]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
15 years agoConvert gdt_segs and ldt_segs initialization to C99 style.
Konstantin Belousov [Thu, 26 Mar 2009 18:07:13 +0000 (18:07 +0000)]
Convert gdt_segs and ldt_segs initialization to C99 style.

Reviewed by: jhb

15 years agoAllow the attach routine to fail gracefully and not panic the system.
Warner Losh [Thu, 26 Mar 2009 17:36:19 +0000 (17:36 +0000)]
Allow the attach routine to fail gracefully and not panic the system.

15 years agoAdd stuff to support upcoming BMC/IPMI flashing of newer Dell machine
Doug Ambrisko [Thu, 26 Mar 2009 17:14:22 +0000 (17:14 +0000)]
Add stuff to support upcoming BMC/IPMI flashing of newer Dell machine
via the Linux tool.
     -  Add Linux shim to ipmi(4)
     -  Create a partitions file to linprocfs to make Linux fdisk see
        disks.  This file is dynamic so we can see disks come and go.
     -  Convert msdosfs to vfat in mtab since Linux uses that for
        msdosfs.
     -  In the Linux mount path convert vfat passed in to msdosfs
        so Linux mount works on FreeBSD.  Note that tasting works
        so that if da0 is a msdos file system
                /compat/linux/bin/mount /dev/da0 /mnt
        works.
     -  fix a 64it bug for l_off_t.
Grabing sh, mount, fdisk, df from Linux, creating a symlink of mtab to
/compat/linux/etc/mtab and then some careful unpacking of the Linux bmc
update tool and hacking makes it work on newer Dell boxes.  Note, probably
if you can't figure out how to do this, then you probably shouldn't be
doing it :-)

15 years agoChange the priority from high to normal. This makes sure that
Marcel Moolenaar [Thu, 26 Mar 2009 16:42:24 +0000 (16:42 +0000)]
Change the priority from high to normal. This makes sure that
the BSD or GPT schemes can take precedence as appropriate.

15 years agothere is no dev_mkdb anymore
Luigi Rizzo [Thu, 26 Mar 2009 10:22:40 +0000 (10:22 +0000)]
there is no dev_mkdb anymore

15 years agobump date.
Weongyo Jeong [Thu, 26 Mar 2009 06:12:04 +0000 (06:12 +0000)]
bump date.

Pointed by: yongari

15 years agoadds malo(4).
Weongyo Jeong [Thu, 26 Mar 2009 04:17:35 +0000 (04:17 +0000)]
adds malo(4).

15 years agobwi(4) and mwl(4) doesn't exist at CURRENT.
Weongyo Jeong [Thu, 26 Mar 2009 04:16:40 +0000 (04:16 +0000)]
bwi(4) and mwl(4) doesn't exist at CURRENT.

15 years agoThe malo(4) device driver first appeared in FreeBSD 7.1 not 8.0.
Weongyo Jeong [Thu, 26 Mar 2009 02:59:10 +0000 (02:59 +0000)]
The malo(4) device driver first appeared in FreeBSD 7.1 not 8.0.

Reported by: John <lists_at_reiteration.net>
MFC after: 3 days

15 years agoFix up waiting on vblank again... This reverts a last minute change that
Robert Noland [Thu, 26 Mar 2009 02:10:18 +0000 (02:10 +0000)]
Fix up waiting on vblank again... This reverts a last minute change that
I made on the last patch, it seems to upset suspend/resume and shutdown.

MFC after: 3 days

15 years agoTweak comments.
Warner Losh [Wed, 25 Mar 2009 22:21:53 +0000 (22:21 +0000)]
Tweak comments.

15 years agoAdd RIOS PCCARD 3.
Warner Losh [Wed, 25 Mar 2009 22:21:38 +0000 (22:21 +0000)]
Add RIOS PCCARD 3.

15 years agoRIOS Systems co was naughty and used a 'random' id. Use it with the
Warner Losh [Wed, 25 Mar 2009 22:20:36 +0000 (22:20 +0000)]
RIOS Systems co was naughty and used a 'random' id.  Use it with the
PCCARD3 name to describe the RIOS PC Card III Ethernet that I have.

15 years agoAdd recently added cards.
Warner Losh [Wed, 25 Mar 2009 22:14:46 +0000 (22:14 +0000)]
Add recently added cards.

15 years agoIt appears that the TDK LAK CD011 works with fe driver. I don't have
Warner Losh [Wed, 25 Mar 2009 22:12:07 +0000 (22:12 +0000)]
It appears that the TDK LAK CD011 works with fe driver.  I don't have
a dongle for the card, but it passes the sanity testsin the
probe/attach routines and returns the correct MAC address.

15 years agoFix a few nits in the earlier changes to prevent local information leakage
John Baldwin [Wed, 25 Mar 2009 22:08:30 +0000 (22:08 +0000)]
Fix a few nits in the earlier changes to prevent local information leakage
in AMD FPUs:
- Do not clear the affected state in the case that the FPU registers for
  the thread that already owns the FPU are changed via fpu_setregs().  The
  only local information the thread would see is its own state in that
  case.
- Fix a type mismatch for the dummy variable used in a "fld".  It accepts
  a float, not a double.

Reviewed by: bde
Approved by: so (cperciva)
MFC after: 1 month

15 years agoupdate; mostly to eliminate ambiguity about what hardware is supported
Sam Leffler [Wed, 25 Mar 2009 21:20:15 +0000 (21:20 +0000)]
update; mostly to eliminate ambiguity about what hardware is supported

15 years agoMan page accompanying r190423 - introduce UFS ID labels.
Ivan Voras [Wed, 25 Mar 2009 20:40:34 +0000 (20:40 +0000)]
Man page accompanying r190423 - introduce UFS ID labels.

Actually-by: pjd
Approved by: gnn (mentor)

15 years agoCreate GEOM labels from UFS IDs, e.g. /dev/ufsid/49c97b1faa2adc43. UFS IDs
Ivan Voras [Wed, 25 Mar 2009 20:38:57 +0000 (20:38 +0000)]
Create GEOM labels from UFS IDs, e.g. /dev/ufsid/49c97b1faa2adc43. UFS IDs
are always present and can be used to identify file systems (useful if
hardware devices move often).

Actually-by: pjd
Approved by: gnn (mentor)

15 years ago- Add proper error checking and printing to the CVSMode code when reading and
Ulf Lilleengen [Wed, 25 Mar 2009 20:15:48 +0000 (20:15 +0000)]
- Add proper error checking and printing to the CVSMode code when reading and
  writing from/to streams, as leaving them out stops csup from cleaning up on
  SIGINT and friends properly.

MFC after:      1 week

15 years agoThis config file uses dynamically linked binaries, which is more
Luigi Rizzo [Wed, 25 Mar 2009 18:48:46 +0000 (18:48 +0000)]
This config file uses dynamically linked binaries, which is more
convenient when you want to import other programs because the
libraries will not be replicated.

Given that there are no floppies around anymore, I have bumped
the fd size to 4MB (which is more than reasonable even for
embedded platforms) and gives some room for other utilities.

MFC after: 3 days

15 years agodevfs is the standard now
Luigi Rizzo [Wed, 25 Mar 2009 17:56:28 +0000 (17:56 +0000)]
devfs is the standard now

15 years agoMove dtnfsclient.c in the cddl tree to nfs_kdtrace.c in the nfsclient
Robert Watson [Wed, 25 Mar 2009 17:47:22 +0000 (17:47 +0000)]
Move dtnfsclient.c in the cddl tree to nfs_kdtrace.c in the nfsclient
directory, since it's under a BSD license, and this keeps NFS internals-
aware tracing parts close to NFS.

MFC after: 1 month
Suggested by: jhb

15 years agoWhen looking up the parent devclass of a new devclass, create the parent
John Baldwin [Wed, 25 Mar 2009 17:02:05 +0000 (17:02 +0000)]
When looking up the parent devclass of a new devclass, create the parent
devclass if it doesn't already exist.

15 years agoAdd support for SCTP to getaddrinfo(3).
Hajimu UMEMOTO [Wed, 25 Mar 2009 16:23:43 +0000 (16:23 +0000)]
Add support for SCTP to getaddrinfo(3).
Now, getaddrinfo(3) returns two SOCK_STREAMs, IPPROTO_TCP and
IPPROTO_SCTP.  It confuses some programs.  If getaddrinfo(3) returns
IPPROTO_SCTP when SOCK_STREAM is specified by hints.ai_socktype, at
least Apache doesn't work.  So, I made getaddrinfo(3) to return
IPPROTO_SCTP with SOCK_STREAM only when IPPROTO_SCTP is specified
explicitly by hints.ai_protocol.

PR: bin/128167
Submitted by: Bruce Cran <bruce__at__cran.org.uk> (partly)
MFC after: 2 week

15 years agoRename (fpu|npx)_cleanstate to (fpu|npx)_initialstate to better reflect
John Baldwin [Wed, 25 Mar 2009 14:17:08 +0000 (14:17 +0000)]
Rename (fpu|npx)_cleanstate to (fpu|npx)_initialstate to better reflect
their purpose.

Inspired by: bde
MFC after: 1 month

15 years agoadd support for shared "crunch" binary.
Luigi Rizzo [Wed, 25 Mar 2009 14:02:26 +0000 (14:02 +0000)]
add support for shared "crunch" binary.

15 years agoImprove the German translation.
Alexander Leidinger [Wed, 25 Mar 2009 12:36:37 +0000 (12:36 +0000)]
Improve the German translation.
Some parts (ERANGE, ETXTBSY) triggered by Christoph Mallon.

Discussed with: brueffer, gabor (previous version)

15 years agoMake programs that define a macro called `dprintf' more likely to work.
David Schultz [Wed, 25 Mar 2009 08:07:52 +0000 (08:07 +0000)]
Make programs that define a macro called `dprintf' more likely to work.

15 years agoAnother akihabra find: Mitsubishi B8895.
Warner Losh [Wed, 25 Mar 2009 07:26:24 +0000 (07:26 +0000)]
Another akihabra find: Mitsubishi B8895.

15 years agoFix typo.
Pyun YongHyeon [Wed, 25 Mar 2009 07:05:23 +0000 (07:05 +0000)]
Fix typo.

15 years ago- Remember to set umask before setting attributes of RCS file.
Ulf Lilleengen [Wed, 25 Mar 2009 07:01:45 +0000 (07:01 +0000)]
- Remember to set umask before setting attributes of RCS file.

Tested by: dougb
MFC after: 2 days

15 years agoUse __packed from cdefs.h
Xin LI [Wed, 25 Mar 2009 06:27:56 +0000 (06:27 +0000)]
Use __packed from cdefs.h

Submitted by: pyunyh
Reviewed by: Shaowei WANG <wsw1wsw2 gmail com>
MFC after: 2 weeks

15 years agoMerge an important change that I mistakenly left out when merging C99
David Schultz [Wed, 25 Mar 2009 05:10:32 +0000 (05:10 +0000)]
Merge an important change that I mistakenly left out when merging C99
inline function support. This should fix instances where gcc
spuriously reports the following error:

    error: nested function 'foo' declared but never defined

15 years agoDisable ATA DMA for ATAPI devices for now. Apparently, certain revisions
Nathan Whitehorn [Wed, 25 Mar 2009 03:42:49 +0000 (03:42 +0000)]
Disable ATA DMA for ATAPI devices for now. Apparently, certain revisions
of this controller, in combination with certain ATAPI devices and phases
of the moon, will cause DMA operations for ATAPI to fail.

15 years agoshuffle code so things build w/o INVARIANTS or IEEE80211_SUPPORT_SUPERG
Sam Leffler [Wed, 25 Mar 2009 03:02:03 +0000 (03:02 +0000)]
shuffle code so things build w/o INVARIANTS or IEEE80211_SUPPORT_SUPERG

Prodded by: several

15 years agoRework the management of vblank interrupts a bit.
Robert Noland [Wed, 25 Mar 2009 01:50:56 +0000 (01:50 +0000)]
Rework the management of vblank interrupts a bit.

When a vt switch occurs the irq handler is uninstalled.  Interrupts
and the state tracking of what was enabled/disabled wasn't working
properly.  This should resolve the reports of "slow windows" after a
vt switch, among other things.  The radeon 2d driver seems to work a
bit more correctly than the Intel driver.  With the Intel driver,
vblank interrupts will be enabled at system startup and will only
be disabled after an additional modeset (vt switch, dpms, randr event).

With this patch, I am able to run glxgears synced to vblank and
vt switch while it is running without ill effects.

MFC after: 3 days

15 years agoIntel handled the management of the breadcrumb counter inconsistently.
Robert Noland [Wed, 25 Mar 2009 01:44:16 +0000 (01:44 +0000)]
Intel handled the management of the breadcrumb counter inconsistently.
Make sure that we always handle it the same way.

MFC after: 3 days

15 years agoThe GART allocations are a propery of the gart, not of scatter-gather
Robert Noland [Wed, 25 Mar 2009 01:41:56 +0000 (01:41 +0000)]
The GART allocations are a propery of the gart, not of scatter-gather
memory.  Track them in the appropriate structure.

MFC after: 3 days

15 years agoUse __attribute__((packed)) for the structure so that hptiop management
Xin LI [Wed, 25 Mar 2009 00:57:09 +0000 (00:57 +0000)]
Use __attribute__((packed)) for the structure so that hptiop management
utility would work.

Submitted by: Shaowei WANG <wsw1wsw2 gmail com>
MFC after: 2 weeks

15 years ago- Add Belarusian catalog
Gabor Kovesdan [Tue, 24 Mar 2009 23:31:41 +0000 (23:31 +0000)]
- Add Belarusian catalog

PR: conf/133004
Submitted by: Tatsiana Elavaya <t.elavaya@gmail.com>

15 years agoFix two bugs in DTrace tracing of accesscache and attrcache load events:
Robert Watson [Tue, 24 Mar 2009 23:16:48 +0000 (23:16 +0000)]
Fix two bugs in DTrace tracing of accesscache and attrcache load events:

- Trace non-error loads into the access cache once, not zero times or
  twice.
- Sometimes attr cache loads fail due to a race, in which case they are
  aborted leading to an invalidation; in this case, trace only the flush,
  not a load.

MFC after: 1 month
Sponsored by: Google, Inc.

15 years agoAllow compile from c++ for libpmc
Fabien Thomas [Tue, 24 Mar 2009 22:35:05 +0000 (22:35 +0000)]
Allow compile from c++ for libpmc

Approved by: jkoshy (mentor)
MFC after: 3 days

15 years agohide linker set stuff from user space
Sam Leffler [Tue, 24 Mar 2009 21:02:19 +0000 (21:02 +0000)]
hide linker set stuff from user space

15 years agosplit Atheros SuperG support out into it's own file that's included only
Sam Leffler [Tue, 24 Mar 2009 20:39:08 +0000 (20:39 +0000)]
split Atheros SuperG support out into it's own file that's included only
with a new IEEE80211_SUPPORT_SUPERG option

15 years agoChange yet another script to use /dev/mdX instead of /dev/mdXc (see
Marius Strobl [Tue, 24 Mar 2009 20:00:28 +0000 (20:00 +0000)]
Change yet another script to use /dev/mdX instead of /dev/mdXc (see
also r188438).

15 years agoWhen a file lookup fails due to encountering a doomed vnode from a forced
John Baldwin [Tue, 24 Mar 2009 18:16:42 +0000 (18:16 +0000)]
When a file lookup fails due to encountering a doomed vnode from a forced
unmount, consistently return ENOENT rather than EBADF.

Reviewed by: kib
MFC after: 1 month

15 years agoFall back to using configuration type 1 accesses for PCI config requests if
John Baldwin [Tue, 24 Mar 2009 18:10:22 +0000 (18:10 +0000)]
Fall back to using configuration type 1 accesses for PCI config requests if
the requested PCI bus falls outside of the bus range given in the ACPI
MCFG table.  Several BIOSes seem to not include all of the PCI busses in
systems in their MCFG tables.  It maybe that the BIOS is simply buggy and
does support all the busses, but it is more conservative to just fall back
to the old method unless it is certain that memory accesses will work.

15 years agoremove stale comments on the size of objects, they are more
Luigi Rizzo [Tue, 24 Mar 2009 18:02:35 +0000 (18:02 +0000)]
remove stale comments on the size of objects, they are more
than 10 years old and do not reflect reality anymore.

Also remove some commented out entries that have no hope to
be useful anymore.

15 years agoadd linker sets for get/set ioctl handlers so optional net80211
Sam Leffler [Tue, 24 Mar 2009 17:57:48 +0000 (17:57 +0000)]
add linker sets for get/set ioctl handlers so optional net80211
code is isolated

15 years agoadd a function to help copying shared binaries from the output
Luigi Rizzo [Tue, 24 Mar 2009 17:47:50 +0000 (17:47 +0000)]
add a function to help copying shared binaries from the output
of a buildworld.

15 years agogetaddrinfo(3) should accept numeric when ai_socktype is not
Hajimu UMEMOTO [Tue, 24 Mar 2009 17:47:24 +0000 (17:47 +0000)]
getaddrinfo(3) should accept numeric when ai_socktype is not
specified in hint or hints is NULL.

PR: bin/51827
Submitted by: Mark Andrews <marka__at__isc.org>
MFC after: 1 week

15 years agouse IEEE80211_SUPPORT_TDMA option to control whether code is
Sam Leffler [Tue, 24 Mar 2009 17:22:10 +0000 (17:22 +0000)]
use IEEE80211_SUPPORT_TDMA option to control whether code is
configured; this allows us to remove an #ifdef

15 years agoAdd DTrace probes to the NFS access and attribute caches. Access cache
Robert Watson [Tue, 24 Mar 2009 17:14:34 +0000 (17:14 +0000)]
Add DTrace probes to the NFS access and attribute caches.  Access cache
events are:

  nfsclient:accesscache:flush:done
  nfsclient:accesscache:get:hit
  nfsclient:accesscache:get:miss
  nfsclient:accesscache:load:done

They pass the vnode, uid, and requested or loaded access mode (if any);
the load event may also report a load error if the RPC fails.

The attribute cache events are:

  nfsclient:attrcache:flush:done
  nfsclient:attrcache:get:hit
  nfsclient:attrcache:get:miss
  nfsclient:attrcache:load:done

They pass the vnode, optionally the vattr if one is present (hit or load),
and in the case of a load event, also a possible RPC error.

MFC after: 1 month
Sponsored by: Google, Inc.

15 years agorequire wlan for ancillary modules; they are required to link
Sam Leffler [Tue, 24 Mar 2009 17:10:42 +0000 (17:10 +0000)]
require wlan for ancillary modules; they are required to link

15 years agoremove unused "package" entry.
Luigi Rizzo [Tue, 24 Mar 2009 16:03:46 +0000 (16:03 +0000)]
remove unused "package" entry.
default to use devfs (we do that anyways)

Many small changes in comments

15 years agoFixed indentation for LINKTYPE_ENC
VANHULLEBUS Yvan [Tue, 24 Mar 2009 15:57:35 +0000 (15:57 +0000)]
Fixed indentation for LINKTYPE_ENC

Approved by: gnn(mentor)

15 years agoRevert unintended part of r190375.
Marius Strobl [Tue, 24 Mar 2009 15:56:15 +0000 (15:56 +0000)]
Revert unintended part of r190375.

15 years agoGiven that zfs.ko depends on opensolaris.ko also build the latter by
Marius Strobl [Tue, 24 Mar 2009 15:52:45 +0000 (15:52 +0000)]
Given that zfs.ko depends on opensolaris.ko also build the latter by
default on sparc64.

15 years agoAs with ZFS use real atomic operations for sparc64.
Marius Strobl [Tue, 24 Mar 2009 15:48:35 +0000 (15:48 +0000)]
As with ZFS use real atomic operations for sparc64.

15 years agoBump __FreeBSD_version to reflect tcpdump 4.0.0 and libpcap 1.0.0
Rui Paulo [Tue, 24 Mar 2009 13:11:16 +0000 (13:11 +0000)]
Bump __FreeBSD_version to reflect tcpdump 4.0.0 and libpcap 1.0.0
import.

15 years agoMFV of tzdata2009d:
Edwin Groothuis [Tue, 24 Mar 2009 10:50:30 +0000 (10:50 +0000)]
MFV of tzdata2009d:

- Morocco will observe DST from 2009-06-01 00:00 to 2009-08-21 00:00
- Tunisia will not observe DST this year.
- Syria will start DST on 2009-03-27 00:00 this year
- the Province of San Luis will go to utc-04:00

MFC after: 1 week

15 years agoVendor import of tzdata2009d
Edwin Groothuis [Tue, 24 Mar 2009 10:47:53 +0000 (10:47 +0000)]
Vendor import of tzdata2009d

- Morocco will observe DST from 2009-06-01 00:00 to 2009-08-21 00:00
- Tunisia will not observe DST this year.
- Syria will start DST on 2009-03-27 00:00 this year
- the Province of San Luis will go to utc-04:00

15 years agoset NULL after free to avoid duplicate free.
Weongyo Jeong [Tue, 24 Mar 2009 04:20:17 +0000 (04:20 +0000)]
set NULL after free to avoid duplicate free.

Tested by: Ganbold <ganbold_at_micom.mng.net>

15 years agoDon't call m_freem() after ip_output(), as it always consumes
Bruce M Simpson [Tue, 24 Mar 2009 01:22:12 +0000 (01:22 +0000)]
Don't call m_freem() after ip_output(), as it always consumes
the mbuf chain provided to it.

Found by: Pierre Guinoiseau

15 years agofix build w/ AH_DEBUG
Sam Leffler [Tue, 24 Mar 2009 00:09:35 +0000 (00:09 +0000)]
fix build w/ AH_DEBUG

15 years agoRemove gratuitous unlock in error case.
Juli Mallett [Tue, 24 Mar 2009 00:08:58 +0000 (00:08 +0000)]
Remove gratuitous unlock in error case.

Reviewed by: sam

15 years agoSave errno before calling _close(), which may clear it.
Xin LI [Mon, 23 Mar 2009 23:43:07 +0000 (23:43 +0000)]
Save errno before calling _close(), which may clear it.

Obtained from: OpenBSD

15 years agouse more proper format string.
Xin LI [Mon, 23 Mar 2009 23:22:09 +0000 (23:22 +0000)]
use more proper format string.

Obtained from: NetBSD via OpenBSD

15 years ago- Clean up suspend/resume code for amd64.
Jung-uk Kim [Mon, 23 Mar 2009 22:35:30 +0000 (22:35 +0000)]
- Clean up suspend/resume code for amd64.
- Call acpi_resync_clock() to reset system time before hardclock is ready
to tick.  Note we assume the current timecounter hardware and RTC are
already available for read operation.

Tested by: mav

15 years agoAdd a function to reset system time after resuming, which will be used
Jung-uk Kim [Mon, 23 Mar 2009 22:12:33 +0000 (22:12 +0000)]
Add a function to reset system time after resuming, which will be used
by amd64 shortly.  It can be turned off by setting "debug.acpi.reset_clock"
tunable to zero.

15 years agoCheck whether devd is running before calling resume notifier and
Jung-uk Kim [Mon, 23 Mar 2009 22:06:09 +0000 (22:06 +0000)]
Check whether devd is running before calling resume notifier and
reshuffle code to reduce unnecessary locking coverage.

15 years agoClean up MI inittodr(9) and kill noop code.
Jung-uk Kim [Mon, 23 Mar 2009 21:16:21 +0000 (21:16 +0000)]
Clean up MI inittodr(9) and kill noop code.

It was derived from i386 version long ago but never resync'ed again.
Originally, i386 version compared the current time from realtime clock
with time_second (which was just `time' in the old days).  When this MI
version was written, it was wrongly compared against `base' AND never
used because of a bug (typo?) in the code.  This check was killed
in i386 version when home-rolled calendaric calculation was removed.
Now, we just remove the code here as well to make the code simpler.

15 years agoImprove the description of a few sysctls.
John Baldwin [Mon, 23 Mar 2009 20:18:06 +0000 (20:18 +0000)]
Improve the description of a few sysctls.

Submitted by: bde (partially)
MFC after: 3 days

15 years agoMinor updates to the Chelsio driver, including removing an LOR.
George V. Neville-Neil [Mon, 23 Mar 2009 19:58:26 +0000 (19:58 +0000)]
Minor updates to the Chelsio driver, including removing an LOR.

Submitted by: Navdeep Parhar at Chelsio
Reviewed by: gnn
MFC after:  3 weeks

15 years agoReenable 64-bit DMA for twa(4) controllers, but use a boundary of 4GB to
John Baldwin [Mon, 23 Mar 2009 19:27:23 +0000 (19:27 +0000)]
Reenable 64-bit DMA for twa(4) controllers, but use a boundary of 4GB to
prevent individual transactions from crossing a 4GB address boundary.  Due
to bus_size_t type limitations, the driver uses a 2GB boundary in PAE
kernels.

Reviewed by: scottl
MFC after: 1 week

15 years agoFix order of debug printf items, addr and config# were swapped.
Andrew Thompson [Mon, 23 Mar 2009 19:10:38 +0000 (19:10 +0000)]
Fix order of debug printf items, addr and config# were swapped.

15 years agoRelease driver lock at the end of the watchdog routine instead of trying to
John Baldwin [Mon, 23 Mar 2009 17:51:07 +0000 (17:51 +0000)]
Release driver lock at the end of the watchdog routine instead of trying to
acquire it again.

Submitted by: bland
MFC after: 3 days

15 years agoSupport for a new environment variable, LD_ELF_HINTS_PATH for overriding
Xin LI [Mon, 23 Mar 2009 16:49:00 +0000 (16:49 +0000)]
Support for a new environment variable, LD_ELF_HINTS_PATH for overriding
the rtld hints file.  This environment variable would be unset if the
process is considered as tainted with setuid/setgid.  This feature gives
a convenient way of using a custom set of shared library that is not
located in the default location and switch back.

Feature requested by: iXsystems
Original patch by: John Hixson
MFC after: 2 weeks

15 years agoFixed comments so it stays in 80 chars by line
VANHULLEBUS Yvan [Mon, 23 Mar 2009 16:20:39 +0000 (16:20 +0000)]
Fixed comments so it stays in 80 chars by line
with hard tabs of 8 chars....

Approved by: gnn(mentor)

15 years agoDon't tempt svn to expand the example $FreeBSD strings
Doug Barton [Mon, 23 Mar 2009 14:42:41 +0000 (14:42 +0000)]
Don't tempt svn to expand the example $FreeBSD strings

15 years ago- Ensure that INTx isn't disabled, as these chips apparently have a
Marius Strobl [Mon, 23 Mar 2009 14:36:50 +0000 (14:36 +0000)]
- Ensure that INTx isn't disabled, as these chips apparently have a
  quirk requiring it to be enabled even when using MSI. This makes
  the latter work again after r189285.
- Remove a comment which no longer applies since r190194.

15 years agoSave (empty) worklist at successful completion
Poul-Henning Kamp [Mon, 23 Mar 2009 11:07:34 +0000 (11:07 +0000)]
Save (empty) worklist at successful completion

15 years agoTogging ALLMULTI does not require controller reset so have Rx
Pyun YongHyeon [Mon, 23 Mar 2009 05:46:28 +0000 (05:46 +0000)]
Togging ALLMULTI does not require controller reset so have Rx
filtering handle this. Introduce a new function msk_rxfilter that
handles Rx filter configuration and multicast setup as well as
promiscuous mode. This simplifies code a lot.
Promiscuous mode always have preference to any other Rx
filtering so don't disable the mode when ALLMULTI is set.

15 years agoDo not pass uninitialized data buffer to devctl_queue_data function.
Alexander Kabaev [Mon, 23 Mar 2009 01:24:17 +0000 (01:24 +0000)]
Do not pass uninitialized data buffer to devctl_queue_data function.
This was botched in revision 190191.

15 years agoAdd safety check that does not allow empty strings to be queued
Alexander Kabaev [Mon, 23 Mar 2009 01:13:34 +0000 (01:13 +0000)]
Add safety check that does not allow empty strings to be queued
to the devctl notification queue. Empty strings cause devctl read
call to return 0 and result in devd exiting prematurely.

The actual offender (ugen notes for root hubs) will be fixed
by separate commit.

15 years agoAdd the URL for RFC2142
Gregory Neil Shapiro [Mon, 23 Mar 2009 00:40:07 +0000 (00:40 +0000)]
Add the URL for RFC2142

PR: conf/127510

15 years agoFix inversed logic. pci_find_extcap() returns 0 when it finds
Pyun YongHyeon [Mon, 23 Mar 2009 00:27:46 +0000 (00:27 +0000)]
Fix inversed logic. pci_find_extcap() returns 0 when it finds
specified capability.

15 years agoCorrectly sanity-check timer IDs. [SA-09:06]
Colin Percival [Mon, 23 Mar 2009 00:00:50 +0000 (00:00 +0000)]
Correctly sanity-check timer IDs. [SA-09:06]

Limit the size of malloced buffer when dumping environment
variables. [EN-09:01]

Approved by: so (cperciva)
Approved by: re (kensmith)
Security: FreeBSD-SA-09:06.ktimer
Errata: FreeBSD-EN-09:01.kenv

15 years agoTest the r190298 change.
Stefan Farfeleder [Sun, 22 Mar 2009 23:00:52 +0000 (23:00 +0000)]
Test the r190298 change.

15 years agoFix the behaviour of the read built-in when IFS is unset.
Stefan Farfeleder [Sun, 22 Mar 2009 22:57:53 +0000 (22:57 +0000)]
Fix the behaviour of the read built-in when IFS is unset.

Obtained from: NetBSD

15 years agoAdd a few tests for the read built-in.
Stefan Farfeleder [Sun, 22 Mar 2009 22:14:46 +0000 (22:14 +0000)]
Add a few tests for the read built-in.

15 years agoImprove the IFS handling of the read built-in.
Stefan Farfeleder [Sun, 22 Mar 2009 22:09:12 +0000 (22:09 +0000)]
Improve the IFS handling of the read built-in.

Obtained from: NetBSD
Submitted by: Jilles Tjoelker

15 years agoAdd dtnfsclient, a first cut at an NFSv2/v3 client reuest DTrace
Robert Watson [Sun, 22 Mar 2009 22:07:52 +0000 (22:07 +0000)]
Add dtnfsclient, a first cut at an NFSv2/v3 client reuest DTrace
provider.  The NFS client exposes 'start' and 'done' probes for NFSv2
and NFSv3 RPCs when using the new RPC implementation, passing in the
vnode, mbuf chain, credential, and NFSv2 or NFSv3 procedure number.
For 'done' probes, the error number is also available.

Probes are named in the following way:

  ...
  nfsclient:nfs2:write:start
  nfsclient:nfs2:write:done
  ...
  nfsclient:nfs3:access:start
  nfsclient:nfs3:access:done
  ...

Access to the unmarshalled arguments is not easily available at this
point in the stack, but the passed probe arguments are sufficient to
to a lot of interesting things in practice.  Technically, these probes
may cover multiple RPC retransmits, and even transactions if the
transaction ID change as a result of authentication failure or a
jukebox error from the server, but usefully capture the intent of a
single NFS request, such as access, getattr, write, etc.

Typical use might involve profiling RPC latency by system call, number
of RPCs, how often a getattr leads to a call to access, when failed
access control checks occur, etc.  More detailed RPC information might
best be provided by adding a krpc provider.  It would also be useful
to add NFS client probes for events such as the access cache or
attribute cache satisfying requests without an RPC.

Sponsored by: Google, Inc.
MFC after: 1 month

15 years agoUpdate test for r190284.
Stefan Farfeleder [Sun, 22 Mar 2009 21:12:00 +0000 (21:12 +0000)]
Update test for r190284.

15 years ago- Apply the r190270 changes to printing of single aliases too.
Stefan Farfeleder [Sun, 22 Mar 2009 21:09:22 +0000 (21:09 +0000)]
- Apply the r190270 changes to printing of single aliases too.
- Sort the aliases before printing them.

15 years agoCorrect the documentation to reflect the fact that BUS_DMA_NOCACHE is a
Robert Noland [Sun, 22 Mar 2009 21:00:58 +0000 (21:00 +0000)]
Correct the documentation to reflect the fact that BUS_DMA_NOCACHE is a
flag to bus_dmamem_alloc() rather than bus_dmamap_load();

Discussed with: kib
MFC after: 3 days

15 years agoFix up the flags to bus_dmamem again. The man page incorrectly showed
Robert Noland [Sun, 22 Mar 2009 20:58:29 +0000 (20:58 +0000)]
Fix up the flags to bus_dmamem again.  The man page incorrectly showed
the BUS_DMA_NOCACHE flags as being a valid flag for load instead of alloc.

Discussed with: kib
MFC after: 3 days

15 years agoUpdate stale comments. The alternate address space mapping was eliminated
Alan Cox [Sun, 22 Mar 2009 18:56:26 +0000 (18:56 +0000)]
Update stale comments.  The alternate address space mapping was eliminated
when PAE support was added to i386.  The direct mapping exists on amd64.

15 years agoUpdate and extend the tests for alias.
Stefan Farfeleder [Sun, 22 Mar 2009 17:22:14 +0000 (17:22 +0000)]
Update and extend the tests for alias.