in double quotes.

19 years agoNits.
Ruslan Ermilov [Thu, 13 Jan 2005 09:33:06 +0000 (09:33 +0000)]
Nits.

19 years agoImplement and document ceill().
Stefan Farfeleder [Thu, 13 Jan 2005 09:11:41 +0000 (09:11 +0000)]
Implement and document ceill().

19 years agoBump .Dd for the last commit.
Stefan Farfeleder [Thu, 13 Jan 2005 09:08:16 +0000 (09:08 +0000)]
Bump .Dd for the last commit.

19 years agoRevert last delta. The ASCII code 0 character is "NUL",
Ruslan Ermilov [Thu, 13 Jan 2005 08:49:45 +0000 (08:49 +0000)]
Revert last delta.  The ASCII code 0 character is "NUL",
not "NULL", see ascii(7).

19 years agoRemoved harmful empty lines that crept in during the copyright
Ruslan Ermilov [Thu, 13 Jan 2005 08:46:19 +0000 (08:46 +0000)]
Removed harmful empty lines that crept in during the copyright
header update.

19 years agoSafer version of the chflags command.
Ruslan Ermilov [Thu, 13 Jan 2005 08:29:51 +0000 (08:29 +0000)]
Safer version of the chflags command.

19 years agoGet rid of the VDESC() macro while the pot is boiling anyway, it is
Poul-Henning Kamp [Thu, 13 Jan 2005 08:05:20 +0000 (08:05 +0000)]
Get rid of the VDESC() macro while the pot is boiling anyway, it is
only used from generate files now, so we might as well generate the
right stuff from the start.

19 years agoChange the generated VOP_ macro implementations to improve type checking
Poul-Henning Kamp [Thu, 13 Jan 2005 07:53:01 +0000 (07:53 +0000)]
Change the generated VOP_ macro implementations to improve type checking
and KASSERT coverage.

After this check there is only one "nasty" cast in this code but there
is a KASSERT to protect against the wrong argument structure behind
that cast.

Un-inlining the meat of VOP_FOO() saves 35kB of text segment on a typical
kernel with no change in performance.

We also now run the checking and tracing on VOP's which have been layered
by nullfs, umapfs, deadfs or unionfs.

    Add new (non-inline) VOP_FOO_AP() functions which take a "struct
    foo_args" argument and does everything the VOP_FOO() macros
    used to do with checks and debugging code.

    Add KASSERT to VOP_FOO_AP() check for argument type being
    correct.

    Slim down VOP_FOO() inline functions to just stuff arguments
    into the struct foo_args and call VOP_FOO_AP().

    Put function pointer to VOP_FOO_AP() into vop_foo_desc structure
    and make VCALL() use it instead of the current offsetoff() hack.

    Retire vcall() which implemented the offsetoff()

    Make deadfs and unionfs use VOP_FOO_AP() calls instead of
    VCALL(), we know which specific call we want already.

    Remove unneeded arguments to VCALL() in nullfs and umapfs bypass
    functions.

    Remove unused vdesc_offset and VOFFSET().

    Generally improve style/readability of the generated code.

19 years agoLocking and cleanup of tty netgraph node. Tty stack is Giant-locked,
Gleb Smirnoff [Thu, 13 Jan 2005 07:43:12 +0000 (07:43 +0000)]
Locking and cleanup of tty netgraph node. Tty stack is Giant-locked,
so we need to acquire Giant in netgraph methods, so that we don't
race with line discipline methods. Remove NET_NEEDS_GIANT.

- Packets coming into node from netgraph are queued in ifqueue
  attached to node private data.
- Mutex in struct ifqueue is used to lock not only the queue, but
  the whole private data, and tp->t_lsc field.
- tp->t_lsc pointer is used to indicate whether line discipline is
  attached to netgraph or not.
- Use FLG_DIE flag to indicate that node may be destroyed.
  (This protection doesn't work, and it didn't before. Must be redesigned.)
- Increment ngt_unit atomically, removing mutex.
- Acquire Giant, when executing ngt_start() from netgraph context.
- Acquire Giant, when {,de}registering line discipline.
- Uncomment forcing queue mode on peers hook, since this is reasonable.
- Force queue mode on our hook, to avoid acquiring Giant when coming from
  network stack. We may already hold some mutexes at this point.

Cleanups:
- Use callout_pending() instead of our own flag.
- Remove spl(9) calls. Now we can use return() instead of ERROUT().

style(9):
- Sort includes.
- Sparse initializer for struct linesw.
- Remove some empty lines, sort declarations.

Reviewed by: julian, phk
MFC after: 1 month

19 years agoWANRS=6 cleanup for vidcontrol(1):
Xin LI [Thu, 13 Jan 2005 03:59:44 +0000 (03:59 +0000)]
WANRS=6 cleanup for vidcontrol(1):
- Use foo(void) instead of foo().
- Use static where applicable.
- Apply more const's when passing parameters
- signed/unsigned madness
- Avoid namespace collision by adding underscores.
- For 64-bit architectures, use %zx instead of %x
  when necessary.
- When storing constants, use const instead of
  variable.
- Bump WARNS?= from 2 to 6

19 years agoImprove previous commit with regards to the user check.
Tom Rhodes [Thu, 13 Jan 2005 00:21:38 +0000 (00:21 +0000)]
Improve previous commit with regards to the user check.

Submitted by: nectar

19 years agoHook up and document floorl().
Stefan Farfeleder [Wed, 12 Jan 2005 22:16:26 +0000 (22:16 +0000)]
Hook up and document floorl().

19 years agoImplement floorl().
Stefan Farfeleder [Wed, 12 Jan 2005 22:10:46 +0000 (22:10 +0000)]
Implement floorl().

19 years agoWhitespace nit.
Stefan Farfeleder [Wed, 12 Jan 2005 22:05:41 +0000 (22:05 +0000)]
Whitespace nit.

19 years agoScheduled mdoc(7) sweep.
Ruslan Ermilov [Wed, 12 Jan 2005 21:48:25 +0000 (21:48 +0000)]
Scheduled mdoc(7) sweep.

19 years agoA few more mdoc fixes I missed in the suggestions of Ruslan.
Giorgos Keramidas [Wed, 12 Jan 2005 21:43:09 +0000 (21:43 +0000)]
A few more mdoc fixes I missed in the suggestions of Ruslan.

19 years agoFix a TCP SACK related crash resulting from incorrect computation
Paul Saab [Wed, 12 Jan 2005 21:40:51 +0000 (21:40 +0000)]
Fix a TCP SACK related crash resulting from incorrect computation
of len in tcp_output(), in the case where the FIN has already been
transmitted. The mis-computation of len is because of a gcc
optimization issue, which this change works around.

Submitted by: Mohan Srinivasan

19 years agoReflect reality when a sudo user builds/installs a kernel. [1]
Tom Rhodes [Wed, 12 Jan 2005 21:28:25 +0000 (21:28 +0000)]
Reflect reality when a sudo user builds/installs a kernel. [1]
/* -> /*- for copyright notices. [2]

[1]:
PR: 41317
Submitted by: marck (original version)

[2]:
Discussed with: imp

19 years agoCross-reference the newly added example.9
Giorgos Keramidas [Wed, 12 Jan 2005 20:56:15 +0000 (20:56 +0000)]
Cross-reference the newly added example.9

19 years agoLink mdoc/example.9 to the build.
Giorgos Keramidas [Wed, 12 Jan 2005 20:55:01 +0000 (20:55 +0000)]
Link mdoc/example.9 to the build.

19 years agoAdd a sample manpage for section 9.
Giorgos Keramidas [Wed, 12 Jan 2005 20:49:25 +0000 (20:49 +0000)]
Add a sample manpage for section 9.

Suggested by: simon
Reviewed by: ru

19 years agoFix typo %d -> %x
Maksim Yevmenkin [Wed, 12 Jan 2005 20:06:16 +0000 (20:06 +0000)]
Fix typo %d -> %x

That should fix the problem with invalid PSM returned from bthidcontrol.
Pointy hat goes to me.

PR: misc/76107
Submitted by: Hiroyuki Aizu < aizu at navi dot org >
MFC after: 1 day

19 years agoTry harder to work with MP table interrupt entries that claim that an
John Baldwin [Wed, 12 Jan 2005 18:24:32 +0000 (18:24 +0000)]
Try harder to work with MP table interrupt entries that claim that an
interrupt is wired up to all the I/O APICs in the system.  If the system
has only one I/O APIC, then just act as if the entry specified that APIC.
We still don't try to handle global entries in a system with multiple I/O
APICs.

Tested by: Peter Trifonov pvtrifonov at mail dot ru
MFC after: 1 week

19 years agoInclude opt_bootp.h for BOOTP_NFSROOT
Brian Somers [Wed, 12 Jan 2005 12:42:46 +0000 (12:42 +0000)]
Include opt_bootp.h for BOOTP_NFSROOT

PR: 73183
Submitted by: Darrin Smith sdar at salseast dot org
MFC after: 7 days

19 years agoSort xrefs.
Ruslan Ermilov [Wed, 12 Jan 2005 10:22:58 +0000 (10:22 +0000)]
Sort xrefs.

19 years agoWhen re-connecting already connected datagram socket ensure to clean
Maxim Sobolev [Wed, 12 Jan 2005 10:15:23 +0000 (10:15 +0000)]
When re-connecting already connected datagram socket ensure to clean
up its pending error state, which may be set in some rare conditions resulting
in connect() syscall returning that bogus error and making application believe
that attempt to change association has failed, while it has not in fact.

There is sockets/reconnect regression test which excersises this bug.

MFC after: 2 weeks

19 years agoScheduled mdoc(7) sweep.
Ruslan Ermilov [Wed, 12 Jan 2005 10:14:43 +0000 (10:14 +0000)]
Scheduled mdoc(7) sweep.

19 years agoComment out debugging printf which doesn't compile on amd64.
Poul-Henning Kamp [Wed, 12 Jan 2005 10:11:31 +0000 (10:11 +0000)]
Comment out debugging printf which doesn't compile on amd64.

19 years agoAdd test which excersises problem with unability to change association of
Maxim Sobolev [Wed, 12 Jan 2005 09:57:18 +0000 (09:57 +0000)]
Add test which excersises problem with unability to change association of
already associated datagram unix domain socket by issuing connect() system
call.

19 years agoCreate three additional X socket directories. Using X applications when another
Eric Anholt [Wed, 12 Jan 2005 07:18:25 +0000 (07:18 +0000)]
Create three additional X socket directories.  Using X applications when another
user owns these directories or the sticky bit is unset may open security holes,
so simply create them at startup with the correct owner/mode.

MFC after: 1 day

19 years agoUse off_t when passing and calculating file offsets. While a single
Scott Long [Wed, 12 Jan 2005 06:42:13 +0000 (06:42 +0000)]
Use off_t when passing and calculating file offsets.  While a single
extent in UDF is only 32 bits, multiple extents can exist in a file.
Also clean up some minor whitespace problems.

Submitted by: John Wehle

19 years agoDon't allow reads past the end of a file.
Scott Long [Wed, 12 Jan 2005 06:17:01 +0000 (06:17 +0000)]
Don't allow reads past the end of a file.

Submitted by: John Wehle, Andriy Gapon
MFC After: 3 days

19 years agoBring some file descriptions in line with reality.
Tim Kientzle [Wed, 12 Jan 2005 05:56:49 +0000 (05:56 +0000)]
Bring some file descriptions in line with reality.

Also, add some hints to help people understand which
functions are most appropriate for typical uses.

19 years agoLet _umtx_op directly return error code rather than from errno because
David Xu [Wed, 12 Jan 2005 05:55:52 +0000 (05:55 +0000)]
Let _umtx_op directly return error code rather than from errno because
errno can be tampered potentially by nested signal handle.
Now all error codes are returned in negative value, positive value are
reserved for future expansion.

19 years agoDon't reprint file names unnecessarily.
Brian Somers [Wed, 12 Jan 2005 04:06:31 +0000 (04:06 +0000)]
Don't reprint file names unnecessarily.

PR: 75028
Submitted by: mteterin at 250-217 dot customer dot cloud9 dot net
MFC after: 7 days

19 years agoFix some signed/unsigned comparisons. Fix prototypes while I'm here.
Brian Somers [Wed, 12 Jan 2005 03:39:34 +0000 (03:39 +0000)]
Fix some signed/unsigned comparisons.  Fix prototypes while I'm here.

PR: 28890
Submitted by: matthias.andree at web dot de
MFC after: 7 days

19 years agoDon't decend into directories when -d is set and the directory itself doesn't
Brian Somers [Wed, 12 Jan 2005 03:25:55 +0000 (03:25 +0000)]
Don't decend into directories when -d is set and the directory itself doesn't
match.

PR: 28095
Submitted by: bill at twwells dot com
MFC after: 7 days

19 years agoSed doesn't grok '[ \t]' -- it doesn't expand the \t :(
Brian Somers [Wed, 12 Jan 2005 01:31:21 +0000 (01:31 +0000)]
Sed doesn't grok '[ \t]' -- it doesn't expand the \t :(

As there are no tabs in maillog, reduce the expression so that only spaces
are used.

Problem raised by:      Leif Neland root at internet dot dk

19 years agoFix handling of the implicit return case for methods called from an
Nate Lawson [Wed, 12 Jan 2005 00:52:40 +0000 (00:52 +0000)]
Fix handling of the implicit return case for methods called from an
external source (i.e., _STA).  The previous case only handled calls
occurring within AML.  This should fix Toshibas, among others.  Thanks
to Robert Moore of Intel for the fix.

MFC after: 2 days

19 years agoThis commit was generated by cvs2svn to compensate for changes in r140094,
Nate Lawson [Wed, 12 Jan 2005 00:52:40 +0000 (00:52 +0000)]
This commit was generated by cvs2svn to compensate for changes in r140094,
which included commits to RCS files with non-trunk default branches.

19 years ago- list the Hama USB RS-232 Serial Adapter as supported
Christian Brueffer [Tue, 11 Jan 2005 23:16:20 +0000 (23:16 +0000)]
- list the Hama USB RS-232 Serial Adapter as supported
- bump .Dd

PR: 76088
Submitted by: Benjamin Lutz <benlutz@datacomm.ch>
MFC after: 3 days

19 years agoRegression tests for [l]lrint[f]() and [l]lround[f]().
David Schultz [Tue, 11 Jan 2005 23:13:36 +0000 (23:13 +0000)]
Regression tests for [l]lrint[f]() and [l]lround[f]().

19 years agoAdd MI implementations of [l]lrint[f]() and [l]lround[f]().
David Schultz [Tue, 11 Jan 2005 23:12:55 +0000 (23:12 +0000)]
Add MI implementations of [l]lrint[f]() and [l]lround[f]().

Discussed with: bde

19 years agoDocument [l]lrint[f]() and [l]lround[f]().
David Schultz [Tue, 11 Jan 2005 23:12:17 +0000 (23:12 +0000)]
Document [l]lrint[f]() and [l]lround[f]().

19 years agoFaster lrint() and llrint() implementations for x86.
David Schultz [Tue, 11 Jan 2005 23:10:53 +0000 (23:10 +0000)]
Faster lrint() and llrint() implementations for x86.

19 years agoMark inline stmxcsr instructions as volatile, since this appears to be
David Schultz [Tue, 11 Jan 2005 22:10:43 +0000 (22:10 +0000)]
Mark inline stmxcsr instructions as volatile, since this appears to be
the only way to convince gcc that they read the MXCSR.  The volatile
annotation may be needed elsewhere as well.

19 years agoAdd missing entry.
Pawel Jakub Dawidek [Tue, 11 Jan 2005 21:15:18 +0000 (21:15 +0000)]
Add missing entry.

Reported by: sos

19 years agoList RFCs under SEE ALSO.
Tom Rhodes [Tue, 11 Jan 2005 21:08:39 +0000 (21:08 +0000)]
List RFCs under SEE ALSO.

PR: 46918

19 years agoDocument that MODULE_DEPEND does not change runtime initialization
Simon L. B. Nielsen [Tue, 11 Jan 2005 20:56:03 +0000 (20:56 +0000)]
Document that MODULE_DEPEND does not change runtime initialization
order.

Mostly submitted by: scottl
Discussed on: cvs-src/src-committers

19 years agoScheduled mdoc(7) sweep.
Ruslan Ermilov [Tue, 11 Jan 2005 20:50:51 +0000 (20:50 +0000)]
Scheduled mdoc(7) sweep.

19 years agoSanitize the markup, as prompted.
Ruslan Ermilov [Tue, 11 Jan 2005 20:16:03 +0000 (20:16 +0000)]
Sanitize the markup, as prompted.

19 years agoGet the vnode from file.f_vnode instead of file.f_data.
Suleiman Souhlal [Tue, 11 Jan 2005 18:52:12 +0000 (18:52 +0000)]
Get the vnode from file.f_vnode instead of file.f_data.

Nowadays, f_data points to the vnode only if the underlying filesystem
doesn't use it for other purposes (devfs uses it to store the cdev,
for example).

Found by: csjp
Reviewed by: csjp
Approved by: phk, wes, grehan (mentor)
MFC after: 1 week

19 years agoRemnant code (broken, unhooked) from a.out which I missed.
Tom Rhodes [Tue, 11 Jan 2005 18:35:40 +0000 (18:35 +0000)]
Remnant code (broken, unhooked) from a.out which I missed.

Noticed by: ru

19 years agoConnect SHSEC GEOM class to the build.
Pawel Jakub Dawidek [Tue, 11 Jan 2005 18:18:40 +0000 (18:18 +0000)]
Connect SHSEC GEOM class to the build.

19 years agoIntroduce a new GEOM class - SHSEC. It provides sharing secret between
Pawel Jakub Dawidek [Tue, 11 Jan 2005 18:06:44 +0000 (18:06 +0000)]
Introduce a new GEOM class - SHSEC. It provides sharing secret between
the given providers. Without even one of the configured components there
should be no way to get the secret.

Supported by: WHEEL Sp. z o.o.
http://www.wheel.pl

19 years agoDocument the -U option a bit more clearly.
Philip Paeps [Tue, 11 Jan 2005 17:16:31 +0000 (17:16 +0000)]
Document the -U option a bit more clearly.

Submitted by: keramida

19 years agoRestore these files to make ldconfig(8) happy.
Tom Rhodes [Tue, 11 Jan 2005 16:40:29 +0000 (16:40 +0000)]
Restore these files to make ldconfig(8) happy.

19 years agoWhen grabbing registers for an lwp, fake the inferior's pid using the
Peter Edwards [Tue, 11 Jan 2005 14:53:16 +0000 (14:53 +0000)]
When grabbing registers for an lwp, fake the inferior's pid using the
lwp ID before invoking the underlying target operation.

For corefiles, we rely on gdb internals to do this, and it uses the
pid as an index, rather than the lwpid, so previously, backtraces
for multithreaded core files wasn't working correctly. For processes,
we currently use ptrace directly, so fixup that code to also use
the pid directly.

Discussed With: marcel, davidxu
MFC After: 4 days

19 years agoMake WARNS=6 happy with our init(8):
Xin LI [Tue, 11 Jan 2005 14:34:29 +0000 (14:34 +0000)]
Make WARNS=6 happy with our init(8):
- Use more ``const''s where suitable.
- Define strk() as a static function in global scope.
  This avoids the "nested extern declaration" warnings.
- Use static initialization of strings, rather than
  referring string constants through char *.
- Bump WARNS from 0 to 6.

19 years agoFix incorrect comment.
Pawel Jakub Dawidek [Tue, 11 Jan 2005 14:11:14 +0000 (14:11 +0000)]
Fix incorrect comment.

MFC after: 3 days

19 years agoUtilize callout_pending() macro
Gleb Smirnoff [Tue, 11 Jan 2005 12:20:28 +0000 (12:20 +0000)]
Utilize callout_pending() macro

19 years agoSilently ignore forced argument to unmount.
Poul-Henning Kamp [Tue, 11 Jan 2005 12:02:26 +0000 (12:02 +0000)]
Silently ignore forced argument to unmount.