]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
19 years agoAdd ulpt(4) and urio(4).
Simon L. B. Nielsen [Sun, 26 Sep 2004 19:06:17 +0000 (19:06 +0000)]
Add ulpt(4) and urio(4).

MFC after: 3 days

19 years agoDon't put devel/gdb6 on disc1 for non-i386 platforms. We have gdb6 in
Marcel Moolenaar [Sun, 26 Sep 2004 19:05:52 +0000 (19:05 +0000)]
Don't put devel/gdb6 on disc1 for non-i386 platforms. We have gdb6 in
the base-system. Adding the port for ia64 on disc1 is especially bogus,
because the port doesn't even build on ia64. It also doesn't support
libthread_db.

19 years ago- Add a HARDWARE section which lists supported devices based on the
Simon L. B. Nielsen [Sun, 26 Sep 2004 19:03:45 +0000 (19:03 +0000)]
- Add a HARDWARE section which lists supported devices based on the
  Hardware Notes.
- Update some text to fit better now that this driver support
  multiple different device types.

MFC after: 3 days

19 years agoEmbellish the man page after trying to migrate to gstripe from ccd(4).
David E. O'Brien [Sun, 26 Sep 2004 17:29:09 +0000 (17:29 +0000)]
Embellish the man page after trying to migrate to gstripe from ccd(4).

19 years agoDon't allow to specify wrong stripe size.
Pawel Jakub Dawidek [Sun, 26 Sep 2004 17:15:42 +0000 (17:15 +0000)]
Don't allow to specify wrong stripe size.

Reported by: obrien

19 years agoPFIL_HOOKS is no longer an optional item.
Max Laier [Sun, 26 Sep 2004 16:10:40 +0000 (16:10 +0000)]
PFIL_HOOKS is no longer an optional item.

Submitted by: Anders Hanssen
MFC after: 1 day

19 years agoPrefer C99's __func__ over GCC's __FUNCTION__.
Stefan Farfeleder [Sun, 26 Sep 2004 13:47:25 +0000 (13:47 +0000)]
Prefer C99's __func__ over GCC's __FUNCTION__.

Approved by: dwmalone

19 years agoAdd a HARDWARE section which lists supported devices based on the
Simon L. B. Nielsen [Sun, 26 Sep 2004 13:38:38 +0000 (13:38 +0000)]
Add a HARDWARE section which lists supported devices based on the
Hardware Notes.

MFC after: 3 days

19 years agoInvalidate dcons buffer address if the magic is wrong.
Hidetoshi Shimokawa [Sun, 26 Sep 2004 12:46:47 +0000 (12:46 +0000)]
Invalidate dcons buffer address if the magic is wrong.

19 years agoBackout manual page updates.
Pawel Jakub Dawidek [Sun, 26 Sep 2004 12:08:31 +0000 (12:08 +0000)]
Backout manual page updates.

Requested by: ru

19 years agoRemove the old ATA_*LOCK_CH macros that used atomic ops and use
Søren Schmidt [Sun, 26 Sep 2004 11:48:43 +0000 (11:48 +0000)]
Remove the old ATA_*LOCK_CH macros that used atomic ops and use
mutexes instead.
This closes the last (known) race issues in ATA which should fix
the various hangs etc seen on heavy loaded systems.

Change from using timeout functions to using callout functions in
the timeout code. This together with above closes the race that could
happen if timeout and device interrupt occured simultaniously.

Also fix the possible recursion in ata_reinit() on very dodgy
devices that could take us down in the probe.

19 years agoRearrange the order of I/O's in dma-start/stop, some chipsets are very
Søren Schmidt [Sun, 26 Sep 2004 11:42:42 +0000 (11:42 +0000)]
Rearrange the order of I/O's in dma-start/stop, some chipsets are very
picky on the order of this, especially in error situations.

19 years agoForced commit to note that the previous commit added /etc/namedb/s
Dag-Erling Smørgrav [Sun, 26 Sep 2004 10:50:06 +0000 (10:50 +0000)]
Forced commit to note that the previous commit added /etc/namedb/s
(where bind stores slave zones), not /etc/namedb/bind.

19 years agoPer style.Makefile(5), don't define MAN if the only page installed
Doug Barton [Sun, 26 Sep 2004 09:38:08 +0000 (09:38 +0000)]
Per style.Makefile(5), don't define MAN if the only page installed
is ${PROG}.1

Reviewed by: ru

19 years agoGet rid of obsolete MAN[0-9] syntax.
Ruslan Ermilov [Sun, 26 Sep 2004 09:25:03 +0000 (09:25 +0000)]
Get rid of obsolete MAN[0-9] syntax.

19 years agostyle: Move the { back to the else line to match the }.
Johan Karlsson [Sun, 26 Sep 2004 09:01:02 +0000 (09:01 +0000)]
style: Move the { back to the else line to match the }.

Discussed with: glebius
X-MFC after: 5.3-Release

19 years agoUpdate the BIND 9 entry with information about files that
Doug Barton [Sun, 26 Sep 2004 07:50:23 +0000 (07:50 +0000)]
Update the BIND 9 entry with information about files that
have moved, or should be removed.

19 years agoIt's not necessary to create an rndc.key file if the user already
Doug Barton [Sun, 26 Sep 2004 07:01:56 +0000 (07:01 +0000)]
It's not necessary to create an rndc.key file if the user already
has an rndc.conf file.

Submitted by: Sergey Mokryshev <mokr@mokr.net>

19 years agoAdd a wrapper for execve(). The exec'd process must be started with
Daniel Eischen [Sun, 26 Sep 2004 06:50:15 +0000 (06:50 +0000)]
Add a wrapper for execve().  The exec'd process must be started with
the signal mask and pending signals of the calling thread.  These
are stored in userland in libpthread.

There is a small race condition in this patch which could cause
problems if a signal arrives after setting the (kernel) signal
mask and before exec'ing.  The thread's set of pending signals
also are not yet installed in the exec'd process.  Both of these
will be corrected with the addition of a special syscall.

Reported & Tested by: Joost Bekkers <joost at jodocus dot org>
Reviewed by: julian, davidxu

19 years agoFix a bug introduced in the previous commit: kdb_cpu_trap() gets to
Marcel Moolenaar [Sun, 26 Sep 2004 06:48:59 +0000 (06:48 +0000)]
Fix a bug introduced in the previous commit: kdb_cpu_trap() gets to
the trapframe via kdb_frame, but kdb_frame was not initialized until
after the call to kdb_cpu_trap(). Ergo: kdb_cpu_trap() was moved too
far up.

Pointy hat: marcel

19 years agoFix up the man file installation for the new BIND 9 sources:
Doug Barton [Sun, 26 Sep 2004 06:36:11 +0000 (06:36 +0000)]
Fix up the man file installation for the new BIND 9 sources:

1. Install man files and links for the lwres library.

2. Fix the path in various files to say /etc/namedb/ instead of just /etc.

3. Correctly install the conf file man pages for named and rndc.

19 years agodetach before ivar delete.
Warner Losh [Sun, 26 Sep 2004 05:51:43 +0000 (05:51 +0000)]
detach before ivar delete.

19 years agoCorrect handling of SADB_UPDATE and SADB_ADD requests. key_align may split
Sam Leffler [Sun, 26 Sep 2004 02:01:27 +0000 (02:01 +0000)]
Correct handling of SADB_UPDATE and SADB_ADD requests.  key_align may split
the mbuf due to use of m_pulldown.  Discarding the result because of this
does not make sense as no subsequent code depends on the entire msg being
linearized (only the individual pieces).  It's likely something else is wrong
here but for now this appears to get things back to a working state.

Submitted by: Roselyn Lee

19 years agoAdd support for i/o-ports. This was cut and pasted from grackle.c
Andrew Gallatin [Sun, 26 Sep 2004 01:56:20 +0000 (01:56 +0000)]
Add support for i/o-ports. This was cut and pasted from grackle.c

19 years agoInvalidate dcons buffer on shutdown.
Hidetoshi Shimokawa [Sun, 26 Sep 2004 01:15:38 +0000 (01:15 +0000)]
Invalidate dcons buffer on shutdown.

19 years agoFix INVARIANTS build: Include <machine/cpu.h>.
Marcel Moolenaar [Sun, 26 Sep 2004 00:38:56 +0000 (00:38 +0000)]
Fix INVARIANTS build: Include <machine/cpu.h>.

19 years agoCreate /etc/namedb/bind with owner / group bind and mode 0750.
Dag-Erling Smørgrav [Sat, 25 Sep 2004 15:55:17 +0000 (15:55 +0000)]
Create /etc/namedb/bind with owner / group bind and mode 0750.

19 years agoTake the lastest fixes from NetBSD.
Pawel Jakub Dawidek [Sat, 25 Sep 2004 14:11:34 +0000 (14:11 +0000)]
Take the lastest fixes from NetBSD.

Obtained from: NetBSD

19 years agoBump document date.
Dag-Erling Smørgrav [Sat, 25 Sep 2004 11:43:08 +0000 (11:43 +0000)]
Bump document date.

19 years agoRuslan teaches me yet another useful fact. "Please sort these lexically"
Doug Barton [Sat, 25 Sep 2004 09:11:39 +0000 (09:11 +0000)]
Ruslan teaches me yet another useful fact. "Please sort these lexically"
does NOT mean in Unix-caps-first order.

19 years agoUse the floppy drive unit number to name the /dev entry, not the
Poul-Henning Kamp [Sat, 25 Sep 2004 08:56:57 +0000 (08:56 +0000)]
Use the floppy drive unit number to name the /dev entry, not the
controller unit number.

Submitted by: Rostislav Krasny <rosti_bsd@yahoo.com>

19 years agoRevert to always descending into the "bsnmp" subdir -- gensnmptree
Ruslan Ermilov [Sat, 25 Sep 2004 07:00:11 +0000 (07:00 +0000)]
Revert to always descending into the "bsnmp" subdir -- gensnmptree
and bsnmpd should not be controlled by the NOATM knob.

19 years agoAllow routing to the SCI even if it's not in the list of valid IRQs.
Nate Lawson [Sat, 25 Sep 2004 06:15:56 +0000 (06:15 +0000)]
Allow routing to the SCI even if it's not in the list of valid IRQs.

MFC if: no problems

19 years agoFix the statements for checking if we're dealing with a 5705/5750.
Paul Saab [Sat, 25 Sep 2004 05:07:20 +0000 (05:07 +0000)]
Fix the statements for checking if we're dealing with a 5705/5750.

Pointhat to: me

19 years agoMove the IA-32 trap handling from trap() to ia32_trap(). Move the
Marcel Moolenaar [Sat, 25 Sep 2004 04:27:44 +0000 (04:27 +0000)]
Move the IA-32 trap handling from trap() to ia32_trap(). Move the
ia32_syscall() function along with it to ia32_trap.c. When COMPAT_IA32
is not defined, we'll raise SIGEMT instead.

19 years agoAdd an entry to tell the world about BIND 9.
Doug Barton [Sat, 25 Sep 2004 03:41:38 +0000 (03:41 +0000)]
Add an entry to tell the world about BIND 9.

19 years agoUncomment bsnmpd so it can build.
Tom Rhodes [Sat, 25 Sep 2004 02:04:54 +0000 (02:04 +0000)]
Uncomment bsnmpd so it can build.

Noticed by: dougb

19 years agoUse the universal 'threaded process' flag rather than the
Julian Elischer [Sat, 25 Sep 2004 00:53:46 +0000 (00:53 +0000)]
Use the universal 'threaded process' flag rather than the
specific tests for different threading systems.

MFC after: 1 week

19 years agoInstall the documentation for bind9, and remove the /usr/share/doc/bind
Doug Barton [Sat, 25 Sep 2004 00:42:38 +0000 (00:42 +0000)]
Install the documentation for bind9, and remove the /usr/share/doc/bind
directory from mtree while we're at it.

Help, advice, and code from: ru, des

19 years agoFix two glitches that appear in the non-chroot case. First, if not
Doug Barton [Fri, 24 Sep 2004 23:49:38 +0000 (23:49 +0000)]
Fix two glitches that appear in the non-chroot case. First, if not
chrooted the pid symlink code should not fire. Also, remove the quotes
around the chroot variable in the rndc-confgen invocation so that if
not chrooted the command will still succeed.

Pointed out by: Sean McNeil <sean@mcneil.com>

19 years agoFor the default FreeBSD install, the file path actually is
Doug Barton [Fri, 24 Sep 2004 22:47:10 +0000 (22:47 +0000)]
For the default FreeBSD install, the file path actually is
/var/run/named/pid. This is done so that named can start
with -u bind and still dump a pid file in that directory,
which is chowned to user bind.

19 years agoAdd support for the BCM5750/5751. Unfortunately the documentation
Paul Saab [Fri, 24 Sep 2004 22:24:33 +0000 (22:24 +0000)]
Add support for the BCM5750/5751.  Unfortunately the documentation
I have from Broadcom does not give much information on these devices,
so the Broadcom Linux driver was used for clues to what these chips
support.  It turns out they are similar to the 5705 with the 5751
being the PCI-Express version and needing special work-arounds and
settings.

19 years agoFix build in the !NOATM case by using the begemot library in place of
Tom Rhodes [Fri, 24 Sep 2004 22:10:34 +0000 (22:10 +0000)]
Fix build in the !NOATM case by using the begemot library in place of
using libisc which was a part of BIND8.

Discussed with: des, re, dougb
Submitted by: harti (one part)
Reviewed by: harti (previous version)

19 years agoThis commit was generated by cvs2svn to compensate for changes in r135768,
Tom Rhodes [Fri, 24 Sep 2004 21:48:46 +0000 (21:48 +0000)]
This commit was generated by cvs2svn to compensate for changes in r135768,
which included commits to RCS files with non-trunk default branches.

19 years agoVendor import of harti's begemot library.
Tom Rhodes [Fri, 24 Sep 2004 21:48:46 +0000 (21:48 +0000)]
Vendor import of harti's begemot library.

19 years agoRuslan has educated me both on the wisdom of why this approach is
Doug Barton [Fri, 24 Sep 2004 21:30:54 +0000 (21:30 +0000)]
Ruslan has educated me both on the wisdom of why this approach is
better than mine, and why to wait for review.

Submitted by: ru

19 years agoFew (trailing) whitespace and spelling fixes.
Joerg Wunsch [Fri, 24 Sep 2004 20:33:52 +0000 (20:33 +0000)]
Few (trailing) whitespace and spelling fixes.

19 years agoProtect fdc->flags within the worker thread, too.
Joerg Wunsch [Fri, 24 Sep 2004 20:28:00 +0000 (20:28 +0000)]
Protect fdc->flags within the worker thread, too.

Use kthread_exit() instead of falling through the end of the worker
thread's main function.  Since kthread_exit() wakeup(9)s everyone
sleeping on the thread handle, drop the superfluous wakeup() call.

19 years agoSort forward declared structures.
John Baldwin [Fri, 24 Sep 2004 20:27:45 +0000 (20:27 +0000)]
Sort forward declared structures.

19 years agoSome more whitespace, style, and comment fixes.
John Baldwin [Fri, 24 Sep 2004 20:27:04 +0000 (20:27 +0000)]
Some more whitespace, style, and comment fixes.

Submitted by: bde (mostly)

19 years agoAdd a proc *p pointer for td->td_proc to make this code easier to read.
John Baldwin [Fri, 24 Sep 2004 20:26:15 +0000 (20:26 +0000)]
Add a proc *p pointer for td->td_proc to make this code easier to read.

19 years agoRemove a duplicated word.
Marc Fonvieille [Fri, 24 Sep 2004 20:09:44 +0000 (20:09 +0000)]
Remove a duplicated word.

Submitted by: Pawel Worach <pawel.worach@telia.com>

19 years agoadd my script that helps me handle MFC's. It takes in a commit message
John-Mark Gurney [Fri, 24 Sep 2004 20:06:49 +0000 (20:06 +0000)]
add my script that helps me handle MFC's.  It takes in a commit message
and generates the proper (hopefully) update -j lines + commit line to do
the MFC...  This has saved me a lot of time doing recent MFC's...

You still should use diff to verify the changes before doing the commit..

19 years agoDocument that kthread_exit()ing will cause a wakeup(9) on the thread
Joerg Wunsch [Fri, 24 Sep 2004 19:58:47 +0000 (19:58 +0000)]
Document that kthread_exit()ing will cause a wakeup(9) on the thread
handle.

MFC after: 1 week

19 years agoRetire the BIND 8 sources.
Dag-Erling Smørgrav [Fri, 24 Sep 2004 19:48:50 +0000 (19:48 +0000)]
Retire the BIND 8 sources.

19 years agoIt's named.pid, not named/pid.
Dag-Erling Smørgrav [Fri, 24 Sep 2004 19:46:18 +0000 (19:46 +0000)]
It's named.pid, not named/pid.

Pointy hat to: dougb@

19 years agoImprove the panic message for a busted MP table with conflicting entries
John Baldwin [Fri, 24 Sep 2004 18:42:54 +0000 (18:42 +0000)]
Improve the panic message for a busted MP table with conflicting entries
for the same PCI interrupt.

Tested by: Pavel Gubin pg at ie dot tusur dot ru
MFC after: 3 days

19 years agoFix the WANT_BIND_LIBS knob by correctly spelling it as WITH_BIND_LIBS
Doug Barton [Fri, 24 Sep 2004 18:42:05 +0000 (18:42 +0000)]
Fix the WANT_BIND_LIBS knob by correctly spelling it as WITH_BIND_LIBS
to match how similar syntax is used in the ports system. Thanks to kris
for pointing out my mistake here.

Install the lwres library unless the user defines NO_BIND, or the new
knob, NO_BIND_LIBS_LWRES. There is at least one potential customer
for this library in the wings. Thanks to nectar for the reminder.

19 years agoAllow %' to be used as a format flag by printf(1). This makes it
Giorgos Keramidas [Fri, 24 Sep 2004 18:20:43 +0000 (18:20 +0000)]
Allow %' to be used as a format flag by printf(1).  This makes it
possible to print the thousands separator in the locale setups that
have one, by something like this:

    $ env -i LC_NUMERIC=en_US.ISO8859-1 ./printf "%'0.2f\n" 12345
    12,345.00

Reviewed by: das

19 years agoPrint number of "waits" per second during transmission, not just waits
Robert Watson [Fri, 24 Sep 2004 18:02:31 +0000 (18:02 +0000)]
Print number of "waits" per second during transmission, not just waits
per call to send().

19 years agoReconnect namedb. While it may not be optimal, our old named.conf from
Dag-Erling Smørgrav [Fri, 24 Sep 2004 16:13:55 +0000 (16:13 +0000)]
Reconnect namedb.  While it may not be optimal, our old named.conf from
BIND 8 is quite usable for BIND 9.

19 years agoDon't look for swap blocks in objects that aren't swap-backed.
David Schultz [Fri, 24 Sep 2004 16:04:20 +0000 (16:04 +0000)]
Don't look for swap blocks in objects that aren't swap-backed.
I expect that this will fix the following panic, reported by Jun:
swap_pager_isswapped: failed to locate all swap meta blocks

MT5 candidate

19 years agoMore remnants of BIND 8; these live in usr.sbin now.
Dag-Erling Smørgrav [Fri, 24 Sep 2004 15:39:01 +0000 (15:39 +0000)]
More remnants of BIND 8; these live in usr.sbin now.

19 years agoAdd mfc.
Dag-Erling Smørgrav [Fri, 24 Sep 2004 15:25:35 +0000 (15:25 +0000)]
Add mfc.

19 years agoA simple shell script to help MFC an entire directory to a branch where it
Dag-Erling Smørgrav [Fri, 24 Sep 2004 15:24:12 +0000 (15:24 +0000)]
A simple shell script to help MFC an entire directory to a branch where it
does not already exist.

19 years agoBreak the NO_BIND knob out from the pack, and begin documenting
Doug Barton [Fri, 24 Sep 2004 13:51:31 +0000 (13:51 +0000)]
Break the NO_BIND knob out from the pack, and begin documenting
the finer granularity for the various BIND features.

19 years agoDon't expose BIND libraries and their headers to the public by default,
Ruslan Ermilov [Fri, 24 Sep 2004 13:42:00 +0000 (13:42 +0000)]
Don't expose BIND libraries and their headers to the public by default,
but have a knob (WANT_BIND_LIBS) to build and install them in /usr/lib
and /usr/include.  Rumors are that this may be useful at a later point,
let's see.

What this really means is that all BIND libraries are now internal to
buildworld (by default, unless WANT_BIND_LIBS is defined), and linked
statically into various BIND executables.

While here, removed redundant -I's from CFLAGS in lib/bind makefiles.

Sponsored by: des
OK'ed by: dougb

19 years agoIndent.
Maxim Konovalov [Fri, 24 Sep 2004 13:22:45 +0000 (13:22 +0000)]
Indent.

19 years agoo Merge rev. 1.5 libexec/ftpd/ftpd.c from DragonflyBSD:
Maxim Konovalov [Fri, 24 Sep 2004 13:21:52 +0000 (13:21 +0000)]
o Merge rev. 1.5 libexec/ftpd/ftpd.c from DragonflyBSD:

  Do not unconditionally fork() after accept().  accept() can
  return -1 due to an interrupted system call (i.e. SIGCHLD).
  If we fork in that case ftpd can get into an
  accept()/SIGCHLD/fork/[fail]/repeat loop.

Reported-by: fabian <fabian.duelli@bluewin.ch>
Obtained from: DragonflyBSD
MFC after: 1 month

19 years agoLock the flags field with the mutex.
Poul-Henning Kamp [Fri, 24 Sep 2004 13:04:49 +0000 (13:04 +0000)]
Lock the flags field with the mutex.

Improve a number of comments.

19 years ago_tcb_ctor takes two args.
Olivier Houchard [Fri, 24 Sep 2004 13:02:30 +0000 (13:02 +0000)]
_tcb_ctor takes two args.

19 years agoAdd ttyu0 as a serial console, as we're using the uart(4) driver on arm.
Olivier Houchard [Fri, 24 Sep 2004 12:51:15 +0000 (12:51 +0000)]
Add ttyu0 as a serial console, as we're using the uart(4) driver on arm.

19 years agoSync with DragonFly BSD.
Hidetoshi Shimokawa [Fri, 24 Sep 2004 12:43:57 +0000 (12:43 +0000)]
Sync with DragonFly BSD.

19 years agoo Turn net.inet.ip.check_interface sysctl off by default.
Maxim Konovalov [Fri, 24 Sep 2004 12:18:40 +0000 (12:18 +0000)]
o Turn net.inet.ip.check_interface sysctl off by default.

When net.inet.ip.check_interface was MFCed to RELENG_4 3+ years ago in
rev. 1.130.2.17 ip_input.c it was 1 by default but shortly changed to
0 (accidently?) in rev. 1.130.2.20 in RELENG_4 only.  Among with the
fact this knob is not documented it breaks POLA especially in bridge
environment.

OK'ed by: andre
Reviewed by: -current

19 years agoFix a comment typo: s/neccessary/necessary/
Giorgos Keramidas [Fri, 24 Sep 2004 11:04:27 +0000 (11:04 +0000)]
Fix a comment typo: s/neccessary/necessary/

19 years agoDocument vfs.root.mountdelay value.
Pawel Jakub Dawidek [Fri, 24 Sep 2004 09:33:48 +0000 (09:33 +0000)]
Document vfs.root.mountdelay value.

Reminded by: ru

19 years agoRename 'mount_root_delay' tunable to 'vfs.root.mountdelay', which fits
Pawel Jakub Dawidek [Fri, 24 Sep 2004 09:19:03 +0000 (09:19 +0000)]
Rename 'mount_root_delay' tunable to 'vfs.root.mountdelay', which fits
a bit better to our current naming scheme.

Discussed with: ru

19 years agoXXX mark two places where we do not hold a threadcount on the dev when
Poul-Henning Kamp [Fri, 24 Sep 2004 08:32:36 +0000 (08:32 +0000)]
XXX mark two places where we do not hold a threadcount on the dev when
frobbing the cdevsw.

In both cases we examine only the cdevsw and it is a good question if we
weren't better off copying those properties into the cdev in the first
place.  This question will be revisited.

19 years agoRemove the cdevsw() function which is now unused.
Poul-Henning Kamp [Fri, 24 Sep 2004 08:30:57 +0000 (08:30 +0000)]
Remove the cdevsw() function which is now unused.

19 years agoHold threadcount while throbbing cdevsw in our underlying driver.
Poul-Henning Kamp [Fri, 24 Sep 2004 08:26:03 +0000 (08:26 +0000)]
Hold threadcount while throbbing cdevsw in our underlying driver.

This is a bit heavyhanded, and will be simplified once the tty code
learns to properly deal with disappearing hw and drivers.

19 years agoHold thread reference while we frob cdevsw.
Poul-Henning Kamp [Fri, 24 Sep 2004 08:12:41 +0000 (08:12 +0000)]
Hold thread reference while we frob cdevsw.

19 years agoes_ES monetdef changed from spanish Ptas. to Euro without
Andrey A. Chernov [Fri, 24 Sep 2004 07:48:05 +0000 (07:48 +0000)]
es_ES monetdef changed from spanish Ptas. to Euro without
switching to cents

PR:             67549
Submitted by:   Jose M Rodriguez <freebsd@wanadoo.es>

19 years agoHold proper thread count while frobbing drivers ioctl.
Poul-Henning Kamp [Fri, 24 Sep 2004 07:24:02 +0000 (07:24 +0000)]
Hold proper thread count while frobbing drivers ioctl.

19 years agoHold threadcount reference when we call into the underlying console
Poul-Henning Kamp [Fri, 24 Sep 2004 07:16:56 +0000 (07:16 +0000)]
Hold threadcount reference when we call into the underlying console
driver.

19 years agoEliminate devsw() call, we are not dereferencing the pointer.
Poul-Henning Kamp [Fri, 24 Sep 2004 07:11:02 +0000 (07:11 +0000)]
Eliminate devsw() call, we are not dereferencing the pointer.

19 years agoRemove devsw() call missed in last commit.
Poul-Henning Kamp [Fri, 24 Sep 2004 07:08:33 +0000 (07:08 +0000)]
Remove devsw() call missed in last commit.

19 years agoDon't read the byte at *argv[strlen(*argv) + 1].
David Schultz [Fri, 24 Sep 2004 06:57:26 +0000 (06:57 +0000)]
Don't read the byte at *argv[strlen(*argv) + 1].

PR: 71743

19 years agoRemove SI_ISDISK, I found a better solution.
Poul-Henning Kamp [Fri, 24 Sep 2004 06:55:32 +0000 (06:55 +0000)]
Remove SI_ISDISK, I found a better solution.

19 years agoAssert topology is held in g_dev_getprovider().
Poul-Henning Kamp [Fri, 24 Sep 2004 06:43:20 +0000 (06:43 +0000)]
Assert topology is held in g_dev_getprovider().

Don't call devsw().  It is not necessary, and we do not need to hold dev_lock
to compare the devsw pointer to our own since we do not dereference it.

19 years agoHold thread reference while frobbing cdevsw.
Poul-Henning Kamp [Fri, 24 Sep 2004 06:37:00 +0000 (06:37 +0000)]
Hold thread reference while frobbing cdevsw.

19 years agoMake sure we don't call _thr_start_sig_daemon() when SYSTEM_SCOPE_ONLY is defined...
Suleiman Souhlal [Fri, 24 Sep 2004 06:36:31 +0000 (06:36 +0000)]
Make sure we don't call _thr_start_sig_daemon() when SYSTEM_SCOPE_ONLY is defined. This makes libpthread usable on powerpc.

Approved by: grehan (mentor), deischen

19 years agoRemove __isnanl() to avoid conflicting with the same function in libm.
David Schultz [Fri, 24 Sep 2004 06:33:22 +0000 (06:33 +0000)]
Remove __isnanl() to avoid conflicting with the same function in libm.
__isnan() and __isnanf() must remain in libc for hysterical raisins.
On the other hand, __isnanl() must live in libm because libm uses it
internally and can't depend on older versions of libc to provide it.
Fortunately, we don't need __isnanl() in both libraries.

Prodded by: ale
PR: 71698
MT5 candidate

19 years agoHold threadref while we throb cdevsw in devtoname()
Poul-Henning Kamp [Fri, 24 Sep 2004 06:29:23 +0000 (06:29 +0000)]
Hold threadref while we throb cdevsw in devtoname()

19 years agoUse vn_isdisk() to check if vnode is a disk.
Poul-Henning Kamp [Fri, 24 Sep 2004 06:23:31 +0000 (06:23 +0000)]
Use vn_isdisk() to check if vnode is a disk.

(repeat, CVS core dumped on me)

19 years agouse vn_isdisk() to see if vnode is a disk.
Poul-Henning Kamp [Fri, 24 Sep 2004 06:21:43 +0000 (06:21 +0000)]
use vn_isdisk() to see if vnode is a disk.

19 years agoHold dev_lock and check for NULL devsw pointer when we service FIODTYPE ioctl.
Poul-Henning Kamp [Fri, 24 Sep 2004 06:16:48 +0000 (06:16 +0000)]
Hold dev_lock and check for NULL devsw pointer when we service FIODTYPE ioctl.

19 years agoHold dev_lock and check for NULL devsw pointer when we determine
Poul-Henning Kamp [Fri, 24 Sep 2004 06:16:08 +0000 (06:16 +0000)]
Hold dev_lock and check for NULL devsw pointer when we determine
if a vnode is a disk.

19 years agoUse dev_re[fl]thread() to maintain a ref on the device driver while
Poul-Henning Kamp [Fri, 24 Sep 2004 05:59:11 +0000 (05:59 +0000)]
Use dev_re[fl]thread() to maintain a ref on the device driver while
we call the ->d_mmap function.

19 years agoUse def_re[fl]thread().
Poul-Henning Kamp [Fri, 24 Sep 2004 05:58:06 +0000 (05:58 +0000)]
Use def_re[fl]thread().

Retire various old compatibility helpers.

19 years agouse dev_re[fl]thread() rather than home rolled versions.
Poul-Henning Kamp [Fri, 24 Sep 2004 05:55:03 +0000 (05:55 +0000)]
use dev_re[fl]thread() rather than home rolled versions.