]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
16 years agoRefactor one of the ISO extraction tests: Move the reference
Tim Kientzle [Sat, 21 Jun 2008 19:11:51 +0000 (19:11 +0000)]
Refactor one of the ISO extraction tests:  Move the reference
file into a separate file (instead of embedding it in the C code)
and use later timestamps (timestamps too close to the Epoch fail
predictably on systems that lack timegm(), whose mktime() doesn't
support dates before the Epoch and which are running in timezones
with negative offsets from GMT).  The goal here is to test the ISO
extraction, not the local platform's time support.

16 years agoMfP4: Joerg Sonnenberg's extensions to the mtree support for more
Tim Kientzle [Sat, 21 Jun 2008 19:06:37 +0000 (19:06 +0000)]
MfP4:  Joerg Sonnenberg's extensions to the mtree support for more
complete quoting.  In particular, this handles escaped newlines
and common C-style escapes.

16 years agoA security-check failure here should be ARCHIVE_FAILED (cannot continue this
Tim Kientzle [Sat, 21 Jun 2008 19:05:29 +0000 (19:05 +0000)]
A security-check failure here should be ARCHIVE_FAILED (cannot continue this
operation) and not ARCHIVE_WARN, since we don't actually open the file.
Both bsdtar and bsdcpio will try to copy file contents after an ARCHIVE_WARN,
which will fail loudly.

16 years agoIf we're using -l and can't hardlink the file because of a cross-device
Tim Kientzle [Sat, 21 Jun 2008 17:47:56 +0000 (17:47 +0000)]
If we're using -l and can't hardlink the file because of a cross-device
link, just ignore the -l option and copy the file instead.
In particular, this should fix the COPYTREE_* macros used in the
ports infrastructure which use -l to preserve space but often get
used for cross-device copies.

16 years agoAdd support for PORT/EPRT FTP commands in lowercase.
Alexander Motin [Sat, 21 Jun 2008 16:22:56 +0000 (16:22 +0000)]
Add support for PORT/EPRT FTP commands in lowercase.
Use strncasecmp() instead of huge local implementation to reduce code size.
Check space presence after command/code.

PR: kern/73034

16 years agoRemoved the no-op -p; documented -P.
Ruslan Ermilov [Sat, 21 Jun 2008 15:48:16 +0000 (15:48 +0000)]
Removed the no-op -p; documented -P.

MFC after: 3 days

16 years agoAdd -v (verbose) option to -l command, to show size and backing store
Dmitry Morozovsky [Sat, 21 Jun 2008 15:04:42 +0000 (15:04 +0000)]
Add -v (verbose) option to -l command, to show size and backing store
of all md devices at one time.

Approved by: phk
MFC after: 2 weeks

16 years agoTo catch up with rev 179872: rename enable_quotas to quota_enable.
Simon L. B. Nielsen [Sat, 21 Jun 2008 13:53:35 +0000 (13:53 +0000)]
To catch up with rev 179872: rename enable_quotas to quota_enable.

16 years agoUse minimum of max_aio_procs and target_aio_procs when spawning new
Oleksandr Tymoshenko [Sat, 21 Jun 2008 11:34:34 +0000 (11:34 +0000)]
Use minimum of max_aio_procs and target_aio_procs when spawning new
aiod since there should be no more then max_aio_procs processes.

16 years agoRework line-processing framework to add support for --null and
Tim Kientzle [Sat, 21 Jun 2008 02:20:20 +0000 (02:20 +0000)]
Rework line-processing framework to add support for --null and
to eliminate a callback.

16 years agoVarious long options for GNU cpio compat.
Tim Kientzle [Sat, 21 Jun 2008 02:18:52 +0000 (02:18 +0000)]
Various long options for GNU cpio compat.

16 years agoMfP4: test improvements, mostly for portability.
Tim Kientzle [Sat, 21 Jun 2008 02:17:18 +0000 (02:17 +0000)]
MfP4: test improvements, mostly for portability.

16 years ago- link powerd(8) man page
Daniel Gerzo [Fri, 20 Jun 2008 23:04:34 +0000 (23:04 +0000)]
- link powerd(8) man page

PR: docs/123111
MFC after: 3 days

16 years agoFix PCI id for 945GME Express Integrated Graphics Controller:
Oleksandr Tymoshenko [Fri, 20 Jun 2008 22:23:41 +0000 (22:23 +0000)]
Fix PCI id for 945GME Express Integrated Graphics Controller:
set to 8086:27AE

PR:     kern/124782
Event:  Bugathon#5

16 years agoFix spelling
Oleksandr Tymoshenko [Fri, 20 Jun 2008 21:41:44 +0000 (21:41 +0000)]
Fix spelling

PR: kern/124723
Event: Bugathon#5

16 years agoMake preparations for increasing the size of the kernel virtual
Alan Cox [Fri, 20 Jun 2008 20:59:31 +0000 (20:59 +0000)]
Make preparations for increasing the size of the kernel virtual
address space on the amd64 architecture.  The amd64 architecture
requires kernel code and global variables to reside in the highest 2GB
of the 64-bit virtual address space.  Thus, KERNBASE cannot change.
However, KERNBASE is sometimes used as the start of the kernel virtual
address space.  Henceforth, VM_MIN_KERNEL_ADDRESS should be used
instead.  Since KERNBASE and VM_MIN_KERNEL_ADDRESS are still the same
address, there should be no visible effect from this change (yet).
That said, kris@ has tested crash dumps under the full patch that
increases the kernel virtual address space on amd64 to 6GB.

Tested by: kris@

16 years ago- Fix spelling errors.
Ulf Lilleengen [Fri, 20 Jun 2008 19:48:18 +0000 (19:48 +0000)]
- Fix spelling errors.

Approved by:    kib (mentor)
PR:             kern/124788
Submitted by:   Hywel Mallett <Hywel -at- hmallett.co.uk>

16 years agoAdd et(4), a port of DragonFly's Agere ET1310 10/100/Gigabit
Xin LI [Fri, 20 Jun 2008 19:30:44 +0000 (19:30 +0000)]
Add et(4), a port of DragonFly's Agere ET1310 10/100/Gigabit
Ethernet device driver, written by sephe@

Obtained from: DragonFly
Sponsored by: iXsystems
MFC after: 2 weeks

16 years agoAdd et(4), a port of DragonFly's Agere ET1310 10/100/Gigabit
Xin LI [Fri, 20 Jun 2008 19:28:33 +0000 (19:28 +0000)]
Add et(4), a port of DragonFly's Agere ET1310 10/100/Gigabit
Ethernet device driver, written by sephe@

Obtained from: DragonFly
Sponsored by: iXsystems
MFC after: 2 weeks

16 years agoAdd support for the optional key in the GRE header.
Andrew Thompson [Fri, 20 Jun 2008 17:26:34 +0000 (17:26 +0000)]
Add support for the optional key in the GRE header.

PR: kern/114714
Submitted by: Cristian KLEIN

16 years agoSplit out the probing magic of device_probe_and_attach into
Warner Losh [Fri, 20 Jun 2008 16:58:15 +0000 (16:58 +0000)]
Split out the probing magic of device_probe_and_attach into
device_probe() so that it can be used by busses that may wish to do
additional processing between probe and attach.

Reviewed by: dfr@

16 years agoAdd Xr to getsockname(2)
Poul-Henning Kamp [Fri, 20 Jun 2008 14:47:06 +0000 (14:47 +0000)]
Add Xr to getsockname(2)

16 years agoMake the search for sources in PATH_PORTS more accurate. I only
Joerg Wunsch [Fri, 20 Jun 2008 08:39:42 +0000 (08:39 +0000)]
Make the search for sources in PATH_PORTS more accurate.  I only
noticed that a "whereis -qs qemu" matched the distfiles subdir of qemu
rather than /usr/ports/emulators/qemu.

It now ignores all dot entries in /usr/ports, plus all entries
starting with a capital letter (maintenance stuff like Templates, but
also includes subdir CVS), plus /usr/ports/distfiles which is simply a
magic name in that respect.

16 years agoEnforce the mapping of kernel loadable modules in the uppermost 2GB of the
Alan Cox [Fri, 20 Jun 2008 06:24:34 +0000 (06:24 +0000)]
Enforce the mapping of kernel loadable modules in the uppermost 2GB of the
kernel virtual address space on amd64.

16 years agoMake preparations for increasing the size of the kernel virtual
Alan Cox [Fri, 20 Jun 2008 05:22:09 +0000 (05:22 +0000)]
Make preparations for increasing the size of the kernel virtual
address space on the amd64 architecture.  The amd64 architecture
requires kernel code and global variables to reside in the highest 2GB
of the 64-bit virtual address space.  Thus, KERNBASE cannot change.
However, KERNBASE is sometimes used as the start of the kernel virtual
address space.  Henceforth, VM_MIN_KERNEL_ADDRESS should be used
instead.  Since KERNBASE and VM_MIN_KERNEL_ADDRESS are still the same
address, there should be no visible effect from this change (yet).

16 years agoImplement fmodl.
David Schultz [Thu, 19 Jun 2008 22:39:53 +0000 (22:39 +0000)]
Implement fmodl.
Document fmodl and fix some errors in the fmod manpage.

16 years agoRevert rev. 178124 as requested by kris@. Having jail id not being
Xin LI [Thu, 19 Jun 2008 21:41:57 +0000 (21:41 +0000)]
Revert rev. 178124 as requested by kris@.  Having jail id not being
reused too frequently is useful for script controlled environment.

16 years agoDocument the _arg versions of the uma_zalloc and uma_zfree functions.
Remko Lodder [Thu, 19 Jun 2008 18:33:38 +0000 (18:33 +0000)]
Document the _arg versions of the uma_zalloc and uma_zfree functions.

PR: docs/120357
Submitted by: gahr
MFC after: 3 days

16 years agoRenew semaphore's pointer after wakeup since during msleep
Oleksandr Tymoshenko [Thu, 19 Jun 2008 18:08:42 +0000 (18:08 +0000)]
Renew semaphore's pointer after wakeup since during msleep
sem_base may have been modified by destroying one of semaphores
and semptr would not be valid in this case.

PR: kern/123731

16 years agoAdd myself. Better late then never.
Alexander Motin [Thu, 19 Jun 2008 17:10:05 +0000 (17:10 +0000)]
Add myself. Better late then never.

16 years agoAdd myself.
Dmitry Marakasov [Thu, 19 Jun 2008 16:30:06 +0000 (16:30 +0000)]
Add myself.

Approved by: miwi (mentor)

16 years agoAdd myself.
Dmitry Marakasov [Thu, 19 Jun 2008 16:29:37 +0000 (16:29 +0000)]
Add myself.

Approved by: miwi (mentor)

16 years agoRemove __restrict keywords from array arguments to make GCC's -std=c99 work.
Ed Schouten [Thu, 19 Jun 2008 07:30:32 +0000 (07:30 +0000)]
Remove __restrict keywords from array arguments to make GCC's -std=c99 work.

When GCC is invoked with -std=c99, the following errors are displayed when
including <spawn.h>:

/usr/include/spawn.h:69: error: static or type qualifiers in abstract declarator
/usr/include/spawn.h:69: error: static or type qualifiers in abstract declarator
/usr/include/spawn.h:72: error: static or type qualifiers in abstract declarator
/usr/include/spawn.h:72: error: static or type qualifiers in abstract declarator

We'd better remove the __restrict keywords here. The same is also done
in <regex.h>.

Submitted by: Andrzej Tobola <ato iem pw edu pl>
Reviewed by: davidxu
Approved by: philip (mentor, implicit)

16 years agoMake quota knob conform to other rc(8) knobs. Keep older knob for
Mike Makonnen [Thu, 19 Jun 2008 07:06:11 +0000 (07:06 +0000)]
Make quota knob conform to other rc(8) knobs. Keep older knob for
compatibility.

Requested by: Volker <volker@vwsoft.com>

16 years agoMove the check for enabled knobs further down in run_rc_command() so
Mike Makonnen [Thu, 19 Jun 2008 06:11:34 +0000 (06:11 +0000)]
Move the check for enabled knobs further down in run_rc_command() so
that bogus commands cause usage information to be printed instead of
diagnostics about enabling the knob.

16 years agoProcess spawn attributes in POSIX document order.
David Xu [Thu, 19 Jun 2008 02:42:50 +0000 (02:42 +0000)]
Process spawn attributes in POSIX document order.

16 years agoAdd roken.h to SRCS. This fixes the compilation of slc during a
Marius Strobl [Wed, 18 Jun 2008 21:20:50 +0000 (21:20 +0000)]
Add roken.h to SRCS. This fixes the compilation of slc during a
buildworld on a host running a world built with WITHOUT_KERBEROS
defined.

16 years agoAdd the option stub for the Globetrotter Max 3.6 Modem.
Attilio Rao [Wed, 18 Jun 2008 20:50:30 +0000 (20:50 +0000)]
Add the option stub for the Globetrotter Max 3.6 Modem.

Submitted by: Greg Rivers <gcr at tharned dot org>

16 years agoPrint out the container lock when showing the thread state in DDB.
Attilio Rao [Wed, 18 Jun 2008 20:42:01 +0000 (20:42 +0000)]
Print out the container lock when showing the thread state in DDB.

Tested by: benjsc

16 years agoAdd the SMB functionality for the MCP65 chipset I happen to
Joerg Wunsch [Wed, 18 Jun 2008 20:39:56 +0000 (20:39 +0000)]
Add the SMB functionality for the MCP65 chipset I happen to
have in my new motherboard.

16 years agoCatch up to recentish kgdb changes:
John Baldwin [Wed, 18 Jun 2008 14:23:28 +0000 (14:23 +0000)]
Catch up to recentish kgdb changes:
- Use ptid_get_pid() rather than ptid_get_tid() (part of the changes to
  let 'tid' work for remote kgdb).
- Add a stub kgdb_trgt_new_objfile() hook.

Silence from: obrien, mips@

16 years agoAdd MIPS to the list of known arches.
David E. O'Brien [Wed, 18 Jun 2008 13:52:58 +0000 (13:52 +0000)]
Add MIPS to the list of known arches.

16 years agohandle .INF files for PCMCIA correctly that specify multiple entries in
Weongyo Jeong [Wed, 18 Jun 2008 06:35:37 +0000 (06:35 +0000)]
handle .INF files for PCMCIA correctly that specify multiple entries in
their [Manufacturer] sections and prevent a case that NDIS_PCI_DEV_TABLE
definition was always emitted that it's only emitted once if a .INF file
is for PCI.

16 years agoImplement the set and unset verbs. While here, have the manpage
Marcel Moolenaar [Wed, 18 Jun 2008 01:46:32 +0000 (01:46 +0000)]
Implement the set and unset verbs. While here, have the manpage
catch up with various changes.

16 years agoAdd the set and unset verbs used to set and clear attributes for
Marcel Moolenaar [Wed, 18 Jun 2008 01:13:34 +0000 (01:13 +0000)]
Add the set and unset verbs used to set and clear attributes for
partition entries. Implement the setunset method for the MBR
scheme to control the active flag.

16 years agoBring this up to date with regard to our other section 4 and Intel manpages,
Christian Brueffer [Tue, 17 Jun 2008 21:14:02 +0000 (21:14 +0000)]
Bring this up to date with regard to our other section 4 and Intel manpages,
also fixes a few mdoc bugs.

16 years agoRemove superfluous eofmarker.
Remko Lodder [Tue, 17 Jun 2008 18:56:04 +0000 (18:56 +0000)]
Remove superfluous eofmarker.

Requested by: Jaakko Heinonen
Discussed with: Jaakko, edwin

Approved by: imp (mentor, implicit)

16 years agoDon't export the unused __use_pts() routine.
Ed Schouten [Tue, 17 Jun 2008 14:05:03 +0000 (14:05 +0000)]
Don't export the unused __use_pts() routine.

The __use_pts() routine was once probably used by libutil to determine
if we are using BSD or UNIX98 style PTY device names. It doesn't seem to
be used outside grantpt.c, which means we can make it static and remove
it from the Symbol.map.

Reviewed by: cognet, kib
Approved by: philip (mentor)

16 years agoBack out rev. 1.352 (SVN rev 179842) as phk pointed out that
Bjoern A. Zeeb [Tue, 17 Jun 2008 11:08:49 +0000 (11:08 +0000)]
Back out rev. 1.352  (SVN rev 179842) as phk pointed out that
SRCCONF was omitted here to be able to build non standard trees
with universe as well.

16 years agoFor make universe, in addition to make.conf, also ignore a
Bjoern A. Zeeb [Tue, 17 Jun 2008 09:06:46 +0000 (09:06 +0000)]
For make universe, in addition to make.conf, also ignore a
src.conf with possibe non-default options.

Reviewed by: ru
MFC after: 10 days

16 years agoStyle fix.
David Xu [Tue, 17 Jun 2008 08:23:45 +0000 (08:23 +0000)]
Style fix.

16 years agoChange my email address to the one from the FreeBSD project.
Ed Schouten [Tue, 17 Jun 2008 07:09:58 +0000 (07:09 +0000)]
Change my email address to the one from the FreeBSD project.

Approved by: philip (mentor, implicit), davidxu

16 years agoAdd POSIX routines called posix_spawn() and posix_spawnp(), which
David Xu [Tue, 17 Jun 2008 06:26:29 +0000 (06:26 +0000)]
Add POSIX routines called posix_spawn() and posix_spawnp(), which
can be used as replacements for exec/fork in a lot of cases. This
change also added execvpe() which allows environment variable
PATH to be used for searching executable file, it is used for
implementing posix_spawnp().

PR: standards/122051

16 years ago- Move ether_ifdetach earlier.
Benno Rice [Tue, 17 Jun 2008 05:48:42 +0000 (05:48 +0000)]
- Move ether_ifdetach earlier.
- Drain callouts after ether_ifdetach.

Suggested by: jhb

16 years agoRemove support for RELENG_4 (__FreeBSD_version < 500039).
Florent Thoumie [Mon, 16 Jun 2008 23:41:11 +0000 (23:41 +0000)]
Remove support for RELENG_4 (__FreeBSD_version < 500039).

MFC after: 1 day

16 years agoInstall extra include files that were forgotten in original GCC 4.2.x
Alexander Kabaev [Mon, 16 Jun 2008 22:49:30 +0000 (22:49 +0000)]
Install extra include files that were forgotten in original GCC 4.2.x
import changes.

PR: 124647
Submitted by: Vlad GALU
MFC after: 2 days

16 years agoChange incorrect stale cookie detection in syncookie_lookup() that prematurely
Stephan Uphoff [Mon, 16 Jun 2008 20:08:22 +0000 (20:08 +0000)]
Change incorrect stale cookie detection in syncookie_lookup() that prematurely
declared a cookie as expired.

Reviewed by: andre@, silby@
Reported by:    Yahoo!

16 years agoFix a check in SYN cache expansion (syncache_expand()) to accept packets that arrive...
Stephan Uphoff [Mon, 16 Jun 2008 19:56:59 +0000 (19:56 +0000)]
Fix a check in SYN cache expansion (syncache_expand()) to accept packets that arrive in the receive window instead of just on the left edge of the receive window.
This is needed for correct behavior when packets are lost or reordered.

PR: kern/123950
Reviewed by: andre@, silby@
Reported by: Yahoo!, Wang Jin
MFC after: 1 week

16 years agoAdd another 8139D variant.
Remko Lodder [Mon, 16 Jun 2008 18:32:20 +0000 (18:32 +0000)]
Add another 8139D variant.

PR: 124622
Submitted by: Evgeny Zhirnov <jirnov at gmail dot com>
Approved by: imp (mentor, implicit)
MFC after: 3 days

16 years agoRecord my roots.
Ermal Luçi [Mon, 16 Jun 2008 17:38:48 +0000 (17:38 +0000)]
Record my roots.

Reviewed by: mlaier (mentor)

16 years agoAdd my birthday to the calendar.
Ermal Luçi [Mon, 16 Jun 2008 17:35:34 +0000 (17:35 +0000)]
Add my birthday to the calendar.

Approved by: mlaier (mentor)

16 years agoStruct cdev is always the member of the struct cdev_priv. When devfs
Konstantin Belousov [Mon, 16 Jun 2008 17:34:59 +0000 (17:34 +0000)]
Struct cdev is always the member of the struct cdev_priv. When devfs
needed to promote cdev to cdev_priv, the si_priv pointer was followed.

Use member2struct() to calculate address of the wrapping cdev_priv.
Rename si_priv to __si_reserved.

Tested by: pho
Reviewed by: ed
MFC after: 2 weeks

16 years agoAdd the member2struct() macro, that returns pointer to the containing
Konstantin Belousov [Mon, 16 Jun 2008 17:25:20 +0000 (17:25 +0000)]
Add the member2struct() macro, that returns pointer to the containing
structure given pointer to some structure member.

MFC after: 2 weeks

16 years agoAdd $FreeBSD$ since we now have local changes.
David E. O'Brien [Mon, 16 Jun 2008 17:06:17 +0000 (17:06 +0000)]
Add $FreeBSD$ since we now have local changes.

16 years agoImplement a workaround for a long-standing problem in
Oliver Fromme [Mon, 16 Jun 2008 17:04:04 +0000 (17:04 +0000)]
Implement a workaround for a long-standing problem in
libi386's time(), caused by a qemu bug.  The bug might
be present in other BIOSes, too.

qemu either does not simulate the AT RTC correctly or
has a broken BIOS 1A/02 implementation, and will return
an incorrect value if the RTC is read while it is being
updated.

The effect is worsened by the fact that qemu's INT 15/86
function ("wait" a.k.a. usleep) is non-implmeneted or
broken and returns immediately, causing beastie.4th to
spin in a tight loop calling the "read RTC" function
millions of times, triggering the problem quickly.

Therefore, we keep reading the BIOS value until we get
the same result twice.  This change fixes beastie.4th's
countdown under qemu.

Approved by: des (mentor)

16 years agoMake it clearer that privilege is needed to reduce as well as
Tony Finch [Mon, 16 Jun 2008 14:50:21 +0000 (14:50 +0000)]
Make it clearer that privilege is needed to reduce as well as
increase group membership.

16 years agoUpdate to include the Camellia algorithm which is in the code but
George V. Neville-Neil [Mon, 16 Jun 2008 14:33:54 +0000 (14:33 +0000)]
Update to include the Camellia algorithm which is in the code but
which was accidentally left undocumented in the manual page.

16 years agoStyle fix (use naked commands).
Florent Thoumie [Mon, 16 Jun 2008 09:15:27 +0000 (09:15 +0000)]
Style fix (use naked commands).

Reported by: obrien

16 years agoInclude bsd.own.mk to pick up the definition of MK_GNU_CPIO
Doug Barton [Mon, 16 Jun 2008 07:24:05 +0000 (07:24 +0000)]
Include bsd.own.mk to pick up the definition of MK_GNU_CPIO

16 years agoProperly alphabetize the BSD_CPIO option
Doug Barton [Mon, 16 Jun 2008 07:23:12 +0000 (07:23 +0000)]
Properly alphabetize the BSD_CPIO option

16 years agoRegenerate to add the definitions of WITHOUT_BSD_CPIO WITH_GNU_CPIO
Doug Barton [Mon, 16 Jun 2008 05:50:21 +0000 (05:50 +0000)]
Regenerate to add the definitions of WITHOUT_BSD_CPIO WITH_GNU_CPIO
and WITHOUT_GNU_GREP

16 years ago1. Make the BSD version of cpio the default [1]
Doug Barton [Mon, 16 Jun 2008 05:48:15 +0000 (05:48 +0000)]
1. Make the BSD version of cpio the default [1]
a. The BSD version will be built and installed unless
WITHOUT_BSD_CPIO is defined.
b. The GNU version will not be built or installed unless
WITH_GNU_CPIO is defined. If this is defined, the symlink
in /usr/bin will be to the GNU version whether the BSD
version is present or not.

When these changes are MFCed the defaults should be flipped.

2. Add a knob to disable the building of GNU grep. This will
make it easier for those that want to test the BSD version in
the ports.

Approved by: kientzle [1]

16 years agoRegenerate using svn $FreeBSD$ tags. There is no change in the actual
Doug Barton [Mon, 16 Jun 2008 05:33:16 +0000 (05:33 +0000)]
Regenerate using svn $FreeBSD$ tags. There is no change in the actual
man page text other than the date.

16 years agoRemove code that isn't required. It actually breaks the case where KDTRACE_HOOKS
John Birrell [Mon, 16 Jun 2008 04:44:29 +0000 (04:44 +0000)]
Remove code that isn't required. It actually breaks the case where KDTRACE_HOOKS
is defined and KDB isn't. This is the case that it was intended for.

16 years agoDo not redo the vnode tear-down work already done by insmntque() when
Konstantin Belousov [Sun, 15 Jun 2008 18:40:58 +0000 (18:40 +0000)]
Do not redo the vnode tear-down work already done by insmntque() when
vnode cannot be put on the vnode list for mount.

Reported and tested by: marck
Guilty party: me
MFC after: 3 days

16 years agoSilence warning about missing IoGetDeviceObjectPointer by implementing
Coleman Kane [Sun, 15 Jun 2008 13:37:29 +0000 (13:37 +0000)]
Silence warning about missing IoGetDeviceObjectPointer by implementing
a simple stub that always returns STATUS_SUCCESS.

Submitted by: Paul B. Mahol <onemda@gmail.com>
Reviewed by: thompsa
MFC after: 1 week

16 years agoremove incorrect comment that I missed in my last change
Kip Macy [Sun, 15 Jun 2008 13:26:25 +0000 (13:26 +0000)]
remove incorrect comment that I missed in my last change

16 years agoStop moused on a detach event. Remove incorrect comment.
Kip Macy [Sun, 15 Jun 2008 13:25:23 +0000 (13:25 +0000)]
Stop moused on a detach event. Remove incorrect comment.

This fixes frequent problems with usb mice and kvm switches caused by moused hanging around.

Suggested by: Matthew Dodd

MFC after: 2 weeks

16 years ago More prep for Vimage:
Randall Stewart [Sun, 15 Jun 2008 12:31:23 +0000 (12:31 +0000)]
 More prep for Vimage:
 - only one functino to destroy an SCTP stack sctp_finish()
 - Make it so this function also arranges for any threads
   created by the image to do a kthread_exit()

16 years agoRework the my_mbtowc_utf8() support function to fully match the mbtowc()
Tim Kientzle [Sun, 15 Jun 2008 11:28:56 +0000 (11:28 +0000)]
Rework the my_mbtowc_utf8() support function to fully match the mbtowc()
calling convention, not the mbrtowc() convention.

16 years agoSince wctomb() returns int, temporaries should be int, not size_t.
Tim Kientzle [Sun, 15 Jun 2008 11:15:12 +0000 (11:15 +0000)]
Since wctomb() returns int, temporaries should be int, not size_t.

16 years agoA number of minor corrections to the support for external compression
Tim Kientzle [Sun, 15 Jun 2008 10:45:57 +0000 (10:45 +0000)]
A number of minor corrections to the support for external compression
programs:
  * Support platforms that have fork() but not vfork()
  * Don't write(), select(), or poll() against closed file descriptors

16 years agoBe a little more careful about closing file descriptors that may not exist.
Tim Kientzle [Sun, 15 Jun 2008 10:43:59 +0000 (10:43 +0000)]
Be a little more careful about closing file descriptors that may not exist.

16 years agoMfP4: test harness improvements.
Tim Kientzle [Sun, 15 Jun 2008 10:35:22 +0000 (10:35 +0000)]
MfP4: test harness improvements.

16 years agoFix a bad cast.
Tim Kientzle [Sun, 15 Jun 2008 10:26:08 +0000 (10:26 +0000)]
Fix a bad cast.

16 years agoMfP4: Minor portability fix.
Tim Kientzle [Sun, 15 Jun 2008 10:08:16 +0000 (10:08 +0000)]
MfP4: Minor portability fix.

16 years agoMfP4: test harness cleanup.
Tim Kientzle [Sun, 15 Jun 2008 10:07:54 +0000 (10:07 +0000)]
MfP4: test harness cleanup.

16 years agoFix reading TOC from zip archives with unsupported
Tim Kientzle [Sun, 15 Jun 2008 05:15:53 +0000 (05:15 +0000)]
Fix reading TOC from zip archives with unsupported
compression.  We can't read the body, but we
shouldn't try to skip the body twice.

16 years agoMark hardlinks that lack any other interesting filetype information
Tim Kientzle [Sun, 15 Jun 2008 05:14:01 +0000 (05:14 +0000)]
Mark hardlinks that lack any other interesting filetype information
with an 'h'.

16 years agoPortability: We can get away with the older and better-supported
Tim Kientzle [Sun, 15 Jun 2008 05:12:47 +0000 (05:12 +0000)]
Portability:  We can get away with the older and better-supported
wctomb() here; we don't need wcrtomb().  In particular, this fixes
libarchive building on FreeBSD 4.

16 years agoPortability: wchar_t is defined in stdlib.h on some systems.
Tim Kientzle [Sun, 15 Jun 2008 05:11:08 +0000 (05:11 +0000)]
Portability:  wchar_t is defined in stdlib.h on some systems.

16 years agoarchive.h is no longer constructed from archive.h.in,
Tim Kientzle [Sun, 15 Jun 2008 05:05:53 +0000 (05:05 +0000)]
archive.h is no longer constructed from archive.h.in,
so we can rename it and drop some no-longer-necessary
build magic from the Makefile.

16 years agoFix the new generic link resolver in libarchive to never match
Tim Kientzle [Sun, 15 Jun 2008 04:31:43 +0000 (04:31 +0000)]
Fix the new generic link resolver in libarchive to never match
dirs as hardlinks.  In particular, this fixes some recent ports
build failures.

Thanks to: Kris Kennaway

16 years agoWhen NETATALK is compiled into the kernel, at_rmx.c is required regardless
Robert Watson [Sat, 14 Jun 2008 15:17:02 +0000 (15:17 +0000)]
When NETATALK is compiled into the kernel, at_rmx.c is required regardless
of whether NETATALKDEBUG is enabled, so make building it conditional on
NETATALK instead.  This problem appears to have been present from the time
that the netatalk implementation was imported.

PR: 124456
Submitted by: Nathan Whitehorn <whitehorn at wisc dot edu>
MFC after: 3 days

16 years ago- Fixes foobar on my part. Some missing virtualization macros from
Randall Stewart [Sat, 14 Jun 2008 13:24:49 +0000 (13:24 +0000)]
- Fixes foobar on my part. Some missing virtualization macros from
  specific logging cases.

16 years agoRemove obselete PECOFF image activator support.
Wojciech A. Koszek [Sat, 14 Jun 2008 12:51:44 +0000 (12:51 +0000)]
Remove obselete PECOFF image activator support.

PRs assigned at the time of removal:    kern/80742

Discussed on:   freebsd-current (silence), IRC
Tested by:      make universe
Approved by:    cognet (mentor)

16 years agoTurn sgtty into a binary-only compatibility interface.
Ed Schouten [Sat, 14 Jun 2008 10:42:18 +0000 (10:42 +0000)]
Turn sgtty into a binary-only compatibility interface.

sgtty was the original interface to configure terminal attributes on my
UNIX-like operating systems. It has been deprecated by the POSIX termios
interface, which is implemented in almost any modern system.

An advantage of turning this into a binary compatibility interface, is
that we can now eventually remove the COMPAT_43TTY switch from kernel
configurations. This removes many ioctl()'s from the TTY layer.

While there, increase the __FreeBSD_version, which may be useful for the
people working on the Ports tree.

Reviewed by: kib
Approved by: philip (mentor)

16 years ago- Macro-izes the packed declaration in all headers.
Randall Stewart [Sat, 14 Jun 2008 07:58:05 +0000 (07:58 +0000)]
- Macro-izes the packed declaration in all headers.
- Vimage prep - these are major restructures to move
  all global variables to be accessed via a macro or two.
  The variables all go into a single structure.
- Asconf address addition tweaks (add_or_del Interfaces)
- Fix rwnd calcualtion to be more conservative.
- Support SACK_IMMEDIATE flag to skip delayed sack
  by demand of peer.
- Comment updates in the sack mapping calculations
- Invarients panic added.
- Pre-support for UDP tunneling (we can do this on
  MAC but will need added support from UDP to
  get a "pipe" of UDP packets in.
- clear trace buffer sysctl added when local tracing on.

Note the majority of this huge patch is all the vimage prep stuff :-)

16 years agoThe TrustedBSD MAC Framework named struct ipq instances 'ipq', which is the
Robert Watson [Fri, 13 Jun 2008 22:14:15 +0000 (22:14 +0000)]
The TrustedBSD MAC Framework named struct ipq instances 'ipq', which is the
same as the global variable defined in ip_input.c.  Instead, adopt the name
'q' as found in about 1/2 of uses in ip_input.c, preventing a collision on
the name.  This is non-harmful, but means that search and replace on the
global works less well (as in the virtualization work), as well as indexing
tools.

MFC after: 1 week
Reported by: julian

16 years agoFix spelling.
Andrew Thompson [Fri, 13 Jun 2008 21:49:21 +0000 (21:49 +0000)]
Fix spelling.

PR: docs/124561
Submitted by: Sam Banks