]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoAlign the stack to a 16 byte boundary so that we can safely call functions
dfr [Thu, 19 May 2005 07:36:07 +0000 (07:36 +0000)]
Align the stack to a 16 byte boundary so that we can safely call functions
that use SSE. The compiler does attempt to do this in main() but not very
successfully - it still manages to use unaligned offsets from %ebp in some
cases. Also we need to have an aligned stack in case something uses SSE
via _init().

MFC After: 1 week

19 years agoAlign the stack to a 16 byte boundary before calling _rtld so that we can
dfr [Thu, 19 May 2005 07:32:42 +0000 (07:32 +0000)]
Align the stack to a 16 byte boundary before calling _rtld so that we can
safely initialise shared libraries that use SSE in their init sections.

MFC After: 1 week

19 years agoKeep the stack aligned to a 16 byte boundary when calling init functions
dfr [Thu, 19 May 2005 07:31:06 +0000 (07:31 +0000)]
Keep the stack aligned to a 16 byte boundary when calling init functions
so that we don't cause a bus error if they start storing SSE math stuff
on the stack.

MFC After: 1 week

19 years agoChange ofw_readin/ofw_copyin to map the entire region before
grehan [Thu, 19 May 2005 07:21:46 +0000 (07:21 +0000)]
Change ofw_readin/ofw_copyin to map the entire region before
copying, rather than a page at a time. This was creating far
too many single-page mappings, and eventually OFW overflowed
some internal data structure and refused to map any more.
The new algorithm creates far less mappings and fixed a bug
where multiple mappings for the same page would be created.

'Twas known this was a problem, but only became urgent when the
install CD's mfs_root grew large enough to cause the overflow.

19 years agoRemove calls to spl*().
alc [Thu, 19 May 2005 06:11:13 +0000 (06:11 +0000)]
Remove calls to spl*().

19 years agoFix low res profiling kernel build. Move two defines to collapse the
njl [Thu, 19 May 2005 05:22:52 +0000 (05:22 +0000)]
Fix low res profiling kernel build.  Move two defines to collapse the
#ifdef GUPROF case.

19 years agoWhitespace only: kill trailing spaces.
sobomax [Thu, 19 May 2005 05:15:08 +0000 (05:15 +0000)]
Whitespace only: kill trailing spaces.

19 years agoFix some of the things I broke so that the SMC2602W (AMD Am1772) driver
wpaul [Thu, 19 May 2005 04:44:26 +0000 (04:44 +0000)]
Fix some of the things I broke so that the SMC2602W (AMD Am1772) driver
works again.

This driver uses NdisScheduleWorkItem(), and we have to take special steps
to insure that its workitems don't collide with any of the other workitems
used by the NDISulator. In particular, if one of the driver's work jobs
blocks, it can prevent NdisMAllocateSharedMemoryAsync() from completing
when expected.

The original hack to fix this was to have NdisMAllocateSharedMemoryAsync()
defer its work to the DPC queue instead of the general task queue. To
fix it now, I decided to add some additional workitem threads. (There's
supposed to be a pool of worker threads in Windows anyway.) Currently,
there are 4. There should be at least 2. One is reserved for the legacy
ExQueueWorkItem() API, while the others are used in round-robin by the
IoQueueWorkItem() API. NdisMAllocateSharedMemoryAsync() uses the latter
API while NdisScheduleWorkItem() uses the former, so the deadlock is
avoided.

Fixed NdisMRegisterDevice()/NdisMDeregisterDevice() to work a little
more sensibly with the new driver_object/device_object framework. It
doesn't really register a working user-mode interface, but the existing
code was completely wrong for the new framework.

Fixed a couple of bugs dealing with the cancellation of events and
DPCs. When cancelling an event that's still on the timer queue (i.e.
hasn't expired yet), reset dh_inserted in its dispatch header to FALSE.
Previously, it was left set to TRUE, which would make a cancelled
timer appear to have not been cancelled. Also, when removing a DPC
from a queue, reset its list pointers, otherwise a cancelled DPC
might mistakenly be treated as still pending.

Lastly, fix the behavior of ntoskrnl_wakeup() when dealing with
objects that have nobody waiting on them: sync event objects get
their signalled state reset to FALSE, but notification objects
should still be set to TRUE.

19 years agoRemove a stale comment concerning spl* usage.
alc [Thu, 19 May 2005 03:53:07 +0000 (03:53 +0000)]
Remove a stale comment concerning spl* usage.

19 years agoFix a bug that caused preemption to happen for a thread in the same
ups [Thu, 19 May 2005 01:08:30 +0000 (01:08 +0000)]
Fix a bug that caused preemption to happen for a thread in the same
ksegrp with the same priority as the currently running thread.
This can cause propagate_priority() to panic.

Pointy hat to: ups

19 years agoFix problem with session termination. bthidd(8) maintains two L2CAP
emax [Wed, 18 May 2005 23:03:44 +0000 (23:03 +0000)]
Fix problem with session termination. bthidd(8) maintains two L2CAP
connections to Bluetooth HID device. As soon as Bluetooth HID device
is powered off (or goes out of RF range) the stack will terminate both
connections. File descriptors for both connections will become active
on next select(2) call. Because bthidd(8) processes file descriptors
in order, it will detect descriptor for one of the closed connections
first and kill the session. However, there is still a second (active)
descriptor that used to point to the same session. bthidd(8) used to
assert() if it cant find session by file descriptor, which was wrong.

While I'm here fix a couple of typos in parser.y

Reported by: Eric Anderson anderson AT centtech DOT com
MFC after: 3 days

19 years agoAllow removal of empty directories with high link counts. These can
mckusick [Wed, 18 May 2005 22:18:21 +0000 (22:18 +0000)]
Allow removal of empty directories with high link counts. These can
occur on a filesystem running with soft updates after a crash and
before a background fsck has been run. To prevent discrepancies
from arising in a background fsck that may already be running,
the directory is removed but its inode is not freed and is left
with the residual reference count. When encountered by the
background fsck it will be reclaimed.

19 years agoUpdate some comments to reflect the change from spl-based to lock-based
alc [Wed, 18 May 2005 22:08:52 +0000 (22:08 +0000)]
Update some comments to reflect the change from spl-based to lock-based
synchronization.

19 years agodevfs_first() return value isn't used, remove it.
pjd [Wed, 18 May 2005 22:05:12 +0000 (22:05 +0000)]
devfs_first() return value isn't used, remove it.

19 years agoCorrect typo.
pjd [Wed, 18 May 2005 21:53:08 +0000 (21:53 +0000)]
Correct typo.

19 years agoRemove calls to spl*().
alc [Wed, 18 May 2005 20:45:33 +0000 (20:45 +0000)]
Remove calls to spl*().

19 years agoRevert revision 1.270: swp_pager_async_iodone() need not perform
alc [Wed, 18 May 2005 17:48:04 +0000 (17:48 +0000)]
Revert revision 1.270: swp_pager_async_iodone() need not perform
VM_LOCK_GIANT().

Discussed with: jeff

19 years agoThe correct copyright notice for manpages that copy material from Open
keramida [Wed, 18 May 2005 17:08:15 +0000 (17:08 +0000)]
The correct copyright notice for manpages that copy material from Open
Group's documentation is `/usr/share/examples/mdoc/POSIX-copyright',
not the one I copied from `/usr/share/examples/etc/bsd-copyright'.

Suggested by: simon
2nd pointy hat of the day: yours truly

19 years agoAdd a manpage for pthread_atfork(3). This copies a lot of the text of
keramida [Wed, 18 May 2005 16:23:11 +0000 (16:23 +0000)]
Add a manpage for pthread_atfork(3).  This copies a lot of the text of
the Open Group manpage for pthread_atfork(), available online at:

http://www.opengroup.org/onlinepubs/009695399/functions/pthread_atfork.html

which should be ok, since Daniel Eischen had mailed me about Open
Group manpages and the fact that they have granted permission to
FreeBSD to use their material.  Any differences from the OG text are
my changes to the original manpage text submitted by Alex Vasylenko:

- In an effort to clean up the part that describes hooks and their
  calling order, I used a list instead of a single paragraph for all the three
  types of fork() hooks.
- After a short discussion with Dima Dorfman a long long time ago in a
  far away galaxy, I changed the RETURN VALUES section to look more
  like the rest of the pthread_xxx.3 manpages.

PR: docs/68201
Submitted by: Alex Vasylenko <lxv@omut.org>

19 years agoGet rid of global variables for argument vectors produced by brk_string()
harti [Wed, 18 May 2005 14:50:35 +0000 (14:50 +0000)]
Get rid of global variables for argument vectors produced by brk_string()
introduce a struct that holds all the information about an argument
vector and pass that around.

Author: Max Okumoto <okumoto@ucsd.edu>
Obtained from: DragonFlyBSD

19 years agoProperly trim the header line too, to fix wrap-around problems that
keramida [Wed, 18 May 2005 13:48:33 +0000 (13:48 +0000)]
Properly trim the header line too, to fix wrap-around problems that
have been noticed by running top(1) in terminals that are too narrow
(or on systems with usernames that were too long, pushing everything
too far to the right).

Note that this does *not* solve the wrap-around problem of the system
statistics, which is an entirely different matter :-/

Tested on: i386, sparc64 (panther), amd64 (sledge)
Approved by: davidxu (in principle)

19 years ago(1) Revert unnecessary indentation changes I committed as part of the
keramida [Wed, 18 May 2005 13:42:51 +0000 (13:42 +0000)]
(1) Revert unnecessary indentation changes I committed as part of the
last version and (2) remove a disabled debugging fprintf() that I
accidentally committed here.

Noticed by: simon (2)

19 years agoMerge the CPU and WCPU columns in a single %6.2f column, add a new 'C'
keramida [Wed, 18 May 2005 13:30:08 +0000 (13:30 +0000)]
Merge the CPU and WCPU columns in a single %6.2f column, add a new 'C'
command that toggles between the two and update the ORDER_PCTCPU()
macro to sort correctly by the visible "cpu" value.

This saves 6 more columns in 80-column terminals, making things a lot
better for the COMMAND column.

Tested on: i386, sparc64 (panther), amd64 (sledge)
Approved by: davidxu (in principle)

19 years agoSwitch to WARNS=2 instead of tinkering directly with CFLAGS (level 2 was
keramida [Wed, 18 May 2005 12:19:50 +0000 (12:19 +0000)]
Switch to WARNS=2 instead of tinkering directly with CFLAGS (level 2 was
picked because higher warn levels are broken for ftp-proxy at the moment).

Approved by: mlaier

19 years agoCorrect 32 vs 64 bit signedness issues.
bz [Wed, 18 May 2005 08:57:31 +0000 (08:57 +0000)]
Correct 32 vs 64 bit signedness issues.

Approved by: pjd (mentor)
MFC after: 2 weeks

19 years agoConvert to the faster bus_dmamap_load_mbuf_sg() interface.
alc [Wed, 18 May 2005 07:45:08 +0000 (07:45 +0000)]
Convert to the faster bus_dmamap_load_mbuf_sg() interface.

MFC after: 1 week

19 years agoGet rid of the ReturnStatus obscuration that was anyway used only
harti [Wed, 18 May 2005 06:50:39 +0000 (06:50 +0000)]
Get rid of the ReturnStatus obscuration that was anyway used only
in two places. While here don't bother returning anything from
Lst_Replace - nobody ever checks the return code.

Suggested by: jmallet

19 years agoSupport passthru ioctl commands from 32bit binaries.
ps [Wed, 18 May 2005 05:31:34 +0000 (05:31 +0000)]
Support passthru ioctl commands from 32bit binaries.

19 years agoUnbreak r1.11 by delhij. Blind change from unsigned to signed char
kan [Wed, 18 May 2005 05:24:08 +0000 (05:24 +0000)]
Unbreak r1.11 by delhij. Blind change from unsigned to signed char
pointer causes problems with sign extension and leads to compress(1)
dumping core on any non-trivial incoming data.

19 years agoUgh. Previous commit got the logic exactly backward.
des [Tue, 17 May 2005 18:23:03 +0000 (18:23 +0000)]
Ugh.  Previous commit got the logic exactly backward.

Submitted by: bland
Pointy hat to: des

19 years agoAdd manpages for devfs.conf and devfs.rules. These have been
keramida [Tue, 17 May 2005 17:52:27 +0000 (17:52 +0000)]
Add manpages for devfs.conf and devfs.rules.  These have been
initially written by Roland, but hacked for a while by me.  Any
good parts are the results of Roland's hard work.  Any typos or
style mistakes are mine.

Submitted by: Roland Smith <rsmith@xs4all.nl>
PR: docs/63808, docs/75433, docs/80458, docs/80459
MFC after: 2 weeks

19 years agoClean up an additional file.
obrien [Tue, 17 May 2005 17:48:26 +0000 (17:48 +0000)]
Clean up an additional file.

19 years agoDon't use a patch w/in /usr/src. Programmatically change files when needed.
obrien [Tue, 17 May 2005 17:46:29 +0000 (17:46 +0000)]
Don't use a patch w/in /usr/src.  Programmatically change files when needed.

19 years agoWhen a drive dies, don't call g_wither_geom() directly, but instead
le [Tue, 17 May 2005 16:38:30 +0000 (16:38 +0000)]
When a drive dies, don't call g_wither_geom() directly, but instead
post an event to the geom event queue that will take care of it,
letting outstanding bios finish, and closing the consumers.

Plus some cosmetic clean ups.

19 years agoRe-enable support for bzip2'ed compressed filesystems.
obrien [Tue, 17 May 2005 16:22:54 +0000 (16:22 +0000)]
Re-enable support for bzip2'ed compressed filesystems.

19 years agoClarify the header.
obrien [Tue, 17 May 2005 15:03:50 +0000 (15:03 +0000)]
Clarify the header.

19 years agoAdd temporary patches to make one-true-awk respect locale's collating order
ru [Tue, 17 May 2005 14:54:33 +0000 (14:54 +0000)]
Add temporary patches to make one-true-awk respect locale's collating order
in [a-z] bracket expressions, until a more complete fix (like handing BREs)
is ready.

Prodded by: ache
OK'ed by: tjr

19 years agoFix the contents of the underneath .depend files and "make checkdpadd".
ru [Tue, 17 May 2005 14:29:06 +0000 (14:29 +0000)]
Fix the contents of the underneath .depend files and "make checkdpadd".

19 years agoStrip the initial world some more.
ru [Tue, 17 May 2005 14:24:33 +0000 (14:24 +0000)]
Strip the initial world some more.

19 years agoRecover the original Berkeley RCS id, and fix the description of
ru [Tue, 17 May 2005 14:20:42 +0000 (14:20 +0000)]
Recover the original Berkeley RCS id, and fix the description of
format for source files when it comes to $FreeBSD$.

19 years agoMake certain the the 48bit flag is reset if we dont translate LBA.
sos [Tue, 17 May 2005 12:31:54 +0000 (12:31 +0000)]
Make certain the the 48bit flag is reset if we dont translate LBA.

19 years ago- Unwind NG_SEND_MSG_PATH() macro and merge it with already unwinded
glebius [Tue, 17 May 2005 12:18:13 +0000 (12:18 +0000)]
- Unwind NG_SEND_MSG_PATH() macro and merge it with already unwinded
  version under TRACE_MESSAGES.
- Pass NG_WAITOK flag to ng_package_data() in unwinded macro.

19 years agoRevision 1.173 broke updating a mount from ro to rw. Fix that by clearing
des [Tue, 17 May 2005 12:00:43 +0000 (12:00 +0000)]
Revision 1.173 broke updating a mount from ro to rw.  Fix that by clearing
the MNT_RDONLY flag if MNT_UPDATE is set and "ro" was not specified.

Suggested by: cognet

19 years agoCorrect the script name to be "zless.sh". This avoids an extra cp/chmod
ru [Tue, 17 May 2005 11:41:01 +0000 (11:41 +0000)]
Correct the script name to be "zless.sh".  This avoids an extra cp/chmod
(from the sys.mk rule) while still installing it under the "zless" name.

19 years agoAdd a zless script which invokes less using the lesspipe script. This is
des [Tue, 17 May 2005 11:14:11 +0000 (11:14 +0000)]
Add a zless script which invokes less using the lesspipe script.  This is
similar to the zmore script that comes with gzip (and in fact, in most
Linux distros, zless is a symlink to that very same zmore script) but has
the advantage that you get the correct file name on the less status line,
and can use :n and :p to navigate back and forth between multiple files.

MFC after: 1 week

19 years agoHandle presumably uncompressed files with cat; exec the handlers.
des [Tue, 17 May 2005 11:08:11 +0000 (11:08 +0000)]
Handle presumably uncompressed files with cat; exec the handlers.

MFC after: 1 week

19 years agoCorrect script path.
jcamou [Tue, 17 May 2005 03:54:52 +0000 (03:54 +0000)]
Correct script path.

PR: docs/80948
Submitted by: anonymous coward
Approved by: trhodes (mentor)
MFC after: 2 days

19 years agoThis commit was generated by cvs2svn to compensate for changes in r146309,
mikeh [Tue, 17 May 2005 03:11:29 +0000 (03:11 +0000)]
This commit was generated by cvs2svn to compensate for changes in r146309,
which included commits to RCS files with non-trunk default branches.

19 years agoImport latest NetBSD version...changes:
mikeh [Tue, 17 May 2005 03:11:29 +0000 (03:11 +0000)]
Import latest NetBSD version...changes:

 * Whitespace and gcc 4.x fixes.
 * Fix deallocation bug in fetch.c.
 * Implement timeouts for accept(2) and connect(2),
   which accept timeout values with '-q quittime'.
   (intended to help with active mode through firewalls)
 * Fix bin/77158 by Ryoji Kanai <rkanai@eeye.com>.

19 years agoTemporarily disable support for bzip2'ed compressed filesystems, until a
obrien [Tue, 17 May 2005 01:44:37 +0000 (01:44 +0000)]
Temporarily disable support for bzip2'ed compressed filesystems, until a
maintainable why of handling them is created.

19 years agoRemove a duplicate 'comment' to fix the syntax of a sentence.
keramida [Tue, 17 May 2005 01:25:57 +0000 (01:25 +0000)]
Remove a duplicate 'comment' to fix the syntax of a sentence.

19 years agoIgnore sigpipe so we can properly detach clients to the pipe.
imp [Mon, 16 May 2005 20:51:46 +0000 (20:51 +0000)]
Ignore sigpipe so we can properly detach clients to the pipe.

Submitted by: Fredrik Lindberg

19 years agoUse NG_QUEUE instead of constant.
glebius [Mon, 16 May 2005 19:52:42 +0000 (19:52 +0000)]
Use NG_QUEUE instead of constant.

19 years agoIntroduce routines to alloc/free sack holes. This cleans up the code
ps [Mon, 16 May 2005 19:26:46 +0000 (19:26 +0000)]
Introduce routines to alloc/free sack holes. This cleans up the code
considerably.

Submitted by:   Noritoshi Demizu.
Reviewed by:    Raja Mukerji, Mohan Srinivasan.

19 years agoUpdate for the 2005/04/24 import.
ru [Mon, 16 May 2005 19:19:32 +0000 (19:19 +0000)]
Update for the 2005/04/24 import.

19 years agoRefactor places where ng_package_data() failed.
glebius [Mon, 16 May 2005 19:12:57 +0000 (19:12 +0000)]
Refactor places where ng_package_data() failed.

Suggested & reviewed by: archie

19 years agoThis commit was generated by cvs2svn to compensate for changes in r146299,
ru [Mon, 16 May 2005 19:11:36 +0000 (19:11 +0000)]
This commit was generated by cvs2svn to compensate for changes in r146299,
which included commits to RCS files with non-trunk default branches.

19 years agoVendor import of bwk's 24-Apr-2005 release.
ru [Mon, 16 May 2005 19:11:36 +0000 (19:11 +0000)]
Vendor import of bwk's 24-Apr-2005 release.

19 years agoUpdate to match the 1.0.3 import.
obrien [Mon, 16 May 2005 18:56:30 +0000 (18:56 +0000)]
Update to match the 1.0.3 import.

19 years ago- Fix build with TRACE_MESSAGES on.
glebius [Mon, 16 May 2005 18:50:26 +0000 (18:50 +0000)]
- Fix build with TRACE_MESSAGES on.
- Reformat code under TRACE_MESSAGES to make it more readable.
- Move linker hackery out of #ifdef.
- Break long lines in linker hackery block.

19 years agoThis commit was generated by cvs2svn to compensate for changes in r146293,
obrien [Mon, 16 May 2005 18:31:55 +0000 (18:31 +0000)]
This commit was generated by cvs2svn to compensate for changes in r146293,
which included commits to RCS files with non-trunk default branches.

19 years agoVirgin import (trimmed) of Bzip2 version 1.0.3.
obrien [Mon, 16 May 2005 18:31:55 +0000 (18:31 +0000)]
Virgin import (trimmed) of Bzip2 version 1.0.3.

19 years agoHardcode username to 8 characters. This makes top(1) output sane when
obrien [Mon, 16 May 2005 18:17:38 +0000 (18:17 +0000)]
Hardcode username to 8 characters.  This makes top(1) output sane when
there are users on the system (even if not running a single process)
with a login > 8 chars.
I'm not all that happy limiting the username width like this, but it
restores sanity to top(1) output.

Discussed with: keramida

19 years agoClose race between node being shutdown and socket being detached. To
glebius [Mon, 16 May 2005 17:25:49 +0000 (17:25 +0000)]
Close race between node being shutdown and socket being detached. To
do this, obtain netgraph locking in detach method via ng_send_fn1().

Reviewed by: julian
MFC after: 2 weeks

19 years agoDont clear all flags in vkbd_clear_state_locked(). Clear only COMPOSE flag.
emax [Mon, 16 May 2005 17:21:10 +0000 (17:21 +0000)]
Dont clear all flags in vkbd_clear_state_locked(). Clear only COMPOSE flag.

MFC after: 3 days

19 years agoEnable lukemftpd for 6.0-RELEASE.
obrien [Mon, 16 May 2005 17:12:23 +0000 (17:12 +0000)]
Enable lukemftpd for 6.0-RELEASE.

19 years agoCatch up with new ng_package_data().
glebius [Mon, 16 May 2005 17:10:08 +0000 (17:10 +0000)]
Catch up with new ng_package_data().

19 years agoCatch up with new ng_package_data(). Use NG_WAITOK on userland
glebius [Mon, 16 May 2005 17:09:35 +0000 (17:09 +0000)]
Catch up with new ng_package_data(). Use NG_WAITOK on userland
path.

19 years ago- Catch up with new interface to ng_package_data().
glebius [Mon, 16 May 2005 17:08:53 +0000 (17:08 +0000)]
- Catch up with new interface to ng_package_data().
- Handle errors from ng_package_data().

19 years agoCatch up with new ng_send_fn1() interface.
glebius [Mon, 16 May 2005 17:07:39 +0000 (17:07 +0000)]
Catch up with new ng_send_fn1() interface.

19 years ago- Extend interface of ng_getqblk(), so that malloc wait flags are
glebius [Mon, 16 May 2005 17:07:03 +0000 (17:07 +0000)]
- Extend interface of ng_getqblk(), so that malloc wait flags are
  specified by caller.
- Change ng_send_item() interface - use 'flags' argument instead of
  boolean 'queue'.
- Extend ng_send_fn(), ng_package_data() and ng_package_msg()
  interface - add possibility to pass flags. Rename ng_send_fn() to
  ng_send_fn1(). Create macro for ng_send_fn().
- Update all macros, that use ng_package_data() and ng_package_msg().

Reviewed by: julian

19 years agoCorrect type for workitem routines.
wpaul [Mon, 16 May 2005 16:50:52 +0000 (16:50 +0000)]
Correct type for workitem routines.

19 years agoEnable building /sbin/ipf (but not the rescue version) with the ability to
darrenr [Mon, 16 May 2005 16:22:55 +0000 (16:22 +0000)]
Enable building /sbin/ipf (but not the rescue version) with the ability to
parse bpf strings for filter rules in ipf.conf

19 years agoRemove harmless bit of leftover debug code.
wpaul [Mon, 16 May 2005 15:44:41 +0000 (15:44 +0000)]
Remove harmless bit of leftover debug code.

19 years agoCorrect some problems with workitem usage. NdisScheduleWorkItem() does
wpaul [Mon, 16 May 2005 15:29:21 +0000 (15:29 +0000)]
Correct some problems with workitem usage. NdisScheduleWorkItem() does
not use exactly the same workitem sturcture as ExQueueWorkItem() like
I originally thought it did.

19 years agoFixed markup from the previous revision.
ru [Mon, 16 May 2005 15:09:13 +0000 (15:09 +0000)]
Fixed markup from the previous revision.

19 years ago longer used (contents added to sys/ata.h)
sos [Mon, 16 May 2005 13:39:49 +0000 (13:39 +0000)]
 longer used (contents added to sys/ata.h)

19 years agoReflect the new usage.
sos [Mon, 16 May 2005 13:32:23 +0000 (13:32 +0000)]
Reflect the new usage.

19 years agoChange the way ioctls are issue to ATA.
sos [Mon, 16 May 2005 13:07:27 +0000 (13:07 +0000)]
Change the way ioctls are issue to ATA.
The most prominent part is that its now possible to issue ata_requests
directly to say acd0, instead of going through the cumbersome /dev/ata
device.

19 years agoAdd the 2nd word of IA32 feature flags. This includes things such as SSE3.
obrien [Mon, 16 May 2005 09:47:53 +0000 (09:47 +0000)]
Add the 2nd word of IA32 feature flags.  This includes things such as SSE3.

Obtained from: sys/amd64/amd64/identcpu.

19 years agoMake <runefile.h> internal to libc.
ru [Mon, 16 May 2005 09:32:41 +0000 (09:32 +0000)]
Make <runefile.h> internal to libc.

Suggested by: phantom

19 years agoCorrect format is 'options<space><tab>'.
glebius [Mon, 16 May 2005 09:05:46 +0000 (09:05 +0000)]
Correct format is 'options<space><tab>'.

Submitted by: ru

19 years agoMake NETGRAPH_DEBUG a kernel option, so that it can't be turned off
glebius [Mon, 16 May 2005 08:25:55 +0000 (08:25 +0000)]
Make NETGRAPH_DEBUG a kernel option, so that it can't be turned off
without hacking source.

In collaboration with: ru, julian

19 years agoAdd convenience functions to get port and interface.
imp [Mon, 16 May 2005 06:58:43 +0000 (06:58 +0000)]
Add convenience functions to get port and interface.

19 years agocleanup a danging reference to cleaning up /etc/fstab. Since we don't
imp [Mon, 16 May 2005 05:37:32 +0000 (05:37 +0000)]
cleanup a danging reference to cleaning up /etc/fstab.  Since we don't
support 4->6 upgrades, this is moot.  Most 4.x installations even,
have things compatible with 5 at thsi point, but some don't, so I'll
leave it in the branch...

submitted by: kevlo

19 years agoA second attempt to adjust option-parsing on a shell command, for the
gad [Mon, 16 May 2005 04:32:41 +0000 (04:32 +0000)]
A second attempt to adjust option-parsing on a shell command, for the
benefit of scripts start out as:     #!/bin/sh -- # -*- perl -*-
With this fix in place, we can commit a change to kern/imgact_shell.c
so FreeBSD will process the `#!' line in shell-scripts in a more
standard fashion.

PR: 16393
Mentioned on: freebsd-arch

19 years agoAdd an #ifdef'd non-blocking version of the test.
rwatson [Mon, 16 May 2005 00:54:47 +0000 (00:54 +0000)]
Add an #ifdef'd non-blocking version of the test.

Update copyright.

19 years agoClose the connect socket as well as the listen socket on completion.
rwatson [Mon, 16 May 2005 00:53:38 +0000 (00:53 +0000)]
Close the connect socket as well as the listen socket on completion.

Update copyright.

19 years agoAdd note about required firmware to save others time and frustration with this device
imp [Sun, 15 May 2005 21:08:51 +0000 (21:08 +0000)]
Add note about required firmware to save others time and frustration with this device

19 years agoPrint a warning once when trying to bring up interface before firmware load.
imp [Sun, 15 May 2005 21:02:51 +0000 (21:02 +0000)]
Print a warning once when trying to bring up interface before firmware load.

19 years ago- The ai_addrlen of a struct addrinfo used to be a size_t, per
ume [Sun, 15 May 2005 20:15:15 +0000 (20:15 +0000)]
- The ai_addrlen of a struct addrinfo used to be a size_t, per
  RFC 2553.  In XNS5.2, and subsequently in POSIX-2001 and RFC
  3493, it was changed to a socklen_t.  And, the n_net of a
  struct netent used to be an unsigned long integer.  In XNS5,
  and subsequently in POSIX-2001, it was changed to an uint32_t.
  To accomodate for this while preserving ABI compatibility with
  the old interface, we need to prepend or append 32 bits of
  padding, depending on the (LP64) architecture's endianness.
- Correct 1st argument of getnetbyaddr() to uint32_t on 32
  bit arch.  Stay as is on 64 bit arch for ABI backward
  compatibility for now.

Reviewed by: das, peter
MFC after: 2 weeks

19 years agoUpdate ndiscvt to handle .INF files that specify multiple entries in
wpaul [Sun, 15 May 2005 19:46:14 +0000 (19:46 +0000)]
Update ndiscvt to handle .INF files that specify multiple entries in
their [Manufacturer] sections.

19 years agoUse \033 instead of \e because \e is not a standard.
nyan [Sun, 15 May 2005 09:43:14 +0000 (09:43 +0000)]
Use \033 instead of \e because \e is not a standard.

Noticed by: stefanf
Reviewed by: diff(1)

19 years agoRemove old epson note support.
nyan [Sun, 15 May 2005 09:07:42 +0000 (09:07 +0000)]
Remove old epson note support.

19 years ago- Add color tables for 16 colors mode and 8 colors mode, use a different
nyan [Sun, 15 May 2005 09:07:04 +0000 (09:07 +0000)]
- Add color tables for 16 colors mode and 8 colors mode, use a different
  table on the pc98 console.
- Remove old epson note support.

19 years ago- Mask an underline attribute on the pc98 console. It enables to use the same
nyan [Sun, 15 May 2005 08:59:00 +0000 (08:59 +0000)]
- Mask an underline attribute on the pc98 console.  It enables to use the same
  color on the pc98 and the others.
- Remove old epson note support.

19 years agoA foreground color number on pc98 console is 7 not 15.
nyan [Sun, 15 May 2005 08:04:46 +0000 (08:04 +0000)]
A foreground color number on pc98 console is 7 not 15.

19 years agoUse \e for escape char instead of ^[ directly.
nyan [Sun, 15 May 2005 07:41:38 +0000 (07:41 +0000)]
Use \e for escape char instead of ^[ directly.

19 years agoRemove the ADJUST_CLOCK ioctl on pc98. It's not used at all.
nyan [Sun, 15 May 2005 07:35:49 +0000 (07:35 +0000)]
Remove the ADJUST_CLOCK ioctl on pc98.  It's not used at all.

19 years agoo -n option added.
marcel [Sun, 15 May 2005 07:35:34 +0000 (07:35 +0000)]
o  -n option added.
o  -x option added.