]> CyberLeo.Net >> Repos - FreeBSD/FreeBSD.git/log
FreeBSD/FreeBSD.git
16 years agoMFC r1.8
gabor [Fri, 4 Jan 2008 13:00:16 +0000 (13:00 +0000)]
MFC r1.8
- This driver will first appear in 6.3, not 7.0

16 years agoMFC r1.9
gabor [Fri, 4 Jan 2008 12:59:56 +0000 (12:59 +0000)]
MFC r1.9
- This driver will first appear in 6.3, not 7.0

16 years agoMFC r1.2
gabor [Fri, 4 Jan 2008 12:59:34 +0000 (12:59 +0000)]
MFC r1.2
- This driver will first appear in 6.3, not 7.0

16 years agoMFC: rev. 1.76: 'netstat -A -p tcp' prints the Tcpcb pointer in the 1st col
obrien [Fri, 4 Jan 2008 03:21:21 +0000 (03:21 +0000)]
MFC: rev. 1.76: 'netstat -A -p tcp' prints the Tcpcb pointer in the 1st col

16 years agoMFC: rev 1.149: Fix a broken DIAGNOSTICS/INVARIANTS check to handle a
obrien [Tue, 1 Jan 2008 03:01:03 +0000 (03:01 +0000)]
MFC: rev 1.149: Fix a broken DIAGNOSTICS/INVARIANTS check to handle a
     block range that spans two block maps.

16 years agoMFC: Happy new year 2008!
delphij [Mon, 31 Dec 2007 22:12:07 +0000 (22:12 +0000)]
MFC: Happy new year 2008!

16 years agoMFC rev. 1.87
mav [Sat, 29 Dec 2007 17:38:24 +0000 (17:38 +0000)]
MFC rev. 1.87
Fix incorrectly placed bracket in pppoe_find_svc().

16 years agoMFC rev. 1.22
mav [Sat, 29 Dec 2007 17:35:32 +0000 (17:35 +0000)]
MFC rev. 1.22
Increase control channel xmit queue to 128 packets.
Previous value 16 was too small for real LAC as temporal activity
spike cound easily overflow queue demanding tunnel disconnection due
to possible state inconsistency.

16 years agoMFC r1.8
thompsa [Sat, 29 Dec 2007 17:29:11 +0000 (17:29 +0000)]
MFC r1.8
  Fix a panic where if the mbuf was consumed by the filter for requeueing
  (dummynet), ipsec_filter() would return the empty error code and the ipsec
  code would continue to forward/deference the null mbuf.

Approved by: re (gnn)

16 years agoMFC v1.117:
kensmith [Sat, 29 Dec 2007 06:20:30 +0000 (06:20 +0000)]
MFC v1.117:
> Plug a memory leak.  Once any given package (e.g. perl) was installed
> we would leak a saved screen for every other package we tried to install
> that listed perl as one of its dependencies.  When installing things
> like gnome and kde that wound up being a LOT of leaked memory.

While here clean up some whitespace nits that are differences between
this branch and the others.

16 years agoMFC v1.78 and v1.79:
kensmith [Sat, 29 Dec 2007 06:19:18 +0000 (06:19 +0000)]
MFC v1.78 and v1.79:
Remove process limits for datasize and stacksize which are unlimited
during a "normal login" (thanks to /etc/login.conf) but not when
sysinstall is running as part of an installation.  Fixups for error
messages and a comment that became untrue.

16 years agoMFC: Update to 2008.
imp [Wed, 26 Dec 2007 16:49:12 +0000 (16:49 +0000)]
MFC: Update to 2008.

Approved by: re@ (kensmith)

16 years agoMFC: properly handle BZ_STREAM_END/Z_STREAM_END.
sobomax [Wed, 26 Dec 2007 10:34:27 +0000 (10:34 +0000)]
MFC: properly handle BZ_STREAM_END/Z_STREAM_END.

16 years agoMFC v1.116:
kensmith [Mon, 24 Dec 2007 06:42:35 +0000 (06:42 +0000)]
MFC v1.116:
> For the INDEX file to be used on disc1 of the 7.0-RELEASE the longest
> "build dependencies" field is 5,108 characters which overflows the
> length of the junk buffer by a teeny bit.  This whole section needs
> much more error checking but for now just completely ignore stuff
> we have no interest in instead of copying it to someplace we don't
> use in the process.

Approved by:    re (bmah)

16 years agoMFC version 1.55, safety belt for the (unlikely) disappearance of
dougb [Mon, 24 Dec 2007 05:50:46 +0000 (05:50 +0000)]
MFC version 1.55, safety belt for the (unlikely) disappearance of
the temproot before a -r run.

16 years agoMFC: 1.8
marius [Sun, 23 Dec 2007 11:28:59 +0000 (11:28 +0000)]
MFC: 1.8

In openprom_ioctl() ensure appropriate permissions and that data isn't
NULL and doesn't point to a NULL pointer before dereferencing it.

16 years agoMerge in_pcb.c:1.198 from HEAD to RELENG_6:
rwatson [Sat, 22 Dec 2007 20:06:54 +0000 (20:06 +0000)]
Merge in_pcb.c:1.198 from HEAD to RELENG_6:

  When IPSEC fails to allocate policy state for an inpcb, and MAC is in use,
  free the MAC label on the inpcb before freeing the inpcb.

  Submitted by:   tanyong <tanyong at ercist dot iscas dot ac dot cn>,
                  zhouzhouyi

Approved by: re (kensmith)

16 years agoMerge mac_posix_sem.c:1.12 from HEAD to RELENG_6:
rwatson [Sat, 22 Dec 2007 15:08:09 +0000 (15:08 +0000)]
Merge mac_posix_sem.c:1.12 from HEAD to RELENG_6:

  Fix a MAC label leak for POSIX semaphores, in which per-policy labels
  would be properly disposed of, but the global label structure for the
  semaphore wouldn't be freed.

  Reported by:    tanyong <tanyong at ercist dot iscas dot ac dot cn>,
                  zhouzhouyi

Approved by: re (kensmith)

16 years agoMerge in6_pcb.c:1.86 from HEAD to RELENG_6:
rwatson [Sat, 22 Dec 2007 15:06:51 +0000 (15:06 +0000)]
Merge in6_pcb.c:1.86 from HEAD to RELENG_6:

  Fix leaking MAC labels for IPv6 inpcbs by adding missing MAC label
  destroy call; this transpired because the inpcb alloc path for IPv4/IPv6
  is the same code, but IPv6 has a separate free path.  The results was
  that as new IPv6 TCP connections were created, kernel memory would
  gradually leak.

  Reported by:    tanyong <tanyong at ercist dot iscas dot ac dot cn>,
                  zhouzhouyi

Approved by: re (kensmith)

16 years agoMFC r1.25
thompsa [Fri, 21 Dec 2007 05:34:35 +0000 (05:34 +0000)]
MFC r1.25
  - Use the macro to check the port status has it will also test if its
    administratively down (!IFF_UP)
  - Use the same parameters to lagg_link_active() to get the backup port as in
    the output path, this didnt actually matter in practice as sc_primary is
    always the first on the port list.

Approved by:    re (kensmith)

16 years agoMFC r1.108
thompsa [Fri, 21 Dec 2007 05:30:47 +0000 (05:30 +0000)]
MFC r1.108
 When the bridge has an address and a packet comes in for it then drop it if
 the link has been marked discarding by Spanning Tree.

Approved by:    re (kensmith)

16 years agoRegen.
jhb [Wed, 19 Dec 2007 23:04:12 +0000 (23:04 +0000)]
Regen.

16 years agoMFC: Add freebsd32 compat wrappers for msgctl() and __semctl().
jhb [Wed, 19 Dec 2007 23:03:50 +0000 (23:03 +0000)]
MFC: Add freebsd32 compat wrappers for msgctl() and __semctl().

16 years agoMFC: Move several data structure definitions out of freebsd32_misc.c and
jhb [Wed, 19 Dec 2007 23:03:02 +0000 (23:03 +0000)]
MFC: Move several data structure definitions out of freebsd32_misc.c and
into freebsd32.h and freebsd32_ipc.h.

16 years agoMFC revision 1.679:
rpaulo [Wed, 19 Dec 2007 21:01:35 +0000 (21:01 +0000)]
MFC revision 1.679:
  Disallow the legacy USB circuit to generate an SMI# via an ICH
  register (MacBooks only).
  This allows MacBooks to boot in SMP mode without any trick and
  solves
  the timer problems with HZ=1000.

Approved by:      re (kensmith), njl (mentor)

16 years agoMFC revision 1.662 and 1.663:
rpaulo [Wed, 19 Dec 2007 20:59:46 +0000 (20:59 +0000)]
MFC revision 1.662 and 1.663:
  Disallow the legacy USB circuit to generate an SMI# via an ICH
  register (MacBooks only).
  This allows MacBooks to boot in SMP mode without any trick and
  solves the timer problems with HZ=1000.

Approved by: re (kensmith), njl (mentor)

16 years agogdbserver came back.
obrien [Wed, 19 Dec 2007 17:08:34 +0000 (17:08 +0000)]
gdbserver came back.

Reminded by: John Hein <jhein@timing.com>

16 years agoMFC: scale TCPTV_MIN with hz, not a fixed number of ticks. Otherwise
peter [Tue, 18 Dec 2007 03:39:17 +0000 (03:39 +0000)]
MFC: scale TCPTV_MIN with hz, not a fixed number of ticks.  Otherwise
machines with hz=1000 drop connections in a matter of a few seconds on a
fast link (eg: ethernet).  With this change, it drops them in the same
time as a machine with hz=100 would.

16 years agoMFC: Only pass paths to directories or config files that exist for ldconfig
jhb [Tue, 18 Dec 2007 01:34:46 +0000 (01:34 +0000)]
MFC: Only pass paths to directories or config files that exist for ldconfig
for 32-bit binaries.

16 years agoMFC: Only attach to a GPT partition if it has the GPT_ENT_TYPE_FREEBSD
jhb [Tue, 18 Dec 2007 01:32:55 +0000 (01:32 +0000)]
MFC: Only attach to a GPT partition if it has the GPT_ENT_TYPE_FREEBSD
type.

16 years agoMFC: Update posix_openpt(3) to handle 512 ptys.
jhb [Tue, 18 Dec 2007 00:36:21 +0000 (00:36 +0000)]
MFC: Update posix_openpt(3) to handle 512 ptys.

Approved by: re (gnn)

16 years agoInitial pass at shifting packages around for 6.3-REL layout. There
kensmith [Mon, 17 Dec 2007 23:54:20 +0000 (23:54 +0000)]
Initial pass at shifting packages around for 6.3-REL layout.  There
is a lot of space left on disc3 that we can put to use by adding back
in lots of the packages we've been slowly ejecting through the years.
We'll decide exactly which ones a little bit later.

Note: not an MFC.  The handling of what packages go where is often
done on a release-by-release basis and 6.3's release cycle is leading
7.0's at the moment.

16 years agoAdd support for disc3. Depending on architecture the combination of
kensmith [Mon, 17 Dec 2007 23:51:29 +0000 (23:51 +0000)]
Add support for disc3.  Depending on architecture the combination of
kde3 and gnome2 is at least 670Mb which would mean if we stick with
just disc1/disc2 the only packages we could provide on the iso's
would be X11, kde3, and gnome2...

This is not an MFC, it was not committed to HEAD or RELENG_7.  It's
looking like we should plan on a bit more radical a change for how
packages get done for 8.X (current thought is cdrom sized things with
no packages at all, and one DVD sized thing with packages).  It's
almost certain this change will be done in RELENG_7 but the 7.0
release cycle hasn't reached the point of dealing with packages quite
yet so I'm holding off on that a bit longer.

16 years agoMFC v1.253 of dist.c and v1.71 of dist.h:
kensmith [Mon, 17 Dec 2007 23:43:42 +0000 (23:43 +0000)]
MFC v1.253 of dist.c and v1.71 of dist.h:

> A few more adjustments needed because of changes in xorg.  Just installing
> xorg-server doesn't include any video drivers so install xorg-drivers as
> well.  And if font-alias isn't installed the X server won't start,
> complaining it can't find the font "fixed".

16 years agoMFC: Autogenerate hardware notes for hptrr(4).
bmah [Mon, 17 Dec 2007 12:22:05 +0000 (12:22 +0000)]
MFC:  Autogenerate hardware notes for hptrr(4).

16 years agoMFC: 1.334
guido [Mon, 17 Dec 2007 10:01:14 +0000 (10:01 +0000)]
MFC: 1.334

16 years agoAdd the hptrr driver.
scottl [Sat, 15 Dec 2007 06:21:58 +0000 (06:21 +0000)]
Add the hptrr driver.

Submitted by: Highpoint
Approved by: re

16 years agoAdd the hptrr driver.
scottl [Sat, 15 Dec 2007 06:03:44 +0000 (06:03 +0000)]
Add the hptrr driver.

Submitted by: Highpoint
Approved by: re

16 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Sat, 15 Dec 2007 00:56:18 +0000 (00:56 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

16 years agoMFC: -r1.296
qingli [Fri, 14 Dec 2007 19:23:31 +0000 (19:23 +0000)]
MFC: -r1.296
use the sequence number comparison macro to compare
projected_offset against isn_offset to account for
wrap around

16 years agoMFC rev 1.84-1.85
mav [Wed, 12 Dec 2007 19:14:05 +0000 (19:14 +0000)]
MFC rev 1.84-1.85
Remove ng_pppoe_sendpacket() function to simplify code as it is called
as much times as it has cases inside of it.
Split ng_pppoe_rcvdata() function into three hook-specific ones
to simplify code and reduce stack usage.

16 years ago- Add a workaround for the DMA bugs on some alpha chipsets that ATA DMA
jhb [Mon, 10 Dec 2007 20:14:16 +0000 (20:14 +0000)]
- Add a workaround for the DMA bugs on some alpha chipsets that ATA DMA
  trips over often.  Specifically, in these chipsets DMA transfers that
  cross a page boundary result in data corruption.  The workaround is to
  not allow any DMA transfers for non-static DMA maps (i.e. "real"
  transfers as opposed to work areas allocated with bus_dmamem_alloc()) to
  cross a page in a single S/G element.  This behavior is enabled by
  setting 'busdma_pyxis_bug' to 1.
- Add a new tunable 'machdep.busdma_pyxis_bug' that can be used to enable
  the workaround from the loader.  This can be used to enable it on
  chipsets where we don't automatically enable it.
- Auto-enable the workaround for buggy PYXIS 1 chipsets supported via
  cia(4).

PR: alpha/75317

16 years agoMFC all updates other than the combination of fortunes-o.real
dougb [Mon, 10 Dec 2007 19:34:26 +0000 (19:34 +0000)]
MFC all updates other than the combination of fortunes-o.real
and fortunes2.o.

16 years agoMFC to RELENG_6
emaste [Mon, 10 Dec 2007 19:33:36 +0000 (19:33 +0000)]
MFC to RELENG_6
aac_pci.c:1.63-1.65
aac.4 1.37-1.39

  Add PCI IDs for controllers obtained from Adaptec's latest drivers.

Approved by: re (kensmith)

16 years agoMFC to RELENG_6
emaste [Mon, 10 Dec 2007 19:17:23 +0000 (19:17 +0000)]
MFC to RELENG_6
aac_debug.c:1.23

  Respell 'Conatainer' as 'Container'.

Approved by: re (kensmith)

16 years agoMFC rev 1.13:
gabor [Mon, 10 Dec 2007 18:51:40 +0000 (18:51 +0000)]
MFC rev 1.13:

- Add doc/hu_* and doc/mn_* to the refuse examples so that include all
  of the currently used languages

16 years agoMFC rev 1.14:
gabor [Mon, 10 Dec 2007 18:51:15 +0000 (18:51 +0000)]
MFC rev 1.14:

- Add doc/hu_* and doc/mn_* to the sample refuse file so that it includes all
  of the languages

16 years agoMFC: 1.3
jkim [Mon, 10 Dec 2007 17:58:47 +0000 (17:58 +0000)]
MFC: 1.3

Teach dhclient(8) about net80211 link (association) status.

Reviewed by: brooks
Approved by: re (kensmith)

16 years agoMFC: Support new ARC120x generation cards.
scottl [Sun, 9 Dec 2007 19:28:21 +0000 (19:28 +0000)]
MFC: Support new ARC120x generation cards.

Approved by: re
Submitted by: Erich Chen @ Areca

16 years agoSync up ata(4) work by sos@ on -HEAD:
delphij [Sun, 9 Dec 2007 19:27:19 +0000 (19:27 +0000)]
Sync up ata(4) work by sos@ on -HEAD:

 * Add preliminary SATA ATAPI support for sii 3132/3124 chips.
 * Fix speed report on Intel SATA chips in compat mode.
 * Fix broken detect of JMicron 368.

Related revisions:
 - ata-chipset.c: 1.208, 1.210
 - ata-disk.c.c:  1.205

Approved by: re (kensmith), sos

16 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Sun, 9 Dec 2007 11:01:59 +0000 (11:01 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.

16 years agoMFC: 1.149-1.150. Hold Giant over device_suspend/resume() methods in apm.
njl [Sun, 9 Dec 2007 00:24:16 +0000 (00:24 +0000)]
MFC: 1.149-1.150.  Hold Giant over device_suspend/resume() methods in apm.

Approved by: re

16 years agoMFC rev 1.6 msk.4
remko [Sat, 8 Dec 2007 12:19:13 +0000 (12:19 +0000)]
MFC rev 1.6 msk.4
MFC rev 1.26 if_msk.c
MFC rev 1.11 if_mskreg.h

  Bring in support for: Marvell Yukon 8039

  PR:             118401
  Submitted by:   Skip Ford <skip at menantico dot com>
  Approved by:    imp (mentor), yongari
  MFC After:      3 days

Approved by: re (kensmith), imp (mentor, implicit)

16 years ago- Add support for all geom class command options in the Fixit environment by
lulf [Sat, 8 Dec 2007 09:41:50 +0000 (09:41 +0000)]
- Add support for all geom class command options in the Fixit environment by
  pointing GEOM_LIBRARY_PATH to the geom libraries.

PR: misc/113543
Submitted by: Ed Schouten <ed -at- fxq.nl>
Approved by: re (kensmith), pjd (mentor)

16 years agoregen.
dwmalone [Sat, 8 Dec 2007 08:50:37 +0000 (08:50 +0000)]
regen.

Approved by: re (kensmith)

16 years agoThe kernel version of Linux statfs64 is actually supposed to take
dwmalone [Sat, 8 Dec 2007 08:48:08 +0000 (08:48 +0000)]
The kernel version of Linux statfs64 is actually supposed to take
3 arguments, but we had forgotten the second argument. Also make the
Linux statfs64 struct depend on the architecture because it has an
extra 4 bytes padding on amd64 compared to i386.

The three argument fix is from David Taylor, the struct statfs64
stuff is my fault. With this patch I can install i386 Linux matlab
on an amd64 machine.

Approved by: re (kensmith)

16 years agoMFC: If "length at end" flag is specified, then a zero
kientzle [Sat, 8 Dec 2007 07:55:40 +0000 (07:55 +0000)]
MFC: If "length at end" flag is specified, then a zero
length at the beginning doesn't mean an empty body.

Approved by: re (Ken Smith)

16 years agoAdd missed closing brace.
kib [Sat, 8 Dec 2007 06:43:25 +0000 (06:43 +0000)]
Add missed closing brace.

16 years agoMFC: 1.317, 1.318, 1.319
kmacy [Fri, 7 Dec 2007 23:22:57 +0000 (23:22 +0000)]
MFC: 1.317, 1.318, 1.319
 Fix bit macros to work on multi-byte types

Approved by: re(kensmith)

16 years agoMFC:
cognet [Fri, 7 Dec 2007 22:08:02 +0000 (22:08 +0000)]
MFC:
sys/arm/include/asmacros.h rev 1.7
sys/arm/include/atomic.h rev 1.23

Close a race.

The RAS implementation would set the end address, then the start
address.  These were used by the kernel to restart a RAS sequence if
it was interrupted.  When the thread switching code ran, it would
check these values and adjust the PC and clear them if it did.

However, there's a small flaw in this scheme.  Thread T1, sets the end
address and gets preempted.  Thread T2 runs and also does a RAS
operation.  This resets end to zero.  Thread T1 now runs again and
sets start and then begins the RAS sequence, but is preempted before
the RAS sequence executes its last instruction.  The kernel code that
would ordinarily restart the RAS sequence doesn't because the PC isn't
between start and 0, so the PC isn't set to the start of the sequence.
So when T1 is resumed again, it is at the wrong location for RAS to
produce the correct results.  This causes the wrong results for the
atomic sequence.

The window for the first race is 3 instructions.  The window for the
second race is 5-10 instructions depending on the atomic operation.
This makes this failure fairly rare and hard to reproduce.

Mutexs are implemented in libthr using atomic operations.  When the
above race would occur, a lock could get stuck locked, causing many
downstream problems, as you might expect.

Also, make sure to reset the start and end address when doing a syscall, or
a malicious process could set them before doing a syscall.

Reviewed by: imp, ups (thanks guys)
Approved by: re (kensmith)
Pointy hat to:  cognet

16 years agoMFC revision 1.209
delphij [Fri, 7 Dec 2007 17:59:39 +0000 (17:59 +0000)]
MFC revision 1.209
date: 2007/11/26 19:08:08;  author: sos;  state: Exp;  lines: +1 -1
Use device_set_desc_copy in the generic ident as its used on a temp buffer.

Approved by: re (kensmith)

16 years agoMFC rev. 1.181:
kib [Fri, 7 Dec 2007 14:30:26 +0000 (14:30 +0000)]
MFC rev. 1.181:
Check for the program headers alignment of the ELF images before
dereferencing. Unaligned access could cause panic on strict alignment
architectures.

Approved by: re (kensmith)

16 years agoMFC rev. 1.5, 1.6:
kib [Fri, 7 Dec 2007 14:16:53 +0000 (14:16 +0000)]
MFC rev. 1.5, 1.6:
Restore the branding of the ELF executables. The structure was optimized out.

Approved by: re (kensmith)

16 years agoMFC support for the Camellia block cipher
gnn [Fri, 7 Dec 2007 08:45:29 +0000 (08:45 +0000)]
MFC support for the Camellia block cipher

HEAD
Revision  Path
  1.16    src/lib/libipsec/pfkey_dump.c
  1.38    src/sbin/setkey/setkey.8
  1.12    src/sbin/setkey/token.l
  1.81    src/secure/lib/libcrypto/Makefile
  1.47    src/secure/lib/libcrypto/Makefile.inc
  1.8     src/secure/lib/libcrypto/opensslconf-amd64.h
  1.8     src/secure/lib/libcrypto/opensslconf-i386.h
  1.1205  src/sys/conf/files
  1.1     src/sys/crypto/camellia/camellia-api.c (new)
  1.1     src/sys/crypto/camellia/camellia.c (new)
  1.1     src/sys/crypto/camellia/camellia.h (new)
  1.6     src/sys/modules/crypto/Makefile
  1.15    src/sys/net/pfkeyv2.h
  1.1     src/sys/netinet6/esp_camellia.c (new)
  1.1     src/sys/netinet6/esp_camellia.h (new)
  1.25    src/sys/netinet6/esp_core.c
  1.19    src/sys/netipsec/xform_esp.c
  1.34    src/sys/opencrypto/cryptodev.c
  1.25    src/sys/opencrypto/cryptodev.h
  1.19    src/sys/opencrypto/cryptosoft.c
  1.9     src/sys/opencrypto/xform.c
  1.4     src/sys/opencrypto/xform.h

Approved by: re (kensmith)
Submitted by: Tomoyuki Okazaki <okazaki at kick dot gr dot jp>

16 years agoMFC r1.18:
kevlo [Fri, 7 Dec 2007 05:17:34 +0000 (05:17 +0000)]
MFC r1.18:

zic(8) should exit the loop when the end of the string is reached
due to the call to error(). The problem found by doing fuzz testing.

Approved by: re (kensmith)

16 years agoDUH, fix breakage of last checkin. Its kinda humorous,
jfv [Fri, 7 Dec 2007 04:01:16 +0000 (04:01 +0000)]
DUH, fix breakage of last checkin. Its kinda humorous,
the code is broken in 7 but that will never be a problem
because of defines, but it broke here, oh well :)

Approved by: r

16 years agoMFC r1.209
thompsa [Fri, 7 Dec 2007 03:46:23 +0000 (03:46 +0000)]
MFC r1.209

  Apply a workaround for the unkillable jail problem where some devices created
  within the jail are never freed.

Approved by: re (kensmith)

16 years agoMFC of a few important fixes done in RELENG_7 to 6
jfv [Fri, 7 Dec 2007 01:06:41 +0000 (01:06 +0000)]
MFC of a few important fixes done in RELENG_7 to 6
  - WITNESS lock problem due to naming of multi locks
  - Race in em_stop if you don't hold the TX lock
  - Bug in TSO code when you have fowarding and different
    MSS sizes (this more for completeness since TSO is not
    enabled in 6.3)

Essentially syncs the two drivers up to version 6.7.3

Approved by:re

16 years agoRegen.
jhb [Thu, 6 Dec 2007 22:00:32 +0000 (22:00 +0000)]
Regen.

16 years agoMFC: Add kern_msgsnd() and kern_msgrcv() and use them to fix msgsnd()
jhb [Thu, 6 Dec 2007 22:00:05 +0000 (22:00 +0000)]
MFC: Add kern_msgsnd() and kern_msgrcv() and use them to fix msgsnd()
and msgrcv() in 32-bit ABIs on 64-bit kernels.

16 years agoMFC: (1.21-1.23) cleanup, additional error checking
des [Thu, 6 Dec 2007 13:35:18 +0000 (13:35 +0000)]
MFC: (1.21-1.23) cleanup, additional error checking

Approved by: re (kensmith)

16 years agoMFC: Add a cross-reference to newgrp(1).
philip [Thu, 6 Dec 2007 10:51:40 +0000 (10:51 +0000)]
MFC: Add a cross-reference to newgrp(1).

Approved by: re (kensmith)

16 years agoFix a use-after-free.
philip [Thu, 6 Dec 2007 10:18:19 +0000 (10:18 +0000)]
Fix a use-after-free.

Submitted by: Ilja van Sprundel <ilja -at- netric.org>
Reviewed by: secteam
Approved by: re (kensmith)

16 years agoMFC if_iwi.c:1.59
philip [Thu, 6 Dec 2007 10:16:34 +0000 (10:16 +0000)]
MFC if_iwi.c:1.59
    if_ipw.c:1.34

  Plug two potential (root-only, local) information leaks.  buf is not
  initialized before use and returned integrally instead of up to size.

Submitted by: Ilja van Sprundel <ilja -at- netric.org>
Reviewed by: secteam
Approved by: re (kensmith)

16 years agoMFC: Always assume type #1 PCI config transactions for amd64 and relax the
jhb [Thu, 6 Dec 2007 08:26:18 +0000 (08:26 +0000)]
MFC: Always assume type #1 PCI config transactions for amd64 and relax the
probe for type #1 on i386.

16 years agoMFC: Change pci_cfgdisable() into a nop for type #1 similar to what other
jhb [Thu, 6 Dec 2007 08:21:23 +0000 (08:21 +0000)]
MFC: Change pci_cfgdisable() into a nop for type #1 similar to what other
OSs do.

16 years agoMFC: Don't delete files in the X11 socket directories under /tmp.
jhb [Thu, 6 Dec 2007 08:10:40 +0000 (08:10 +0000)]
MFC: Don't delete files in the X11 socket directories under /tmp.

16 years agoMFC: Properly sort 'sync' in the list of options passed to -o.
jhb [Thu, 6 Dec 2007 08:08:47 +0000 (08:08 +0000)]
MFC: Properly sort 'sync' in the list of options passed to -o.

16 years agoMFC if_re.c, rev 1.98 to RELENG_6
yongari [Thu, 6 Dec 2007 06:01:47 +0000 (06:01 +0000)]
MFC if_re.c, rev 1.98 to RELENG_6
Always honor promiscuous flag prior to programming Rx multicast
filter. This fixes a regression introduced in rev 1.89.

PR: 114632

16 years agoHeadphone / speakers automute fixup for Acer Aspire 4710 @ ALC268 laptop.
ariff [Thu, 6 Dec 2007 05:20:46 +0000 (05:20 +0000)]
Headphone / speakers automute fixup for Acer Aspire 4710 @ ALC268 laptop.

Tested by: Liyu, She

16 years agoVery belated MFC.
kmacy [Wed, 5 Dec 2007 22:05:49 +0000 (22:05 +0000)]
Very belated MFC.

  - fix qset to port binding as a proper fix for the problems encountered on the 4-port
  - fix the use after free seen when sending packets small enough to fit as an immediate
    and bpf peers are present
  - update to firmware rev 4.7 along with various small vendor fixes

MFC after: 3 days

16 years agoCall NET_LOCK_GIANT/NET_UNLOCK_GIANT around calls to protocol drain
rwatson [Wed, 5 Dec 2007 00:00:10 +0000 (00:00 +0000)]
Call NET_LOCK_GIANT/NET_UNLOCK_GIANT around calls to protocol drain
methods in mb_reclaim().  This is not an MFC, as debug.mpsafenet is
not present in 7.x or 8.x.  In practice, this likely resulted in
instability only on kernels with INVARIANTS, as the protocol drain
paths are generally MPSAFE.

PR: 116034
Reported by: Dan Lukes <dan@obluda.cz>
Discussed with: kmacy, alc
Approved by: re (kensmith)
Tested by: Eugene Grosbein <eugen at grosbein dot pp dot ru>

16 years agoMFC revision 1.3: Add /root/, /.cshrc, and /.profile to the default
cperciva [Tue, 4 Dec 2007 22:51:36 +0000 (22:51 +0000)]
MFC revision 1.3: Add /root/, /.cshrc, and /.profile to the default
UpdateIfUnmodified directive in order to avoid overwriting these when
users upgrade to future versions of FreeBSD.

Approved by:    re (kensmith)

16 years agoMFC: less v416.
delphij [Tue, 4 Dec 2007 22:41:44 +0000 (22:41 +0000)]
MFC: less v416.

Approved by: re (kensmith)

16 years agoMFC: libexec/rtld-elf/sparc64/reloc.c 1.16,
marius [Tue, 4 Dec 2007 20:00:48 +0000 (20:00 +0000)]
MFC: libexec/rtld-elf/sparc64/reloc.c 1.16,
     sys/sparc64/sparc64/elf_machdep.c 1.24

- Fix the handling of R_SPARC_OLO10, which is a bit of a special case
  in the way we implement handling of relocations.
  While at it replace some more ELF_R_TYPE which should have been
  ELF64_R_TYPE_ID but didn't cause problems so far.
- Sync a sanity check between kernel and rtld(1) and change it to be
  maintenance free regarding the type used for the lookup table.
- Sprinkle const on lookup tables.
- Use __FBSDID.

16 years agoMFC: turn on gdbserver for the arch that supports it.
obrien [Tue, 4 Dec 2007 18:00:17 +0000 (18:00 +0000)]
MFC: turn on gdbserver for the arch that supports it.

16 years agoMFC: product of patch(1)
obrien [Tue, 4 Dec 2007 16:28:10 +0000 (16:28 +0000)]
MFC: product of patch(1)

16 years agoGC this patch.
obrien [Mon, 3 Dec 2007 22:08:49 +0000 (22:08 +0000)]
GC this patch.

16 years agoRegen.
jhb [Mon, 3 Dec 2007 21:54:17 +0000 (21:54 +0000)]
Regen.

16 years agoMFC: Sync up SYSV IPC with HEAD some:
jhb [Mon, 3 Dec 2007 21:53:23 +0000 (21:53 +0000)]
MFC: Sync up SYSV IPC with HEAD some:
- Add kern_semctl() and use it for Linux, IBCS2, and SVR4 ABIs.
- Don't pass a bufsz pointer to kern_shmctl() for operations that don't
  need it.
- Add module dependencies for SYSV IPC modules to IBCS2 and SVR4.
- Use kern_shmctl() in SVR4.
- Use kern_msgctl(), kern_semctl(), and kern_shmctl() in IBCS2.
- Mark linux_ipc(), svr4_sys_shmsys(), and svr4_sys_semsys() MPSAFE.

16 years agoMFC 1.109: Remove call to ipcperm() in shmget_existing().
jhb [Mon, 3 Dec 2007 21:47:33 +0000 (21:47 +0000)]
MFC 1.109: Remove call to ipcperm() in shmget_existing().

16 years agoMFC: Unexpand PTRIN() in a few places.
jhb [Mon, 3 Dec 2007 21:42:29 +0000 (21:42 +0000)]
MFC: Unexpand PTRIN() in a few places.

16 years agoMFC: sys/amd64/include/elf.h 1.19; sys/arm/include/elf.h 1.8;
marius [Mon, 3 Dec 2007 21:30:36 +0000 (21:30 +0000)]
MFC: sys/amd64/include/elf.h 1.19; sys/arm/include/elf.h 1.8;
     sys/i386/include/elf.h 1.17; sys/ia64/include/elf.h 1.15;
     sys/powerpc/include/elf.h 1.8; sys/sparc64/include/elf.h 1.13, 1.14, 1.15;
     sys/sys/elf32.h 1.11, 1.12, 1.13; sys/sys/elf64.h 1.14, 1.16, 1.17;
     sys/sys/elf_common.h 1.17, 1.18 (partial), 1.19, 1.20, 1.21, 1.22;
     sys/sys/elf_generic.h 1.8

- Add GNU symbol versioning constants.
- Correct URL to ELF header documantation.
- Add Sparc TLS relocation definitions.
- Move the relocation definitions to the common elf header so that DTrace
  can use them on one architecture targeted to a different one.
  Add the additional ELF types defines in Sun's "Linker and Libraries"
  manual.
- Add `PN_XNUM', for supporting ELF objects with a large number of program
  header table entries.

Discussed with: jb (who in turn discussed the MFC with jkoshy)

16 years agoMFC:
obrien [Mon, 3 Dec 2007 17:32:02 +0000 (17:32 +0000)]
MFC:
+ Turn on gdbserver for the arch that supports it.
+ Fix static/extern mismatch.
+ Remove extern int verbose declaration.

16 years agoMFC r1.6:
jkoshy [Mon, 3 Dec 2007 11:00:39 +0000 (11:00 +0000)]
MFC r1.6:
"Revert revision 1.4.
 Intel CPUs with family 0x6, model 0xE and later (i.e., Intel Core(TM))
 have a PMC architecture that differs somewhat from previous CPUs in
 family 0x6.  Even though the basic programming model is similar, the
 documented set of legal values that may be loaded into their PMC MSRs
 differs from that of the previous PMCs in family 0x6 and reusing bit
 values legal for the older PMCs could result in undefined behaviour in
 the general case."

16 years agoForced commit to correct file handle reversal in previous commit. The
marcus [Sun, 2 Dec 2007 23:46:27 +0000 (23:46 +0000)]
Forced commit to correct file handle reversal in previous commit.  The
message should have read:

This is a partial MFC of rev 1.126 from HEAD's libkse.  This only merges the
fix which causes stderr_debug to write output to STDOUT_FILENO instead
of STDERR_FILENO.  This is known to fix (or workaround) some mono-related
build problems.

Approved by: re (kensmith)

16 years agoMFC: rev 1.126
marcus [Sun, 2 Dec 2007 23:42:38 +0000 (23:42 +0000)]
MFC: rev 1.126

This is a partial MFC of rev 1.126 from HEAD's libkse.  This only merges the
fix which causes stderr_debug to write output to STDERR_FILENO instead
of STDOUT_FILENO.  This is known to fix (or workaround) some mono-related
build problems.

Approved by: re (kensmith)

16 years agoMFC:
cognet [Sun, 2 Dec 2007 14:38:21 +0000 (14:38 +0000)]
MFC:
gnu/usr.bin/gdb/arch/arm/nm-fbsd.h rev 1.2
gnu/usr.bin/gdb/arch/arm/Makefile rev 1.2
gnu/usr.bin/gdb/arch/arm/armfbsd-nat.c 1.2
gnu/usr.bin/gdb/arch/arm/armfbsd-tdep.c rev 1.2
gnu/usr.bin/gdb/arch/arm/init.c rev 1.2
Allow gdb to attach live process.
Thread support for arm.

Approved by: re (kensmith)

16 years agoMFC:
cognet [Sun, 2 Dec 2007 14:21:27 +0000 (14:21 +0000)]
MFC:
arm/at91/kb920x_machdep.c rev 1.24, 1.25 and 1.26
arm/at91/std.at91 rev 1.4
arm/at91/std.kb920x rev 1.5
arm/at91/at91rm92reg.h rev 1.6
conf/options.arm 1.18
Small tweaks to be able to load the kernel at a different address than
0x20000000 + add an option to override the default value of the master clock
frequency.
arm/at91/if_ate.c rev 1.21:
In ate_get_mac(), try to get the mac address in the right order, at least
in the same order as it's set in ate_set_mac.

arm/conf/HL200: config file for the Hot-e HL200

Approved by: re (kensmith)

16 years agoThis commit was manufactured by cvs2svn to create branch 'RELENG_6'.
cvs2svn [Sun, 2 Dec 2007 13:10:43 +0000 (13:10 +0000)]
This commit was manufactured by cvs2svn to create branch 'RELENG_6'.