openwrt/staging/blogic.git
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Wed, 19 Apr 2006 01:02:55 +0000 (18:02 -0700)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  [SPARC64]: Fix dependencies of HUGETLB_PAGE_SIZE_64K

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Wed, 19 Apr 2006 01:01:47 +0000 (18:01 -0700)]
Merge /pub/scm/linux/kernel/git/davej/cpufreq

* master.kernel.org:/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] powernow-k8.c: fix a check-after-use
  [CPUFREQ] Remove duplicate check in powernow-k8
  [CPUFREQ] drivers/cpufreq/cpufreq.c: static functions mustn't be exported
  [CPUFREQ] If max_freq got reduced (e.g. by _PPC) a write to sysfs scaling_governor let cpufreq core stuck at low max_freq for ever
  [CPUFREQ] x86_64: Proper null pointer check in powernow_k8_get
  [CPUFREQ] x86_64: Revert earlier powernow-k8 change
  [CPUFREQ] Update LART site URL
  [CPUFREQ] Remove pointless check in conservative governor.
  [CPUFREQ] trailing whitespace removal de-jour.
  [CPUFREQ] extra debugging in cpufreq_add_dev()

18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Wed, 19 Apr 2006 01:01:19 +0000 (18:01 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 3479/1: Corgi SSP: Fix potential concurrent access problem
  [ARM] 3478/1: SharpSL SCOOP: Fix potenial build failure

18 years ago[SPARC64]: Fix dependencies of HUGETLB_PAGE_SIZE_64K
Jean-Luc Léger [Tue, 18 Apr 2006 23:19:53 +0000 (16:19 -0700)]
[SPARC64]: Fix dependencies of HUGETLB_PAGE_SIZE_64K

This patch fixes dependencies of HUGETLB_PAGE_SIZE_64K

Signed-off-by: Jean-Luc Léger <jean-luc.leger@dspnet.fr.eu.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[CPUFREQ] powernow-k8.c: fix a check-after-use
Adrian Bunk [Tue, 18 Apr 2006 22:07:28 +0000 (00:07 +0200)]
[CPUFREQ] powernow-k8.c: fix a check-after-use

This patch fixes a check-after-use introduced by commit
4211a30349e8d2b724cfb4ce2584604f5e59c299 and spotted by the Coverity checker.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dave Jones <davej@redhat.com>
18 years ago[CPUFREQ] Remove duplicate check in powernow-k8
Tobias Klauser [Mon, 10 Apr 2006 17:20:12 +0000 (19:20 +0200)]
[CPUFREQ] Remove duplicate check in powernow-k8

Remove a duplicate NULL pointer check introduced by commit
4211a30349e8d2b724cfb4ce2584604f5e59c299

Signed-off-by: Tobias Klauser <tklauser@nuerscht.ch>
Signed-off-by: Dave Jones <davej@redhat.com>
18 years ago[CPUFREQ] drivers/cpufreq/cpufreq.c: static functions mustn't be exported
Adrian Bunk [Tue, 18 Apr 2006 15:06:13 +0000 (17:06 +0200)]
[CPUFREQ] drivers/cpufreq/cpufreq.c: static functions mustn't be exported

This patch removes the EXPORT_SYMBOL_GPL of the static function cpufreq_parse_governor().

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Dave Jones <davej@redhat.com>
18 years ago[CPUFREQ] If max_freq got reduced (e.g. by _PPC) a write to sysfs scaling_governor...
Thomas Renninger [Thu, 13 Apr 2006 13:14:04 +0000 (15:14 +0200)]
[CPUFREQ] If max_freq got reduced (e.g. by _PPC) a write to sysfs scaling_governor let cpufreq core stuck at low max_freq for ever

The previous patch had bugs (locking and refcount).

This one could also be related to the latest DELL reports.
But they only slip into this if a user prog (e.g. powersave daemon does when
AC got (un) plugged due to a scheme change) echos something to
/sys/../cpufreq/scaling_governor
while the frequencies got limited by BIOS.

This one works:

Subject: Max freq stucks at low freq if reduced by _PPC and sysfs gov access

The problem is reproducable by(if machine is limiting freqs via BIOS):
 - Unplugging AC -> max freq gets limited
 - echo ${governor} >/sys/.../cpufreq/scaling_governor (policy->user_data.max
   gets overridden with policy->max and will never come up again.)

This patch exchanged the cpufreq_set_policy call to __cpufreq_set_policy and
duplicated it's functionality but did not override user_data.max.
The same happens with overridding min/max values. If freqs are limited and
you override the min freq value, the max freq global value will also get
stuck to the limited freq, even if BIOS allows all freqs again.
Last scenario does only happen if BIOS does not reduce the frequency
to the lowest value (should never happen, just for correctness...)

 drivers/cpufreq/cpufreq.c |   17 +++++++++++++++--
 1 files changed, 15 insertions(+), 2 deletions(-)

Signed-off-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: "Pallipadi, Venkatesh" <venkatesh.pallipadi@intel.com>
Signed-off-by: Dave Jones <davej@redhat.com>
18 years agoMerge ../linus
Dave Jones [Tue, 18 Apr 2006 22:19:55 +0000 (17:19 -0500)]
Merge ../linus

18 years ago[ARM] 3479/1: Corgi SSP: Fix potential concurrent access problem
Richard Purdie [Tue, 18 Apr 2006 22:18:54 +0000 (23:18 +0100)]
[ARM] 3479/1: Corgi SSP: Fix potential concurrent access problem

Patch from Richard Purdie

corgi_ssp_probe() should not access GPDR directly but should use
pxa_gpio_mode() which has appropriate locking and other safeguards.

Signed-off-by: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
18 years ago[ARM] 3478/1: SharpSL SCOOP: Fix potenial build failure
Richard Purdie [Tue, 18 Apr 2006 22:18:53 +0000 (23:18 +0100)]
[ARM] 3478/1: SharpSL SCOOP: Fix potenial build failure

Patch from Richard Purdie

Move platform_scoop_config from the SharpSL scoop PCMCIA driver to
the SCOOP driver. This avoids build failures when PCMCIA is not built
or is modular (scoop.c itself cannot be modular).

Signed-off-by: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
18 years agox86: be careful about tailcall breakage for sys_open[at] too
Linus Torvalds [Tue, 18 Apr 2006 20:22:59 +0000 (13:22 -0700)]
x86: be careful about tailcall breakage for sys_open[at] too

Came up through a quick grep for other cases similar to the ftruncate()
one in commit 0a489cb3b6a7b277030cdbc97c2c65905db94536.

Also, add a comment, so that people who read the code understand why we
do what looks like a no-op.

(Again, this won't actually matter to any sane user, since libc will
save and restore the register gcc stomps on, but it's still wrong to
stomp on it)

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agox86: don't allow tail-calls in sys_ftruncate[64]()
Linus Torvalds [Tue, 18 Apr 2006 20:02:48 +0000 (13:02 -0700)]
x86: don't allow tail-calls in sys_ftruncate[64]()

Gcc thinks it owns the incoming argument stack, but that's not true for
"asmlinkage" functions, and it corrupts the caller-set-up argument stack
when it pushes the third argument onto the stack.  Which can result in
%ebx getting corrupted in user space.

Now, normally nobody sane would ever notice, since libc will save and
restore %ebx anyway over the system call, but it's still wrong.

I'd much rather have "asmlinkage" tell gcc directly that it doesn't own
the stack, but no such attribute exists, so we're stuck with our hacky
manual "prevent_tail_call()" macro once more (we've had the same issue
before with sys_waitpid() and sys_wait4()).

Thanks to Hans-Werner Hilse <hilse@sub.uni-goettingen.de> for reporting
the issue and testing the fix.

Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Tue, 18 Apr 2006 17:49:11 +0000 (10:49 -0700)]
Merge branch 'drm-patches' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-patches' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm: Fix further issues in drivers/char/drm/via_irq.c
  drivers/char/drm/drm_memory.c: possible cleanups
  drm: deline a few large inlines in DRM code
  drm: remove master setting from add/remove context
  drm: drm_pci needs dma-mapping.h
  [PATCH] drm: Fix issue reported by Coverity in drivers/char/drm/via_irq.c

18 years ago[PATCH] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.
Andi Kleen [Tue, 18 Apr 2006 10:35:22 +0000 (12:35 +0200)]
[PATCH] i386: Move CONFIG_DOUBLEFAULT into arch/i386 where it belongs.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Add tee and sync_file_range
Andi Kleen [Tue, 18 Apr 2006 10:35:19 +0000 (12:35 +0200)]
[PATCH] x86_64: Add tee and sync_file_range

tee was already there for some reason for native 64bit, but
sys_sync_file_range was missing. Also add it to the compat layer.

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Increase NUMA hash function nodemap
Andi Kleen [Tue, 18 Apr 2006 10:35:16 +0000 (12:35 +0200)]
[PATCH] x86_64: Increase NUMA hash function nodemap

Needed for some big Opteron systems to compute a numa hash function
They have more than 12 bits significant address.

TBD switch this over to dynamic allocation or use better hash

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: x86_64 add crashdump trigger points
Vivek Goyal [Tue, 18 Apr 2006 10:35:13 +0000 (12:35 +0200)]
[PATCH] x86_64: x86_64 add crashdump trigger points

o Start booting into the capture kernel after an Oops if system is in a
  unrecoverable state. System will boot into the capture kernel, if one is
  pre-loaded by the user, and capture the kernel core dump.

o One of the following conditions should be true to trigger the booting of
  capture kernel.
        - panic_on_oops is set.
        - pid of current thread is 0
        - pid of current thread is 1
        - Oops happened inside interrupt context.

Signed-off-by: Vivek Goyal <vgoyal@in.ibm.com>
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] i386/x86-64: Fix ACPI disabled LAPIC handling mismerge
Andi Kleen [Tue, 18 Apr 2006 10:35:10 +0000 (12:35 +0200)]
[PATCH] i386/x86-64: Fix ACPI disabled LAPIC handling mismerge

The patch I submitted earlier to fix disabled LAPIC handling in ACPI
was mismerged for some reason I still don't quite understand. Parts
of it was applied to the wrong function.

This patch fixes it up.

Cc: len.brown@intel.com
Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] x86_64: Update defconfig
Andi Kleen [Tue, 18 Apr 2006 10:35:07 +0000 (12:35 +0200)]
[PATCH] x86_64: Update defconfig

Signed-off-by: Andi Kleen <ak@suse.de>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge branch 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block
Linus Torvalds [Tue, 18 Apr 2006 17:36:49 +0000 (10:36 -0700)]
Merge branch 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block

* 'for-linus' of git://brick.kernel.dk/data/git/linux-2.6-block:
  [PATCH] cfq: Further rbtree traversal and cfq_exit_queue() race fix
  [PATCH 2/2] cfq: fix cic's rbtree traversal
  [PATCH 1/2] iosched: fix typo and barrier()

18 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc-merge
Linus Torvalds [Tue, 18 Apr 2006 17:34:24 +0000 (10:34 -0700)]
Merge git://git./linux/kernel/git/paulus/powerpc-merge

* git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc-merge:
  powerpc: Use correct sequence for putting CPU into nap mode
  [PATCH] spufs: fix context-switch decrementer code
  [PATCH] powerpc32: Set cpu explicitly in kernel compiles
  [PATCH] powerpc/pseries: bugfix: balance calls to pci_device_put
  [PATCH] powerpc: Fix machine detection in prom_init.c
  [PATCH] ppc32: Fix string comparing in platform_notify_map
  [PATCH] powerpc: Avoid __initcall warnings
  [PATCH] powerpc: Ensure runlatch is off in the idle loop
  powerpc: Fix CHRP booting - needs a define_machine call
  powerpc: iSeries has only 256 IRQs

18 years ago[PATCH] cfq: Further rbtree traversal and cfq_exit_queue() race fix
OGAWA Hirofumi [Tue, 18 Apr 2006 17:18:31 +0000 (19:18 +0200)]
[PATCH] cfq: Further rbtree traversal and cfq_exit_queue() race fix

In current code, we are re-reading cic->key after dead cic->key check.
So, in theory, it may really re-read *after* cfq_exit_queue() seted NULL.

To avoid race, we copy it to stack, then use it. With this change, I
guess gcc will assign cic->key to a register or stack, and it wouldn't
be re-readed.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
18 years agopowerpc: Use correct sequence for putting CPU into nap mode
Paul Mackerras [Tue, 18 Apr 2006 11:49:11 +0000 (21:49 +1000)]
powerpc: Use correct sequence for putting CPU into nap mode

We weren't using the recommended sequence for putting the CPU into
nap mode.  When I changed the idle loop, for some reason 7447A cpus
started hanging when we put them into nap mode.  Changing to the
recommended sequence fixes that.

The complexity here is that the recommended sequence is a loop that
keeps putting the cpu back into nap mode.  Clearly we need some way
to break out of the loop when an interrupt (external interrupt,
decrementer, performance monitor) occurs.  Here we use a bit in
the thread_info struct to indicate that we need this, and the exception
entry code notices this and arranges for the exception to return
to the value in the link register, thus breaking out of the loop.
We use a new `local_flags' field in the thread_info which we can
alter without needing to use an atomic update sequence.

The PPC970 has the same recommended sequence, so we do the same thing
there too.

This also fixes a bug in the kernel stack overflow handling code on
32-bit, since it was causing a value that we needed in a register to
get trashed.

Signed-off-by: Paul Mackerras <paulus@samba.org>
18 years agodrm: Fix further issues in drivers/char/drm/via_irq.c
Jayachandran C [Tue, 11 Apr 2006 06:18:28 +0000 (23:18 -0700)]
drm: Fix further issues in drivers/char/drm/via_irq.c

Fix de-reference of 'dev_priv' before NULL check.

Signed-off-by: Jayachandran C. <c.jayachandran@gmail.com>
Cc: Dave Airlie <airlied@linux.ie>
Signed-off-by: Andrew Morton <akpm@osdl.org>
18 years agodrivers/char/drm/drm_memory.c: possible cleanups
Adrian Bunk [Tue, 11 Apr 2006 06:18:27 +0000 (23:18 -0700)]
drivers/char/drm/drm_memory.c: possible cleanups

- #if 0 the following unused global function:
  - drm_ioremap_nocache()

- make the following needlessly global functions static:
  - agp_remap()
  - drm_lookup_map()

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Dave Airlie <airlied@linux.ie>
Signed-off-by: Andrew Morton <akpm@osdl.org>
18 years ago[PATCH 2/2] cfq: fix cic's rbtree traversal
OGAWA Hirofumi [Tue, 18 Apr 2006 07:45:18 +0000 (09:45 +0200)]
[PATCH 2/2] cfq: fix cic's rbtree traversal

When queue dies, we set cic->key=NULL as dead mark. So, when we
traverse a rbtree, we must check whether it's still valid key. if it
was invalidated, drop it, then restart the traversal from top.

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
18 years ago[PATCH 1/2] iosched: fix typo and barrier()
OGAWA Hirofumi [Tue, 18 Apr 2006 07:44:06 +0000 (09:44 +0200)]
[PATCH 1/2] iosched: fix typo and barrier()

On rmmod path, cfq/as waits to make sure all io-contexts was
freed. However, it's using complete(), not wait_for_completion().

I think barrier() is not enough in here. To avoid the following case,
this patch replaces barrier() with smb_wmb().

cpu0 visibility cpu1
                [ioc_gnone=NULL,ioc_count=1]

ioc_gnone = &all_gone NULL,ioc_count=1
atomic_read(&ioc_count) NULL,ioc_count=1
wait_for_completion() NULL,ioc_count=0 atomic_sub_and_test()
NULL,ioc_count=0 if ( && ioc_gone)
    [ioc_gone==NULL,
    so doesn't call complete()]
   &all_gone,ioc_count=0

Signed-off-by: OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>
Signed-off-by: Jens Axboe <axboe@suse.de>
18 years ago[PATCH] IPC: access to unmapped vmalloc area in grow_ary()
Alexey Kuznetsov [Mon, 17 Apr 2006 11:39:23 +0000 (15:39 +0400)]
[PATCH] IPC: access to unmapped vmalloc area in grow_ary()

grow_ary() should not copy struct ipc_id_ary (it copies new->p, not
new). Due to this, memcpy() src pointer could hit unmapped vmalloc page
when near page boundary.

Found during OpenVZ stress testing

Signed-off-by: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Signed-off-by: Kirill Korotaev <dev@openvz.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] Fix MADV_REMOVE protection checking
Hugh Dickins [Mon, 17 Apr 2006 21:46:32 +0000 (22:46 +0100)]
[PATCH] Fix MADV_REMOVE protection checking

madvise_remove needs to respect file and mmap protections.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
[ Will the real CVE-2006-1524 stand up, please.. ]
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 18 Apr 2006 00:44:47 +0000 (17:44 -0700)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6:
  [IPV4]: ip_route_input panic fix

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/stable-2.6
Linus Torvalds [Tue, 18 Apr 2006 00:44:17 +0000 (17:44 -0700)]
Merge /pub/scm/linux/kernel/git/gregkh/stable-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/stable-2.6:
  [PATCH] shmat: stop mprotect from giving write permission to a readonly attachment (CVE-2006-1524)
  [PATCH] cciss: bug fix for crash when running hpacucli
  [PATCH] ext3: Fix missed mutex unlock
  [PATCH] Fix block device symlink name
  [PATCH] isd200: limit to BLK_DEV_IDE

18 years ago[IPV4]: ip_route_input panic fix
Stephen Hemminger [Tue, 18 Apr 2006 00:27:11 +0000 (17:27 -0700)]
[IPV4]: ip_route_input panic fix

This fixes http://bugzilla.kernel.org/show_bug.cgi?id=6388
The bug is caused by ip_route_input dereferencing skb->nh.protocol of
the dummy skb passed dow from inet_rtm_getroute (Thanks Thomas for seeing
it). It only happens if the route requested is for a multicast IP
address.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[PATCH] shmat: stop mprotect from giving write permission to a readonly attachment...
Hugh Dickins [Wed, 12 Apr 2006 21:34:27 +0000 (14:34 -0700)]
[PATCH] shmat: stop mprotect from giving write permission to a readonly attachment (CVE-2006-1524)

I found that all of 2.4 and 2.6 have been letting mprotect give write
permission to a readonly attachment of shared memory, whether or not IPC
would give the caller that permission.

SUS says "The behaviour of this function [mprotect] is unspecified if the
mapping was not established by a call to mmap", but I don't think we can
interpret that as allowing it to subvert IPC permissions.

I haven't tried 2.2, but the 2.2.26 source looks like it gets it right; and
the patch below reproduces that behaviour - mprotect cannot be used to add
write permission to a shared memory segment attached readonly.

This patch is simple, and I'm sure it's what we should have done in 2.4.0:
if you want to go on to switch write permission on and off with mprotect,
just don't attach the segment readonly in the first place.

However, we could have accumulated apps which attach readonly (even though
they would be permitted to attach read/write), and which subsequently use
mprotect to switch write permission on and off: it's not unreasonable.

I was going to add a second ipcperms check in do_shmat, to check for
writable when readonly, and if not writable find_vma and clear VM_MAYWRITE.
 But security_ipc_permission might do auditing, and it seems wrong to
report an attempt for write permission when there has been none.  Or we
could flag the vma as SHM, note the shmid or shp in vm_private_data, and
then get mprotect to check.

But the patch below is a lot simpler: I'd rather stick with it, if we can
convince ourselves somehow that it'll be safe.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] cciss: bug fix for crash when running hpacucli
Mike Miller [Mon, 10 Apr 2006 22:38:07 +0000 (15:38 -0700)]
[PATCH] cciss: bug fix for crash when running hpacucli

Fix a crash when running hpacucli with multiple logical volumes on a cciss
controller.  We were not properly initializing the disk->queue and causing
a fault.

Thanks to Hasso Tepper for reporting the problem.  Thanks to Steve Cameron
for root causing the problem.  Most of the patch just moves things around.
The fix is a one-liner.

Signed-off-by: Mike Miller <mike.miller@hp.com>
Signed-off-by: Stephen Cameron <steve.cameron@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] ext3: Fix missed mutex unlock
Ananiev, Leonid I [Tue, 11 Apr 2006 05:54:38 +0000 (22:54 -0700)]
[PATCH] ext3: Fix missed mutex unlock

Missed unlock_super()call is added in error condition code path.

Signed-off-by: Leonid Ananiev <leonid.i.ananiev@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] Fix block device symlink name
Stephen Rothwell [Mon, 10 Apr 2006 07:17:20 +0000 (00:17 -0700)]
[PATCH] Fix block device symlink name

As noted further on the this file, some block devices have a / in their
name, so fix the "block:..." symlink name the same as the /sys/block name.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] isd200: limit to BLK_DEV_IDE
Randy Dunlap [Wed, 5 Apr 2006 19:03:45 +0000 (12:03 -0700)]
[PATCH] isd200: limit to BLK_DEV_IDE

Limit USB_STORAGE_ISD200 to whatever BLK_DEV_IDE and USB_STORAGE
are set to (y, m) since isd200 calls ide_fix_driveid() in the
BLK_DEV_IDE code.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[SPARC64]: Export pcibios_resource_to_bus().
David S. Miller [Mon, 17 Apr 2006 20:34:44 +0000 (13:34 -0700)]
[SPARC64]: Export pcibios_resource_to_bus().

SYM2 driver uses it.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-serial
Linus Torvalds [Sat, 15 Apr 2006 23:02:04 +0000 (16:02 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-serial

* master.kernel.org:/home/rmk/linux-2.6-serial:
  [SERIAL] Update serial driver documentation

18 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Sat, 15 Apr 2006 23:01:39 +0000 (16:01 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 3477/1: ARM EABI: undefine removed syscalls
  [ARM] 3475/1: S3C2410: fix spelling mistake in SMDK partition table
  [ARM] 3474/1: S3C2440: USB rate writes wrong var to CLKDIVN

18 years ago[SERIAL] Update serial driver documentation
Russell King [Sat, 15 Apr 2006 19:46:11 +0000 (20:46 +0100)]
[SERIAL] Update serial driver documentation

Improve serial driver documentation:
- Remove CVS id.
- Update pointer to reference driver documentation.
- Add comments about new uart_write_console function.
- Add TIOCM_LOOP modem control bit description.
- Add commentry about enable_ms method being called multiple times.
- Add commentry about startup/shutdown method calling.
- Mention that dereferencing port->info after shutdown is invalid.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
18 years ago[ARM] 3477/1: ARM EABI: undefine removed syscalls
Nicolas Pitre [Sat, 15 Apr 2006 15:10:43 +0000 (16:10 +0100)]
[ARM] 3477/1: ARM EABI: undefine removed syscalls

Patch from Nicolas Pitre

Avoid confusion for libraries assuming that a given syscall is available
when corresponding symbol is defined.

Signed-off-by: Nicolas Pitre <nico@cam.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
18 years agoFix AT91RM9200 build breakage
David Brownell [Sat, 15 Apr 2006 01:05:38 +0000 (18:05 -0700)]
Fix AT91RM9200 build breakage

The at91_cf driver got out of sync with certain changes in the PCMCIA
layer, notably getting rid of some duplication of data ... causing the
version merged to kernel.org to fail compiling.

This patch gives the at91_cf platform device a new iomem resource, using
it so this new pcmcia scheme works.  It also cleans up some whitepsace
bugs that have accumulated over time (mostly too-long lines).

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[PATCH] kill unushed __put_task_struct_cb
Eric W. Biederman [Fri, 14 Apr 2006 18:43:15 +0000 (12:43 -0600)]
[PATCH] kill unushed __put_task_struct_cb

Somehow in the midst of dotting i's and crossing t's during
the merge up to rc1 we wound up keeping __put_task_struct_cb
when it should have been killed as it no longer has any users.
Sorry I probably should have caught this while it was
still in the -mm tree.

Having the old code there gets confusing when reading
through the code and trying to understand what is
happening.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years ago[efficeon-agp] Add missing memory mask
H. Peter Anvin [Sat, 15 Apr 2006 00:25:30 +0000 (17:25 -0700)]
[efficeon-agp] Add missing memory mask

Original patch by Benjamin Herrenschmidt after debugging by Brian Hinz.

Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Brian Hinz <bphinz@hotmail.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/sam/kbuild
Linus Torvalds [Sat, 15 Apr 2006 00:30:55 +0000 (17:30 -0700)]
Merge /pub/scm/linux/kernel/git/sam/kbuild

* master.kernel.org:/pub/scm/linux/kernel/git/sam/kbuild:
  kbuild: fix false section mismatch warnings

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 15 Apr 2006 00:10:20 +0000 (17:10 -0700)]
Merge /pub/scm/linux/kernel/git/davem/net-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6: (170 commits)
  commit 3d9dd7564d5d7c28eb87b14c13a23806484667f3
  Author: Zach Brown <zach.brown@oracle.com>
  Date:   Fri Apr 14 16:04:18 2006 -0700

      [PATCH] ip_output: account for fraggap when checking to add trailer_len

      During other work I noticed that ip_append_data() seemed to be forgetting to
      include the frag gap in its calculation of a fragment that consumes the rest of
      the payload.  Herbert confirmed that this was a bug that snuck in during a
      previous rework.

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
  commit 08d099974a09faf4cb11ffc46da87073fa132fc0
  Author: Linus Walleij <triad@df.lth.se>
  Date:   Fri Apr 14 16:03:33 2006 -0700

      [IRDA]: smsc-ircc2, smcinit support for ALi ISA bridges

  ...

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6
Linus Torvalds [Sat, 15 Apr 2006 00:09:41 +0000 (17:09 -0700)]
Merge /pub/scm/linux/kernel/git/davem/sparc-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/davem/sparc-2.6:
  commit 5fdef394953d8660c70cc27b27db421582c42bf9
  Author: David S. Miller <davem@sunset.davemloft.net>
  Date:   Fri Apr 14 15:29:32 2006 -0700

      [SPARC]: Hook up sys_tee() into syscall tables.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Sat, 15 Apr 2006 00:09:08 +0000 (17:09 -0700)]
Merge /linux/kernel/git/jejb/scsi-rc-fixes-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6: (679 commits)
  commit 7676f83aeb774e7a3abe6af06ec92b29488b5b79
  Author: James Bottomley <James.Bottomley@steeleye.com>
  Date:   Fri Apr 14 09:47:59 2006 -0500

      [SCSI] scsi_transport_sas: don't scan a non-existent end device

      Any end device that can't support any of the scanning protocols
      shouldn't be scanned, so set its id to -1 to prevent
      scsi_scan_target() being called for it.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
  commit 3c0c25b97c7d020ef07f6366cf1d668a8e980c7c
  Author: Moore, Eric <Eric.Moore@lsil.com>
  Date:   Thu Apr 13 16:08:17 2006 -0600

      [SCSI] mptfusion - fix panic in mptsas_slave_configure

      Driver panic when RAID logical volume was present when driver
      loaded, or when a RAID logical volume was created on the fly.
  ...

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/i2c-2.6
Linus Torvalds [Sat, 15 Apr 2006 00:08:41 +0000 (17:08 -0700)]
Merge /pub/scm/linux/kernel/git/gregkh/i2c-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/i2c-2.6: (78 commits)
  commit e97b81ddbb8b8c72b85330ac4a454a4513dcba8a
  Author: Mark M. Hoffman <mhoffman@lightlink.com>
  Date:   Thu Mar 23 16:50:25 2006 +0100

      [PATCH] i2c-parport: Make type parameter mandatory

      This patch forces the user to specify what type of adapter is present when
      loading i2c-parport or i2c-parport-light.  If none is specified, the driver
      init simply fails - instead of assuming adapter type 0.

      This alleviates the sometimes lengthy boot time delays which can be caused
      by accidentally building one of these into a kernel along with several i2c
      slave drivers that have lengthy probe routines (e.g. hwmon drivers).

      Kconfig and documentation updated accordingly.

Signed-off-by: Mark M. Hoffman <mhoffman@lightlink.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
  ...

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/pci-2.6
Linus Torvalds [Sat, 15 Apr 2006 00:08:18 +0000 (17:08 -0700)]
Merge /pub/scm/linux/kernel/git/gregkh/pci-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/pci-2.6: (169 commits)
  commit 78a596b4490e17b9990d87b9d468ef5bb70daa10
  Author: Adrian Bunk <bunk@stusta.de>
  Date:   Fri Mar 31 01:38:12 2006 -0800

      [PATCH] remove kernel/power/pm.c:pm_unregister()

      Since the last user is removed in -mm, we can now remove this long deprecated
      function.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
  commit 21440d313358043b0ce5e43b00ff3c9b35a8616c
  Author: David Brownell <david-b@pacbell.net>
  Date:   Sat Apr 1 10:21:52 2006 -0800

      [PATCH] dma doc updates

  ...

18 years agoMerge master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6
Linus Torvalds [Sat, 15 Apr 2006 00:07:57 +0000 (17:07 -0700)]
Merge /pub/scm/linux/kernel/git/gregkh/driver-2.6

* master.kernel.org:/pub/scm/linux/kernel/git/gregkh/driver-2.6: (158 commits)
  commit 4f705ae3e94ffaafe8d35f71ff4d5c499bb06814
  Author: Bjorn Helgaas <bjorn.helgaas@hp.com>
  Date:   Mon Apr 3 17:09:22 2006 -0700

      [PATCH] DMI: move dmi_scan.c from arch/i386 to drivers/firmware/

      dmi_scan.c is arch-independent and is used by i386, x86_64, and ia64.
      Currently all three arches compile it from arch/i386, which means that ia64
      and x86_64 depend on things in arch/i386 that they wouldn't otherwise care
      about.

      This is simply "mv arch/i386/kernel/dmi_scan.c drivers/firmware/" (removing
      trailing whitespace) and the associated Makefile changes.  All three
      architectures already set CONFIG_DMI in their top-level Kconfig files.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Andi Kleen <ak@muc.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Andrey Panin <pazke@orbita1.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
  ...

18 years ago[PATCH] ip_output: account for fraggap when checking to add trailer_len
Zach Brown [Fri, 14 Apr 2006 23:04:18 +0000 (16:04 -0700)]
[PATCH] ip_output: account for fraggap when checking to add trailer_len

During other work I noticed that ip_append_data() seemed to be forgetting to
include the frag gap in its calculation of a fragment that consumes the rest of
the payload.  Herbert confirmed that this was a bug that snuck in during a
previous rework.

Signed-off-by: Zach Brown <zach.brown@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IRDA]: smsc-ircc2, smcinit support for ALi ISA bridges
Linus Walleij [Fri, 14 Apr 2006 23:03:33 +0000 (16:03 -0700)]
[IRDA]: smsc-ircc2, smcinit support for ALi ISA bridges

From: Linus Walleij <triad@df.lth.se>

This patch enables support for ALi ISA bridges when we run the smcinit
code.  It is needed to properly configure some Toshiba laptops.

Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IRDA]: irda-usb, unregister netdev when patch upload fails
Samuel Ortiz [Fri, 14 Apr 2006 23:02:07 +0000 (16:02 -0700)]
[IRDA]: irda-usb, unregister netdev when patch upload fails

In the STIR421x case, when the firmware upload fails, we need to
unregister_netdev. Otherwise we hit a BUG on free_netdev(), if sysfs
is enabled.

Signed-off-by: Samuel Ortiz <samuel.ortiz@nokia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM] clip: add module info
Stephen Hemminger [Fri, 14 Apr 2006 23:01:26 +0000 (16:01 -0700)]
[ATM] clip: add module info

Add module information

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM] clip: notifier related cleanups
Stephen Hemminger [Fri, 14 Apr 2006 23:00:59 +0000 (16:00 -0700)]
[ATM] clip: notifier related cleanups

Cleanup some code around notifier.  Don't need (void) casts to ignore
return values, and use C90 style initializer. Just ignore unused device
events.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM] clip: get rid of PROC_FS ifdef
Stephen Hemminger [Fri, 14 Apr 2006 23:00:25 +0000 (16:00 -0700)]
[ATM] clip: get rid of PROC_FS ifdef

Don't need the ifdef here since create_proc_entry() is stubbed to
always return NULL.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM] clip: run through Lindent
Stephen Hemminger [Fri, 14 Apr 2006 22:59:37 +0000 (15:59 -0700)]
[ATM] clip: run through Lindent

Run CLIP driver through Lindent script to fix formatting.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM]: Clip timer race.
Stephen Hemminger [Fri, 14 Apr 2006 22:56:02 +0000 (15:56 -0700)]
[ATM]: Clip timer race.

By inspection, the clip idle timer code is racy on SMP.
Here is a safe version of timer management.
Untested, I don't have ATM hardware.

Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years agokbuild: fix false section mismatch warnings
Sam Ravnborg [Fri, 14 Apr 2006 21:54:13 +0000 (23:54 +0200)]
kbuild: fix false section mismatch warnings

Darren Jenkins <darrenrjenkins@gmail.com> pointed out a
number of false positives where we referenced variables
from a _driver variable.
Fix it by check for that pattern and ignore it.

Randy.Dunlap <rdunlap@xenotime.net> pointed out a similar
set of warnings for a number of scsi drivers.
In scsi world they misname their variables *_template or
*_sht so add these to list of variables that may have references
to .init.text with no warning.

Randy.Dunlap <rdunlap@xenotime.net> also pointed out a scsi driver
with many references to .exit.text from .rodata. This is compiler
generated references and we already ignore these for .init.text, so
ignore them for .exit.text also.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
18 years ago[SPARC]: Hook up sys_tee() into syscall tables.
David S. Miller [Fri, 14 Apr 2006 22:29:32 +0000 (15:29 -0700)]
[SPARC]: Hook up sys_tee() into syscall tables.

Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[ATM]: clip causes unregister hang
Stephen Hemminger [Fri, 14 Apr 2006 22:07:27 +0000 (15:07 -0700)]
[ATM]: clip causes unregister hang

If Classical IP over ATM module is loaded, its neighbor table gets
populated when permanent neighbor entries are created; but these entries
are not flushed when the device is removed. Since the entry never gets
flushed the unregister of the network device never completes.

This version of the patch also adds locking around the reference to
the atm arp daemon to avoid races with events and daemon state changes.
(Note: barrier() was never really safe)

Bug-reference: http://bugzilla.kernel.org/show_bug.cgi?id=6295
Signed-off-by: Stephen Hemminger <shemminger@osdl.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[XFRM]: Add documentation for async events.
Jamal Hadi Salim [Fri, 14 Apr 2006 22:05:16 +0000 (15:05 -0700)]
[XFRM]: Add documentation for async events.

Documentation to describe asynchronous xfrm events to help people
writting HA code in user space.

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[XFRM]: Fix aevent timer.
Jamal Hadi Salim [Fri, 14 Apr 2006 22:03:05 +0000 (15:03 -0700)]
[XFRM]: Fix aevent timer.

Send aevent immediately if we have sent nothing since last timer and
this is the first packet.

Fixes a corner case when packet threshold is very high, the timer low
and a very low packet rate input which is bursty.

Signed-off-by: Jamal Hadi Salim <hadi@cyberus.ca>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[IPV4]: Possible cleanups.
Adrian Bunk [Wed, 12 Apr 2006 20:57:59 +0000 (13:57 -0700)]
[IPV4]: Possible cleanups.

This patch contains the following possible cleanups:
- make the following needlessly global function static:
  - arp.c: arp_rcv()
- remove the following unused EXPORT_SYMBOL's:
  - devinet.c: devinet_ioctl
  - fib_frontend.c: ip_rt_ioctl
  - inet_hashtables.c: inet_bind_bucket_create
  - inet_hashtables.c: inet_bind_hash
  - tcp_input.c: sysctl_tcp_abc
  - tcp_ipv4.c: sysctl_tcp_tw_reuse
  - tcp_output.c: sysctl_tcp_mtu_probing
  - tcp_output.c: sysctl_tcp_base_mss

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
18 years ago[SCSI] scsi_transport_sas: don't scan a non-existent end device
James Bottomley [Fri, 14 Apr 2006 14:47:59 +0000 (09:47 -0500)]
[SCSI] scsi_transport_sas: don't scan a non-existent end device

Any end device that can't support any of the scanning protocols
shouldn't be scanned, so set its id to -1 to prevent
scsi_scan_target() being called for it.

Signed-off-by: James Bottomley <James.Bottomley@SteelEye.com>
18 years ago[PATCH] remove kernel/power/pm.c:pm_unregister()
Adrian Bunk [Fri, 31 Mar 2006 09:38:12 +0000 (01:38 -0800)]
[PATCH] remove kernel/power/pm.c:pm_unregister()

Since the last user is removed in -mm, we can now remove this long deprecated
function.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Cc: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] dma doc updates
David Brownell [Sat, 1 Apr 2006 18:21:52 +0000 (10:21 -0800)]
[PATCH] dma doc updates

This updates the DMA API documentation to address a few issues:

 - The dma_map_sg() call results are used like pci_map_sg() results:
   using sg_dma_address() and sg_dma_len().  That's not wholly obvious
   to folk reading _only_ the "new" DMA-API.txt writeup.

 - Buffers allocated by dma_alloc_coherent() may not be completely
   free of coherency concerns ... some CPUs also have write buffers
   that may need to be flushed.

 - Cacheline coherence issues are now mentioned as being among issues
   which affect dma buffers, and complicate/prevent using of static and
   (especially) stack based buffers with the DMA calls.

I don't think many drivers currently need to worry about flushing write
buffers, but I did hit it with one SOC using external SDRAM for DMA
descriptors:  without explicit writebuffer flushing, the on-chip DMA
controller accessed descriptors before the CPU completed the writes.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] PCI: Add PCI quirk for SMBus on the Asus A6VA notebook
Jean Delvare [Sat, 1 Apr 2006 14:46:35 +0000 (16:46 +0200)]
[PATCH] PCI: Add PCI quirk for SMBus on the Asus A6VA notebook

The Asus A6VA notebook was reported to need a PCI quirk to unhide
the SMBus.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] arch/i386/pci/irq.c - new VIA chipsets (fwd)
Grzegorz Janoszka [Tue, 21 Mar 2006 10:57:19 +0000 (11:57 +0100)]
[PATCH] arch/i386/pci/irq.c - new VIA chipsets (fwd)

I use 2.6.15.6 Linux kernel and found some problems. I have about 100
Linux boxes (all with the same (binary the same) kernel). Last time I have
upgraded all those boxes from 2.4.32 to 2.6.15.6 (first 2.6.15.1, next .2,
.4 and .6) and I have found some problems on VIA based PC's. Probably the
reason of this is that some VIA chipsets are unrecognized by IRQ router.

In line 586 there is: /* FIXME: add new ones for 8233/5 */

There were only a few of chipsets ID's there, some of my VIA chipsets were
not present and kernel used default IRQ router.

I have added three entries, so that the code looks like:

        case PCI_DEVICE_ID_VIA_82C596:
        case PCI_DEVICE_ID_VIA_82C686:
        case PCI_DEVICE_ID_VIA_8231:
        case PCI_DEVICE_ID_VIA_8233A:
        case PCI_DEVICE_ID_VIA_8235:
        case PCI_DEVICE_ID_VIA_8237:
        case PCI_DEVICE_ID_VIA_8237_SATA:
                /* FIXME: add new ones for 8233/5 */
                r->name = "VIA";
                r->get = pirq_via_get;
                r->set = pirq_via_set;
                return 1;
        }

The kernel goes fine but I haven't testes it for weeks, I'm just a moment
after reboot :)
One thing is different (better?):
Using previus kernel I had:
PCI: Via IRQ fixup for 0000:00:0f.1, from 255 to 0
now I have:
PCI: Via IRQ fixup for 0000:00:0f.1, from 255 to 11

Maybe it is good idea to add there some more VIA chipsets?
The ones I have added seem to be OK.

From: Grzegorz Janoszka <Grzegorz@Janoszka.pl>
Acked-by: Martin Mares <mj@ucw.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] PCI: rpaphp: remove init error condition
John Rose [Thu, 23 Mar 2006 20:21:14 +0000 (14:21 -0600)]
[PATCH] PCI: rpaphp: remove init error condition

The init function for the RPA PCI Hotplug driver returns -ENODEV in the
case that no hotplug-capable slots are detected in the system.  This is
bad, since hot-capable slots can be added after boot to a purely virtual
POWER partition.  This is also bad because DLPAR I/O operations depend
on the rpaphp module.

Change the rpaphp init module to return success for the case of
partitions that own no hotplug-capable slots at boot.  Such slots can be
dynamically added after boot.

Signed-off-by: John Rose <johnrose@austin.ibm.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] PCI: fix sparse warning about pci_bus_flags
Roland Dreier [Thu, 30 Mar 2006 20:55:10 +0000 (12:55 -0800)]
[PATCH] PCI: fix sparse warning about pci_bus_flags

Sparse warns about casting to a __bitwise type.  However, it's correct
to do when defining the enum for pci_bus_flags_t, so add a __force to
quiet the warnings.  This will fix getting

    include/linux/pci.h:100:26: warning: cast to restricted type

from sparse all over the build.

Signed-off-by: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] pci_ids.h: correct naming of 1022:7450 (AMD 8131 Bridge)
John W. Linville [Mon, 20 Mar 2006 19:33:56 +0000 (14:33 -0500)]
[PATCH] pci_ids.h: correct naming of 1022:7450 (AMD 8131 Bridge)

The naming of the constant defined for PCI ID 1022:7450 does not seem
to match the information at http://pciids.sourceforge.net/:

http://pci-ids.ucw.cz/iii/?i=1022

There 1022:7450 is listed as "AMD-8131 PCI-X Bridge" while 1022:7451
is listed as "AMD-8131 PCI-X IOAPIC".  Yet, the current definition for
0x7450 is PCI_DEVICE_ID_AMD_8131_APIC. It seems to me like that name
should map to 0x7451, while a name like PCI_DEVICE_ID_AMD_8131_BRIDGE
should map to 0x7450.

Signed-off-by: John W. Linville <linville@tuxdriver.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] PCI: MSI(X) save/restore for suspend/resume
Shaohua Li [Wed, 8 Feb 2006 09:11:38 +0000 (17:11 +0800)]
[PATCH] PCI: MSI(X) save/restore for suspend/resume

Add MSI(X) configure sapce save/restore in generic PCI helper.

Signed-off-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] DMI: move dmi_scan.c from arch/i386 to drivers/firmware/
Bjorn Helgaas [Tue, 4 Apr 2006 00:09:22 +0000 (17:09 -0700)]
[PATCH] DMI: move dmi_scan.c from arch/i386 to drivers/firmware/

dmi_scan.c is arch-independent and is used by i386, x86_64, and ia64.
Currently all three arches compile it from arch/i386, which means that ia64
and x86_64 depend on things in arch/i386 that they wouldn't otherwise care
about.

This is simply "mv arch/i386/kernel/dmi_scan.c drivers/firmware/" (removing
trailing whitespace) and the associated Makefile changes.  All three
architectures already set CONFIG_DMI in their top-level Kconfig files.

Signed-off-by: Bjorn Helgaas <bjorn.helgaas@hp.com>
Cc: Andi Kleen <ak@muc.de>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Andrey Panin <pazke@orbita1.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] pm: print name of failed suspend function
Andrew Morton [Thu, 23 Mar 2006 09:38:34 +0000 (01:38 -0800)]
[PATCH] pm: print name of failed suspend function

Print more diagnostic info to help identify the source of power management
suspend failures.

Example:

usb_hcd_pci_suspend(): pci_set_power_state+0x0/0x1af() returns -22
pci_device_suspend(): usb_hcd_pci_suspend+0x0/0x11b() returns -22
suspend_device(): pci_device_suspend+0x0/0x34() returns -22

Work-in-progress.  It needs lots more suspend_report_result() calls sprinkled
everywhere.

Cc: Patrick Mochel <mochel@digitalimplant.org>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Nigel Cunningham <nigel@suspend2.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] driver core: driver_bind attribute returns incorrect value
Ryan Wilson [Wed, 22 Mar 2006 21:26:25 +0000 (16:26 -0500)]
[PATCH] driver core: driver_bind attribute returns incorrect value

The manual driver <-> device binding attribute in sysfs doesn't return
the correct value on failure or success of driver_probe_device.
driver_probe_device returns 1 on success (the driver accepted the
device) or 0 on probe failure (when the driver didn't accept the
device but no real error occured). However, the attribute can't just
return 0 or 1, it must return the number of bytes consumed from buf
or an error value. Returning 0 indicates to userspace that nothing
was written (even though the kernel has tried to do the bind/probe and
failed). Returning 1 indicates that only one character was accepted in
which case userspace will re-try the write with a partial string.

A more correct version of driver_bind would return count (to indicate
the entire string was consumed) when driver_probe_device returns 1
and -ENODEV when driver_probe_device returns 0. This patch makes that
change.

Signed-off-by: Ryan Wilson <hap9@epoch.ncsc.mil>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] driver core: fix unnecessary NULL check in drivers/base/class.c
Jayachandran C [Mon, 3 Apr 2006 19:31:53 +0000 (12:31 -0700)]
[PATCH] driver core: fix unnecessary NULL check in drivers/base/class.c

This patch tries to fix an issue in drivers/base/class.c, please
review and apply if correct.

Patch Description:
  "parent_class" is checked for NULL already, so removed the unnecessary
  check.

Signed-off-by: Jayachandran C. <c.jayachandran@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] BLOCK: delay all uevents until partition table is scanned
Kay Sievers [Fri, 24 Mar 2006 19:45:35 +0000 (20:45 +0100)]
[PATCH] BLOCK: delay all uevents until partition table is scanned

[BLOCK] delay all uevents until partition table is scanned

Here we delay the annoucement of all block device events until the
disk's partition table is scanned and all partition devices are already
created and sysfs is populated.

We have a bunch of old bugs for removable storage handling where we
probe successfully for a filesystem on the raw disk, but at the
same time the kernel recognizes a partition table and creates partition
devices.
Currently there is no sane way to tell if partitions will show up or not
at the time the disk device is announced to userspace. With the delayed
events we can simply skip any probe for a filesystem on the raw disk when
we find already present partitions.

Signed-off-by: Kay Sievers <kay.sievers@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] driver core: safely unbind drivers for devices not on a bus
Alan Stern [Fri, 31 Mar 2006 16:52:25 +0000 (11:52 -0500)]
[PATCH] driver core: safely unbind drivers for devices not on a bus

This patch (as667) changes the __device_release_driver() routine to
prevent it from crashing when it runs across a device not on any bus.
This seems logical, inasmuch as the corresponding bus_add_device()
routine has an explicit check allowing it to accept such devices.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] sysfs: Allow sysfs attribute files to be pollable
NeilBrown [Mon, 20 Mar 2006 06:53:53 +0000 (17:53 +1100)]
[PATCH] sysfs: Allow sysfs attribute files to be pollable

It works like this:
  Open the file
  Read all the contents.
  Call poll requesting POLLERR or POLLPRI (so select/exceptfds works)
  When poll returns,
     close the file and go to top of loop.
   or lseek to start of file and go back to the 'read'.

Events are signaled by an object manager calling
   sysfs_notify(kobj, dir, attr);

If the dir is non-NULL, it is used to find a subdirectory which
contains the attribute (presumably created by sysfs_create_group).

This has a cost of one int  per attribute, one wait_queuehead per kobject,
one int per open file.

The name "sysfs_notify" may be confused with the inotify
functionality.  Maybe it would be nice to support inotify for sysfs
attributes as well?

This patch also uses sysfs_notify to allow /sys/block/md*/md/sync_action
to be pollable

Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] i2c-parport: Make type parameter mandatory
Mark M. Hoffman [Thu, 23 Mar 2006 15:50:25 +0000 (16:50 +0100)]
[PATCH] i2c-parport: Make type parameter mandatory

This patch forces the user to specify what type of adapter is present when
loading i2c-parport or i2c-parport-light.  If none is specified, the driver
init simply fails - instead of assuming adapter type 0.

This alleviates the sometimes lengthy boot time delays which can be caused
by accidentally building one of these into a kernel along with several i2c
slave drivers that have lengthy probe routines (e.g. hwmon drivers).

Kconfig and documentation updated accordingly.

Signed-off-by: Mark M. Hoffman <mhoffman@lightlink.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] i2c-sis96x: Remove an init-time log message
Mark M. Hoffman [Thu, 23 Mar 2006 15:49:34 +0000 (16:49 +0100)]
[PATCH] i2c-sis96x: Remove an init-time log message

This patch removes an init-time kernel log message.
http://marc.theaimsgroup.com/?l=linux-kernel&m=114232987208628&w=3

Signed-off-by: Mark M. Hoffman <mhoffman@lightlink.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] w83792d: Be quiet on misdetection
Jean Delvare [Thu, 23 Mar 2006 15:38:21 +0000 (16:38 +0100)]
[PATCH] w83792d: Be quiet on misdetection

Make the w83792d driver keep quiet when misdetecting a chip. This can
happen, and the user doesn't need to know.

Also renumber the messages, and add one, for consistency.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] i2c: convert m41t00 to use a workqueue
Mark A. Greer [Fri, 31 Mar 2006 21:06:03 +0000 (23:06 +0200)]
[PATCH] i2c: convert m41t00 to use a workqueue

The m41t00 i2c/rtc driver currently uses a tasklet to schedule
interrupt-level writes to the rtc.  This patch causes the driver
to use a workqueue instead.

Signed-off-by: Mark A. Greer <mgreer@mvista.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] i2c: convert ds1374 to use a workqueue
Jean Delvare [Fri, 31 Mar 2006 21:05:01 +0000 (23:05 +0200)]
[PATCH] i2c: convert ds1374 to use a workqueue

A tasklet is not suitable for what the ds1374 driver does: neither sleeping
nor mutex operations are allowed in tasklets, and ds1374_set_tlet may do
both.

We can use a workqueue instead, where both sleeping and mutex operations
are allowed.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Acked-by: Randy Vinson <rvinson@mvista.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: add driver for funsoft usb serial device
Greg Kroah-Hartman [Thu, 13 Apr 2006 15:09:52 +0000 (08:09 -0700)]
[PATCH] USB: add driver for funsoft usb serial device

Cc: David Clare <david@funsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: remove __init from usb_console_setup
Paul Fulghum [Wed, 12 Apr 2006 21:41:59 +0000 (23:41 +0200)]
[PATCH] USB: remove __init from usb_console_setup

This prevents an Oops if booted with "console=ttyUSB0" but without a
USB-serial dongle, and plugged one in afterwards.

Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: pl2303: added support for OTi's DKU-5 clone cable
Tomasz Kazmierczak [Thu, 6 Apr 2006 20:07:12 +0000 (22:07 +0200)]
[PATCH] USB: pl2303: added support for OTi's DKU-5 clone cable

This patch adds support for a clone of Nokia DKU-5 cable made by
Ours Technology Inc for Nokia phones with PopPort (Nokia 3100 and others).
The cable uses PL2303 USB-to-serial converter from Prolific Technology Inc.

Signed-off-by: Tomasz Kazmierczak <tomek.fizyk@op.pl>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: S3C2410: use clk_enable() to ensure 48MHz to OHCI core
Ben Dooks [Sun, 2 Apr 2006 00:45:00 +0000 (01:45 +0100)]
[PATCH] USB: S3C2410: use clk_enable() to ensure 48MHz to OHCI core

Get the "usb-bus" clock and ensure it is enabled
when the OHCI core is in use.

It seems that a few bootloaders do not enable the
UPLL at startup, which stops the OHCI core having
a 48MHz bus clock. The improvements to the clock
framework for the s3c24xx now allow the USB PLL
to be started and stopped when being used.

Signed-off-by: Ben Dooks <ben-linux@fluff.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] hid-core.c: fix "input irq status -32 received" for Silvercrest USB Keyboard
Jeffrey Vandenbroucke sign [Tue, 28 Mar 2006 23:21:36 +0000 (15:21 -0800)]
[PATCH] hid-core.c: fix "input irq status -32 received" for Silvercrest USB Keyboard

When not using this patch, the kernel will continuously return "input irq
status -32 received", while making the keyboard unusable.  This can be
easely resolved using HID_QUIRK_NOGET.  Vendor-ID and Device-ID should be
applied to hid-core.c, and making an entry to make use of it.

Signed-off-by: Jeffrey Vandenbroucke <jeffrey@wirehead.be>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: Greg KH <greg@kroah.com>
Cc: David Brownell <david-b@pacbell.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: UHCI: don't track suspended ports
Alan Stern [Tue, 4 Apr 2006 18:47:44 +0000 (14:47 -0400)]
[PATCH] USB: UHCI: don't track suspended ports

Someone recently posted a bug report where it turned out that uhci-hcd
was disagreeing with the UHCI controller over whether or not a port was
suspended: The driver thought it wasn't and the hardware thought it was.
This patch (as665) fixes the problem and simplifies the driver by
removing the internal state-tracking completely.  Now the driver just
asks the hardware whether a port is suspended.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: keyspan-remote bugfix
Michael Downey [Mon, 3 Apr 2006 14:58:07 +0000 (08:58 -0600)]
[PATCH] USB: keyspan-remote bugfix

Signed-off-by: Michael Downey <downey@zymeta.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: Rename ax8817x_func() to asix_func() and add utility functions to reduce...
David Hollis [Wed, 29 Mar 2006 01:15:42 +0000 (20:15 -0500)]
[PATCH] USB: Rename ax8817x_func() to asix_func() and add utility functions to reduce bloat

Now that the ASIX code is supporting more than just the AX88172 devices,
make the utility function names more generic: ax8817x_func -> asix_func.
Functions that are chip specific now indicate as such: ax88772_func.

Additionally, pull some common routines used in initialization and such
into simple functions to reduce the verbosity of certain functions such
as
the bind() routines and to make the error handling consistent across the
board.

Signed-off-by: David Hollis <dhollis@davehollis.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: linux/usb/net2280.h common definitions
Pete Zaitcev [Sun, 2 Apr 2006 18:21:26 +0000 (10:21 -0800)]
[PATCH] USB: linux/usb/net2280.h common definitions

Move common definitions for NET2280 to <linux/usb/net2280.h>, so that I can
use them in prism54usb (it is not merged yet, but I plan to do it soon).

Signed-off-by: Pete Zaitcev <zaitcev@redhat.com>
Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: g_ether, highspeed conformance fix
David Brownell [Sun, 2 Apr 2006 18:20:43 +0000 (10:20 -0800)]
[PATCH] USB: g_ether, highspeed conformance fix

Be sure to record the peripheral's ep0 maxpacket size BEFORE using
that to initialize the (high speed) device qualifier; that helps a
lot with USBCV testing.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: usbtest: scatterlist OUT data pattern testing
David Brownell [Sun, 2 Apr 2006 18:20:15 +0000 (10:20 -0800)]
[PATCH] USB: usbtest: scatterlist OUT data pattern testing

Previously, scatterlist tests didn't write patterned data.  Given how many
corner cases are addresed by them, this was a significant gap in Linux-USB
test coverage.  Moreover, when peripherals checked for correct data patterns,
false error reports would drown out the true ones.

This adds the pattern on the way OUT from the host, so scatterlist tests can
now be used to uncover bugs like host TX or peripheral RX paths failing for
back-to-back short packets.  It's easy enough to get an error there with at
least one of the {DMA,PIO}{RX,TX} code paths, or run into hardware races
that need to be defended against.

Note this patch doesn't add checking for correct data patterns on the way
IN from peripherals, just a FIXME for later.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
18 years ago[PATCH] USB: at91 usb driver supend/resume fixes
David Brownell [Mon, 3 Apr 2006 04:26:21 +0000 (20:26 -0800)]
[PATCH] USB: at91 usb driver supend/resume fixes

AT91: the two USB drivers (OHCI, UDC) got out of sync with various
usbcore and driver model PM updates; fix.

Also minor fixes to ohci:  whitespace/style, MODULE_ALIAS so coldplug works
using /sys/.../modalias, and turn off _both_ clocks during suspend.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>