13 years ago[POWERPC] clean up pseries hcall interfaces
Anton Blanchard [Tue, 18 Jul 2006 22:01:28 +0000]
[POWERPC] clean up pseries hcall interfaces

Our pseries hcall interfaces are out of control:

plpar_hcall_norets
plpar_hcall
plpar_hcall_8arg_2ret
plpar_hcall_4out
plpar_hcall_7arg_7ret
plpar_hcall_9arg_9ret

Create 3 interfaces to cover all cases:

plpar_hcall_norets: 7 arguments no returns
plpar_hcall: 6 arguments 4 returns
plpar_hcall9: 9 arguments 9 returns

There are only 2 cases in the kernel that need plpar_hcall9, hopefully
we can keep it that way.

Pass in a buffer to stash return parameters so we avoid the &dummy1,
&dummy2 madness.

Signed-off-by: Anton Blanchard <anton@samba.org>
--
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years agoMerge branch 'merge'
Paul Mackerras [Tue, 1 Aug 2006 00:37:25 +0000]
Merge branch 'merge'

13 years agoMerge branch 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc
Linus Torvalds [Mon, 31 Jul 2006 20:39:52 +0000]
Merge branch 'merge' of git://git./linux/kernel/git/paulus/powerpc

* 'merge' of git://git.kernel.org/pub/scm/linux/kernel/git/paulus/powerpc:
  [POWERPC] Minor comment fix for misc_64.S
  [POWERPC] Use H_CEDE on non-SMT
  [POWERPC] force 64bit mode in fwnmi handlers to workaround firmware bugs
  [POWERPC] PMAC_APM_EMU should depend on ADB_PMU
  [POWERPC] Fix new interrupt code (MPIC detection)
  [POWERPC] Fix new interrupt code (MPIC endianness)
  [POWERPC] Add cpufreq support for Xserve G5
  [POWERPC] Xserve G5 thermal control fixes
  [POWERPC] Fix mem= handling when the memory limit is > RMO size
  [POWERPC] More offb/bootx fixes
  [POWERPC] Fix legacy_serial.c error handling on 32 bits
  [POWERPC] Fix default clock for udbg_16550
  [POWERPC] Fix non-MPIC CHRPs with CONFIG_SMP set
  [POWERPC] Fix 32 bits warning in prom_init.c
  [POWERPC] Workaround Pegasos incorrect ISA "ranges"
  [POWERPC] fix up front-LED Kconfig

13 years ago[PATCH] rivafb/nvidiafb: race between register_framebuffer and *_bl_init
Guido Guenther [Sun, 30 Jul 2006 10:04:21 +0000]
[PATCH] rivafb/nvidiafb: race between register_framebuffer and *_bl_init

Since we now use the generic backlight infrastructure, I think we need to
call rivafb_bl_init before calling register_framebuffer since otherwise
rivafb_bl_init might race with the framebuffer layer already opening the
device and setting up the video mode.  In this case we might end up with a
not yet fully intialized backlight (info->bl_dev still NULL) when calling
riva_bl_set_power via rivafb_set_par/rivafb_load_video_mode and the kernel
dies without any further notice during boot.

This fixes booting current git on a PB 6,1.  In this case radeonfb/atyfb
would be affected too - I can fix that too but don't have any hardware to
test this on.

Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] nvidiafb: remove redundant CONFIG_PCI check
Arthur Othieno [Sun, 30 Jul 2006 10:04:20 +0000]
[PATCH] nvidiafb: remove redundant CONFIG_PCI check

CONFIG_FB_NVIDIA already depends on CONFIG_PCI in drivers/video/Kconfig.
Driver does an extra ``sanity check'' which is then redundant.

Signed-off-by: Arthur Othieno <apgo@patchbomb.org>
Cc: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] powermac: More powermac backlight fixes
Michael Hanselmann [Sun, 30 Jul 2006 10:04:19 +0000]
[PATCH] powermac: More powermac backlight fixes

This patch fixes several problems:
- The legacy backlight value might be set at interrupt time. Introduced
  a worker to prevent it from directly calling the backlight code.
- via-pmu allows the backlight to be grabbed, in which case we need to
  prevent other kernel code from changing the brightness.
- Don't send PMU requests in via-pmu-backlight when the machine is about
  to sleep or waking up.
- More Kconfig fixes.

Signed-off-by: Michael Hanselmann <linux-kernel@hansmi.ch>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] radeonfb sleep fixes
Volker Braun [Sun, 30 Jul 2006 10:04:18 +0000]
[PATCH] radeonfb sleep fixes

Many IBM Thinkpad T4* models and some R* and X* with radeon video cards draw
too much power when suspended to RAM, reducing drastically the battery
lifetime.  The solution is to enable suspend-to-D2 on these machines.  They
are whitelisted through their subsystem vendor/device ID.  This fixes
http://bugzilla.kernel.org/show_bug.cgi?id=3022

The patch introduces a framework to alter the pm_mode and reinit_func fields
of the radeonfb_info structure based on a whitelist.  This should facilitate
future hardware-dependent workarounds.  The workaround for the Samsung P35
that is already in the radeonfb code has been rewritten using this framework.

The behavior can be overridden with module options:

i)  video=radeonfb:force_sleep=1
    enable suspend-to-D2 also on non-whitelisted machines (useful for
    testing new notebook models),

ii) video=radeonfb:ignore_devlist=1
    Disable checking the whitelist and do not apply any workarounds.

Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: "Antonino A. Daplas" <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fbdev: statically link the framebuffer notification functions
Antonino A. Daplas [Sun, 30 Jul 2006 10:04:17 +0000]
[PATCH] fbdev: statically link the framebuffer notification functions

The backlight and lcd subsystems can be notified by the framebuffer layer
of blanking events.  However, these subsystems, as a whole, can function
independently from the framebuffer layer.  But in order to enable to the
lcd and backlight subsystems, the framebuffer has to be compiled also,
effectively sucking in a huge amount of unneeded code.

To prevent dependency problems, separate out the framebuffer notification
mechanism from the framebuffer layer and permanently link it to the kernel.

Signed-off-by: Antonino Daplas <adaplas@pol.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] 9p: fix fid behavior on failed remove
Eric Van Hensbergen [Sun, 30 Jul 2006 10:04:16 +0000]
[PATCH] 9p: fix fid behavior on failed remove

Based on a bug report from Russ Ross <russruss@gmail.com>

According to the spec:

"The remove request asks the file server both to remove the file
 represented by fid and to clunk the fid, even if the remove fails."

but the Linux client seems to expect the fid to be valid after a failed
remove attempt.  Specifically, I'm getting this behavior when attempting to
remove a non-empty directory.

Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] update KJ details
Randy Dunlap [Sun, 30 Jul 2006 10:04:16 +0000]
[PATCH] update KJ details

Use preferred email address.  Remove sf.net project reference.  It is no
longer used.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] 9p: fix marshalling bug in tcreate with empty extension field
Russ Ross [Sun, 30 Jul 2006 10:04:15 +0000]
[PATCH] 9p: fix marshalling bug in tcreate with empty extension field

Signed-off-by: Russ Ross <russross@gmail.com>
Signed-off-by: Eric Van Hensbergen <ericvh@gmail.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] ext3 -nobh option causes oops
Badari Pulavarty [Sun, 30 Jul 2006 10:04:14 +0000]
[PATCH] ext3 -nobh option causes oops

For files other than IFREG, nobh option doesn't make sense.  Modifications
to them are journalled and needs buffer heads to do that.  Without this
patch, we get kernel oops in page_buffers().

Signed-off-by: Badari Pulavarty <pbadari@us.ibm.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] timer: Fix tvec_bases initializer
Josh Triplett [Sun, 30 Jul 2006 10:04:14 +0000]
[PATCH] timer: Fix tvec_bases initializer

kernel/timer.c defines a (per-cpu) pointer to tvec_base_t, but initializes
it using { &a_tvec_base_t }, which sparse warns about; change this to just
&a_tvec_base_t.

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] freevxfs: Add missing lock_kernel() to vxfs_readdir
Josh Triplett [Sun, 30 Jul 2006 10:04:12 +0000]
[PATCH] freevxfs: Add missing lock_kernel() to vxfs_readdir

Commit 7b2fd697427e73c81d5fa659efd91bd07d303b0e in the historical GIT tree
stopped calling the readdir member of a file_operations struct with the big
kernel lock held, and fixed up all the readdir functions to do their own
locking.  However, that change added calls to unlock_kernel() in
vxfs_readdir, but no call to lock_kernel().  Fix this by adding a call to
lock_kernel().

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Cc: Christoph Hellwig <hch@lst.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] documentation: Documentation/initrd.txt
Thomas Horsley [Sun, 30 Jul 2006 10:04:12 +0000]
[PATCH] documentation: Documentation/initrd.txt

I spent a long time the other day trying to examine an initrd image on a
fedora core 5 system because the initrd.txt file is apparently obsolete.
Here is a patch which I hope will reduce future confusion for others.

Signed-off-by: Thomas Horsley <tom.horsley@ccur.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] ipc/msg.c: clean up coding style
Ingo Molnar [Sun, 30 Jul 2006 10:04:11 +0000]
[PATCH] ipc/msg.c: clean up coding style

Clean up ipc/msg.c to conform to Documentation/CodingStyle.  (before it was
an inconsistent hodepodge of various coding styles)

Verified that the before/after .o's are identical.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fuse: fix typo
Miklos Szeredi [Sun, 30 Jul 2006 10:04:10 +0000]
[PATCH] fuse: fix typo

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fuse: use jiffies_64
Miklos Szeredi [Sun, 30 Jul 2006 10:04:10 +0000]
[PATCH] fuse: use jiffies_64

It is entirely possible (though rare) that jiffies half-wraps around, while a
dentry/inode remains in the cache.  This could mean that the dentry/inode is
not invalidated for another half wraparound-time.

To get around this problem, use 64-bit jiffies.  The only problem with this is
that dentry->d_time is 32 bits on 32-bit archs.  So use d_fsdata as the high
32 bits.  This is an ugly hack, but far simpler, than having to allocate
private data just for this purpose.

Since 64-bit jiffies can be assumed never to wrap around, simple comparison
can be used, and a zero time value can represent "invalid".

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fuse: fix zero timeout
Miklos Szeredi [Sun, 30 Jul 2006 10:04:08 +0000]
[PATCH] fuse: fix zero timeout

An attribute and entry timeout of zero should mean, that the entity is
invalidated immediately after the operation.  Previously invalidation only
happened at the next clock tick.

Reported and tested by Craig Davies.

Signed-off-by: Miklos Szeredi <miklos@szeredi.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Fix trivial unwind info bug
Markus Armbruster [Sun, 30 Jul 2006 10:04:08 +0000]
[PATCH] Fix trivial unwind info bug

CFA needs to be adjusted upwards for push, and downwards for pop.
arch/i386/kernel/entry.S gets it wrong in one place.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Jan Beulich <jbeulich@novell.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] vDSO hash-style fix
Roland McGrath [Sun, 30 Jul 2006 10:04:06 +0000]
[PATCH] vDSO hash-style fix

The latest toolchains can produce a new ELF section in DSOs and
dynamically-linked executables.  The new section ".gnu.hash" replaces
".hash", and allows for more efficient runtime symbol lookups by the
dynamic linker.  The new ld option --hash-style={sysv|gnu|both} controls
whether to produce the old ".hash", the new ".gnu.hash", or both.  In some
new systems such as Fedora Core 6, gcc by default passes --hash-style=gnu
to the linker, so that a standard invocation of "gcc -shared" results in
producing a DSO with only ".gnu.hash".  The new ".gnu.hash" sections need
to be dealt with the same way as ".hash" sections in all respects; only the
dynamic linker cares about their contents.  To work with older dynamic
linkers (i.e.  preexisting releases of glibc), a binary must have the old
".hash" section.  The --hash-style=both option produces binaries that a new
dynamic linker can use more efficiently, but an old dynamic linker can
still handle.

The new section runs afoul of the custom linker scripts used to build vDSO
images for the kernel.  On ia64, the failure mode for this is a boot-time
panic because the vDSO's PT_IA_64_UNWIND segment winds up ill-formed.

This patch addresses the problem in two ways.

First, it mentions ".gnu.hash" in all the linker scripts alongside ".hash".
 This produces correct vDSO images with --hash-style=sysv (or old tools),
with --hash-style=gnu, or with --hash-style=both.

Second, it passes the --hash-style=sysv option when building the vDSO
images, so that ".gnu.hash" is not actually produced.  This is the most
conservative choice for compatibility with any old userland.  There is some
concern that some ancient glibc builds (though not any known old production
system) might choke on --hash-style=both binaries.  The optimizations
provided by the new style of hash section do not really matter for a DSO
with a tiny number of symbols, as the vDSO has.  If someone wants to use
=gnu or =both for their vDSO builds and worry less about that
compatibility, just change the option and the linker script changes will
make any choice work fine.

Signed-off-by: Roland McGrath <roland@redhat.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Kyle McMartin <kyle@mcmartin.ca>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Jeff Dike <jdike@addtoit.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Sam Ravnborg <sam@ravnborg.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] hwrng: fix geode probe error unwind
Michael Buesch [Sun, 30 Jul 2006 10:04:05 +0000]
[PATCH] hwrng: fix geode probe error unwind

The geode hwrng leaks an iomapped resource, if hwrng_register() fails.
This fixes it.

Signed-off-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] hwrng: fix intel probe error unwind
Michael Buesch [Sun, 30 Jul 2006 10:04:04 +0000]
[PATCH] hwrng: fix intel probe error unwind

The intel hwrng leaks an iomapped resource, if hwrng_register() failes.
This fixes it.

Signed-off-by: Michael Buesch <mb@bu3sch.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Fix kmem_cache_alloc() been documented twice
Rolf Eike Beer [Sun, 30 Jul 2006 10:04:04 +0000]
[PATCH] Fix kmem_cache_alloc() been documented twice

kmem_cache_alloc() was documented twice, but kmem_cache_zalloc() never.
Fix this obvious typo to get things right.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] reference rt-mutex-design in rtmutex.c
Steven Rostedt [Sun, 30 Jul 2006 10:04:03 +0000]
[PATCH] reference rt-mutex-design in rtmutex.c

In order to prevent Doc Rot, this patch adds a reference to the design
document for rtmutex.c in rtmutex.c.  So when someone needs to update or
change the design of that file they will know that a document actually
exists that explains the design (helping them change it), and hopefully
that they will update the document if they too change the design.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Add parentheses around arguments in the SH_DIV macro.
Uwe Zeisberger [Sun, 30 Jul 2006 10:04:02 +0000]
[PATCH] Add parentheses around arguments in the SH_DIV macro.

There is currently no affected user in the tree, but usage is less
surprising that way.

Signed-off-by: Uwe Zeisberger <Uwe_Zeisberger@digi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Reducing local_bh_enable/disable overhead in irqtrace
Tim Chen [Sun, 30 Jul 2006 10:04:02 +0000]
[PATCH] Reducing local_bh_enable/disable overhead in irqtrace

The recent changes from irqtrace feature has added overheads to
local_bh_disable and local_bh_enable that reduces UDP performance across
x86_64 and IA64, even though IA64 does not support the irqtrace feature.
Patch in question is

[PATCH]lockdep: irqtrace subsystem, core
http://www.kernel.org/git/?p=linux/kernel/git/torvalds/linux-2.6.git;a=c
ommit;h=de30a2b355ea85350ca2f58f3b9bf4e5bc007986

Prior to this patch, local_bh_disable was a short macro.  Now it is a
function which calls __local_bh_disable with added irq flags save and
restore.  The irq flags save and restore were also added to
local_bh_enable, probably for injecting the trace irqs code.

This overhead is on the generic code path across all architectures.  On a
IA_64 test machine (Itanium-2 1.6 GHz) running a benchmark like netperf's
UDP streaming test, the added overhead results in a drop of 3% in
throughput, as udp_sendmsg calls the local_bh_enable/disable several times.

Other workloads that have heavy usages of local_bh_enable/disable could
also be affected.  The patch ideally should not have affected IA-64
performance as it does not have IRQ tracing support.  A significant portion
of the overhead is in the added irq flags save and restore, which I think
is not needed if IRQ tracing is unused.  A suggested patch is attached
below that recovers the lost performance.  However, the "ifdef"s in the
patch are a bit ugly.

Signed-off-by: Tim Chen <tim.c.chen@intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Typo in ub clause of devices.txt
Pete Zaitcev [Sun, 30 Jul 2006 10:04:01 +0000]
[PATCH] Typo in ub clause of devices.txt

Change "Thrid" into "Third", and realign similarly to other entries.

Signed-off-by: Pete Zaitcev <zaitcev@redhat.com>
Cc: <device@lanana.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Fix typo in MAINTAINERS: s/DEVICS/DEVICES/
Josh Triplett [Sun, 30 Jul 2006 10:04:00 +0000]
[PATCH] Fix typo in MAINTAINERS: s/DEVICS/DEVICES/

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] ufs: remove incorrect unlock_kernel from failure path in ufs_symlink()
Josh Triplett [Sun, 30 Jul 2006 10:03:59 +0000]
[PATCH] ufs: remove incorrect unlock_kernel from failure path in ufs_symlink()

ufs_symlink, in one of its error paths, calls unlock_kernel without ever
having called lock_kernel(); fix this by creating and jumping to a new
label out_notlocked rather than the out label used after calling
lock_kernel().

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Cc: Evgeniy Dushistov <dushistov@mail.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] efs: add entry for EFS filesystem to MAINTAINERS as Orphan
Josh Triplett [Sun, 30 Jul 2006 10:03:58 +0000]
[PATCH] efs: add entry for EFS filesystem to MAINTAINERS as Orphan

The EFS filesystem does not have an entry in MAINTAINERS; add one, giving
the EFS filesystem and listing the status as Orphan, per the note on that
page saying "I'm no longer actively maintaining EFS".

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] efs: Remove incorrect unlock_kernel from failure path in efs_symlink_readpage()
Josh Triplett [Sun, 30 Jul 2006 10:03:58 +0000]
[PATCH] efs: Remove incorrect unlock_kernel from failure path in efs_symlink_readpage()

If efs_symlink_readpage hits the -ENAMETOOLONG error path, it will call
unlock_kernel without ever having called lock_kernel(); fix this by
creating and jumping to a new label fail_notlocked rather than the fail
label used after calling lock_kernel().

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Cc: Marcelo Tosatti <marcelo.tosatti@cyclades.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Remove incorrect unlock_kernel from allocation failure path in coda_open()
Josh Triplett [Sun, 30 Jul 2006 10:03:56 +0000]
[PATCH] Remove incorrect unlock_kernel from allocation failure path in coda_open()

Commit 398c53a757702e1e3a7a2c24860c7ad26acb53ed (in the historical GIT
tree) moved the lock_kernel() in coda_open after the allocation of a
coda_file_info struct, but left an unlock_kernel() in the allocation
failure error path; remove it.

Signed-off-by: Josh Triplett <josh@freedesktop.org>
Acked-by: Jan Harkes <jaharkes@cs.cmu.edu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Fix swsusp with PNP BIOS
Ondrej Zary [Sun, 30 Jul 2006 10:03:55 +0000]
[PATCH] Fix swsusp with PNP BIOS

swsusp is unable to suspend my machine (DTK FortisPro TOP-5A notebook) with
kernel 2.6.17.5 because it's unable to suspend PNP device 00:16 (mouse).

The problem is in PNP BIOS.  pnp_bus_suspend() calls pnp_stop_dev() for the
device if the device can be disabled according to pnp_can_disable().  The
problem is that pnpbios_disable_resources() returns -EPERM if the device is
not dynamic (!pnpbios_is_dynamic()) but insert_device() happily sets
PNP_DISABLE capability/flag even if the device is not dynamic.  So we try
to disable non-dynamic devices which will fail.  This patch prevents
insert_device() from setting PNP_DISABLE if the device is not dynamic and
fixes suspend on my system.

Signed-off-by: Ondrej Zary <linux@rainbow-software.org>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] inotify: fix deadlock found by lockdep
Arjan van de Ven [Sun, 30 Jul 2006 10:03:54 +0000]
[PATCH] inotify: fix deadlock found by lockdep

This is a real deadlock, a nice complex one:
(warning: long explanation follows so that Andrew can have a complete
patch description)

it's an ABCDA deadlock:

A iprune_mutex
B inode->inotify_mutex
C ih->mutex
D dev->ev_mutex

The AB relationship comes straight from invalidate_inodes()

int invalidate_inodes(struct super_block * sb)
{
        int busy;
        LIST_HEAD(throw_away);

        mutex_lock(&iprune_mutex);
        spin_lock(&inode_lock);
        inotify_unmount_inodes(&sb->s_inodes);

where inotify_umount_inodes() takes the
                mutex_lock(&inode->inotify_mutex);

The BC relationship comes directly from inotify_find_update_watch():
s32 inotify_find_update_watch(struct inotify_handle *ih, struct inode *inode,
                              u32 mask)
{
   ...
        mutex_lock(&inode->inotify_mutex);
        mutex_lock(&ih->mutex);

The CD relationship comes from inotify_rm_wd:
inotify_rm_wd does
        mutex_lock(&inode->inotify_mutex);
        mutex_lock(&ih->mutex)
and then calls inotify_remove_watch_locked() which calls
notify_dev_queue_event() which does
        mutex_lock(&dev->ev_mutex);

(this strictly is a BCD relationship)

The DA relationship comes from the most interesting part:

  [<ffffffff8022d9f2>] shrink_icache_memory+0x42/0x270
  [<ffffffff80240dc4>] shrink_slab+0x11d/0x1c9
  [<ffffffff802b5104>] try_to_free_pages+0x187/0x244
  [<ffffffff8020efed>] __alloc_pages+0x1cd/0x2e0
  [<ffffffff8025e1f8>] cache_alloc_refill+0x3f8/0x821
  [<ffffffff8020a5e5>] kmem_cache_alloc+0x85/0xcb
  [<ffffffff802db027>] kernel_event+0x2e/0x122
  [<ffffffff8021d61c>] inotify_dev_queue_event+0xcc/0x140

inotify_dev_queue_event schedules a kernel_event which does a
kmem_cache_alloc( , GFP_KERNEL) which may try to shrink slabs, including
the inode cache .. which then takes iprune_mutex.

And voila, there is an AB, a BC, a CD relationship (even a direct BCD),
and also now a DA relationship -> a circular type AB-BA deadlock but
involving 4 locks.

The solution is simple: kernel_event() is NOT allowed to use GFP_KERNEL,
but must use GFP_NOFS to not cause recursion into the VFS.

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Robert Love <rml@novell.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Add linux-mm mailing list for memory management in MAINTAINERS file
Steven Rostedt [Sun, 30 Jul 2006 10:03:53 +0000]
[PATCH] Add linux-mm mailing list for memory management in MAINTAINERS file

Since I didn't know about the linux-mm mailing list until I spammed all
those that had their names anywhere in the mm directory, I'm sending this
patch to add the linux-mm mailing list to the MAINTAINERS file.

Also, since mm is so broad, it doesn't have a single person to maintain it,
and thus no maintainer is listed.  I also left the status as Maintained,
since it obviously is.

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] pi-futex: missing pi_waiters plist initialization
Heiko Carstens [Sun, 30 Jul 2006 10:03:52 +0000]
[PATCH] pi-futex: missing pi_waiters plist initialization

Initialize init task's pi_waiters plist.  Otherwise cpu hotplug of cpu 0
might crash, since rt_mutex_getprio() accesses an uninitialized list head.

call chain which led to crash:

take_cpu_down
sched_idle_next
__setscheduler
rt_mutex_getprio

Using PLIST_HEAD_INIT in the INIT_TASK macro doesn't work unfortunately,
since the pi_waiters member is only conditionally present.

Cc: Arjan van de Ven <arjan@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] hide onboard graphics drivers on G5
Olaf Hering [Sun, 30 Jul 2006 10:03:52 +0000]
[PATCH] hide onboard graphics drivers on G5

Hide the video drivers for onboard graphics found in early PCI PowerMacs in
Apple G5 config files.

drivers/built-in.o: In function `.platinumfb_probe':
platinumfb.c:(.text+0x377a0): undefined reference to `.nvram_read_byte'
platinumfb.c:(.text+0x37830): undefined reference to `.nvram_read_byte'
drivers/built-in.o: In function `.control_init':
controlfb.c:(.init.text+0x1938): undefined reference to `.nvram_read_byte'
controlfb.c:(.init.text+0x1968): undefined reference to `.nvram_read_byte'
drivers/built-in.o: In function `.valkyriefb_init':
(.init.text+0x2300): undefined reference to `.nvram_read_byte'
drivers/built-in.o:(.init.text+0x239c): more undefined references to `.nvram_read_byte' follow

Signed-off-by: Olaf Hering <olh@suse.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] enable mac partition label per default on pmac
Olaf Hering [Sun, 30 Jul 2006 10:03:51 +0000]
[PATCH] enable mac partition label per default on pmac

Enable mac partition table support per default also for a powermac config.

Signed-off-by: Olaf Hering <olh@suse.de>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] pcmcia: fix ioctl GET_CONFIGURATION_INFO for pcmcia_cards
Daniel Ritz [Sun, 30 Jul 2006 10:03:49 +0000]
[PATCH] pcmcia: fix ioctl GET_CONFIGURATION_INFO for pcmcia_cards

Values displayed when by cardctl config are horribly wrong for 16bit cards.
 this fixes it up by not using memcpy() since source and target struct are
very different.

Signed-off-by: Daniel Ritz <daniel.ritz@gmx.ch>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] pcmcia: fix ioctl for GET_STATUS and GET_CONFIGURATION_INFO
Daniel Ritz [Sun, 30 Jul 2006 10:03:47 +0000]
[PATCH] pcmcia: fix ioctl for GET_STATUS and GET_CONFIGURATION_INFO

the p_dev == NULL checks are wrong.  the called functions handle a NULL
p_dev on their own.  w/o this patch output of cardcctl status and cardctl
config is broken for cardbus cards or when the slot is empty.

Signed-off-by: Daniel Ritz <daniel.ritz@gmx.ch>
Cc: Dominik Brodowski <linux@dominikbrodowski.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] mdacon: fix __init section warnings
Frederik Deweerdt [Sun, 30 Jul 2006 10:03:46 +0000]
[PATCH] mdacon: fix __init section warnings

WARNING: drivers/video/console/mdacon.o - Section mismatch: reference to .init.text: from .text between 'mdacon_startup' (at offset 0x123) and 'mdacon_init'
WARNING: drivers/video/console/mdacon.o - Section mismatch: reference to .init.text: from .text between 'mdacon_startup' (at offset 0x18b) and 'mdacon_init'

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] sgiioc4: Always share IRQ
Brent Casavant [Sun, 30 Jul 2006 10:03:46 +0000]
[PATCH] sgiioc4: Always share IRQ

The SGI IOC4 IDE device always shares an interrupt with other devices which
are part of IOC4.  As such, IDEPCI_SHARE_IRQ should always be enabled when
BLK_DEV_SGIIOC4 is enabled.

Signed-off-by: Brent Casavant <bcasavan@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Doc/SubmittingPatches cleanups
Randy Dunlap [Sun, 30 Jul 2006 10:03:45 +0000]
[PATCH] Doc/SubmittingPatches cleanups

A few cleanups to SubmittingPatches:
- mention SubmitChecklist
- remove mention of my simple patch script tools
- remove last-updated line

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Add DocBook documentation for workqueue functions
Rolf Eike Beer [Sun, 30 Jul 2006 10:03:42 +0000]
[PATCH] Add DocBook documentation for workqueue functions

kernel/workqueue.c was omitted from generating kernel documentation.  This
adds a new section "Workqueues and Kevents" and adds documentation for some
of the functions.

Some functions in this file already had DocBook-style comments, now they
finally become visible.

Signed-off-by: Rolf Eike Beer <eike-kernel@sf-tec.de>
Cc: "Randy.Dunlap" <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] pci/search: cleanups, add to kernel-api.tmpl
Randy Dunlap [Sun, 30 Jul 2006 10:03:41 +0000]
[PATCH] pci/search: cleanups, add to kernel-api.tmpl

Clean up kernel-doc comments in drivers/pci/search.c (line sizes and typos).

Enable that source file in DocBook/kernel-api.tmpl.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] kernel-doc: ignore __devinit
Randy Dunlap [Sun, 30 Jul 2006 10:03:41 +0000]
[PATCH] kernel-doc: ignore __devinit

Ignore __devinit in function definitions so that kernel-doc won't fail on
them.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fix kernel-api doc for kernel/resource.c
Randy Dunlap [Sun, 30 Jul 2006 10:03:40 +0000]
[PATCH] fix kernel-api doc for kernel/resource.c

insert_resource() was unexported, so kernel-doc needs to be told to search
kernel/resource.c for internal functions instead of exported functions so that
it won't report an error.

Signed-off-by: Randy Dunlap <rdunlap@xenotime.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fix cond_resched() fix
Jim Houston [Sun, 30 Jul 2006 10:03:39 +0000]
[PATCH] fix cond_resched() fix

In cond_resched_lock() it calls __resched_legal() before dropping the spin
lock.  __resched_legal() will always finds the preempt_count non-zero and
will prevent the call to __cond_resched().

The attached patch adds a parameter to __resched_legal() with the expected
preempt_count value.

Cc: Ingo Molnar <mingo@elte.hu>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] fix bad macro param in timer.c
Steven Rostedt [Sun, 30 Jul 2006 10:03:38 +0000]
[PATCH] fix bad macro param in timer.c

We have

#define INDEX(N) (base->timer_jiffies >> (TVR_BITS + N * TVN_BITS)) & TVN_MASK

and it's used via

list = varray[i + 1]->vec + (INDEX(i + 1));

So, due to underparenthesisation, this INDEX(i+1) is now a ...  (TVR_BITS + i
+ 1 * TVN_BITS)) ...

So this bugfix changes behaviour.  It worked before by sheer luck:

  "If i was anything but 0, it was broken.  But this was only used by
   s390 and arm.  Since it was for the next interrupt, could that next
   interrupt be a problem (going into the second cascade)? But it was
   probably seldom wrong.  That is, this would fail if the next
   interrupt was in the second cascade, and was wrapped.  Which may
   never of happened.  Also if it did happen, it would have just missed
   the interrupt.

   If an interrupt was missed, and no one was there to miss it, was it
   really missed :-)"

Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] cpu hotplug: use hotplug version of registration in late inits
Chandra Seetharaman [Sun, 30 Jul 2006 10:03:37 +0000]
[PATCH] cpu hotplug: use hotplug version of registration in late inits

Use hotplug version of register_cpu_notifier in late init functions.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Andi Kleen <ak@muc.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] cpu hotplug: fix hotplug cpu documentation for proper usage
Chandra Seetharaman [Sun, 30 Jul 2006 10:03:36 +0000]
[PATCH] cpu hotplug: fix hotplug cpu documentation for proper usage

Update hotplug cpu documentation to clearly state when to use
register_cpu_notifier() and register_hotcpu_notifier.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] cpu hotplug: replace __devinit* with __cpuinit* for cpu notifications
Chandra Seetharaman [Sun, 30 Jul 2006 10:03:35 +0000]
[PATCH] cpu hotplug: replace __devinit* with __cpuinit* for cpu notifications

Few of the callback functions and notifier blocks that are associated with cpu
notifications incorrectly have __devinit and __devinitdata.  They should be
__cpuinit and __cpuinitdata instead.

It makes no functional difference but wastes text area when CONFIG_HOTPLUG is
enabled and CONFIG_HOTPLUG_CPU is not.

This patch fixes all those instances.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Cc: Ashok Raj <ashok.raj@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] panic_on_oops: remove ssleep()
Horms [Sun, 30 Jul 2006 10:03:34 +0000]
[PATCH] panic_on_oops: remove ssleep()

This patch is part of an effort to unify the panic_on_oops behaviour across
all architectures that implement it.

It was pointed out to me by Andi Kleen that if an oops has occured in
interrupt context, then calling sleep() in the oops path will only cause a
panic, and that it would be really better for it not to be in the path at
all.

This patch removes the ssleep() call and reworks the console message
accordinly.  I have a slght concern that the resulting console message is
too long, feedback welcome.

For powerpc it also unifies the 32bit and 64bit behaviour.

Fror x86_64, this patch only updates the console message, as ssleep() is
already not present.

Signed-off-by: Horms <horms@verge.net.au>
Acked-by: Paul Mackerras <paulus@samba.org>
Cc: Russell King <rmk@arm.linux.org.uk>
Cc: "Luck, Tony" <tony.luck@intel.com>
Cc: Andi Kleen <ak@muc.de>
Cc: Chris Zankel <chris@zankel.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] always define IRQ_PER_CPU
Yoichi Yuasa [Sun, 30 Jul 2006 10:03:33 +0000]
[PATCH] always define IRQ_PER_CPU

Reduce the likelihood of someone accidentally introducing namespace
collisions.

Signed-off-by: Yoichi Yuasa <yoichi_yuasa@tripeaks.co.jp>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] nbd: Abort request on data reception failure
Michal Feix [Sun, 30 Jul 2006 10:03:32 +0000]
[PATCH] nbd: Abort request on data reception failure

When reading from nbd device, we need to receive all the data after
receiving reply packet from the server - otherwise such request will never
be ended.

If socket is closed right after accepting reply control packet and in the
middle of waiting for read data, nbd_read_stat() returns NULL and
nbd_end_request() is not called.

This patch fixes it.

Signed-off-by: Michal Feix <michal@feix.cz>
Acked-by: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] nbd: Check magic before doing anything else
Michal Feix [Sun, 30 Jul 2006 10:03:31 +0000]
[PATCH] nbd: Check magic before doing anything else

We should check magic sequence in reply packet before trying to find
request with it's request handle.  This also solves the problem with
"Unexpected reply" message beeing logged, when packet with invalid magic is
received.

Signed-off-by: Michal Feix <michal@feix.cz>
Acked-by: Paul Clements <paul.clements@steeleye.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] lockdep: annotate pktcdvd natural device hierarchy
Arjan van de Ven [Sun, 30 Jul 2006 10:03:30 +0000]
[PATCH] lockdep: annotate pktcdvd natural device hierarchy

The pkt_*_dev functions operate on not-this-blockdevice, and that is
sufficiently checked at setup time.  As a result there is a natural
hierarchy, which needs nesting annotations

Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Cc: Peter Osterlund <petero2@telia.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] IDE: Touch NMI watchdog during resume from STR
Michal Schmidt [Sun, 30 Jul 2006 10:03:29 +0000]
[PATCH] IDE: Touch NMI watchdog during resume from STR

When resuming from suspend-to-RAM, the NMI watchdog detects a lockup in
ide_wait_not_busy.  Here's a screenshot of the trace taken by a digital
camera: http://www.uamt.feec.vutbr.cz/rizeni/pom/DSC03510-2.JPG

Let's touch the NMI watchdog in ide_wait_not_busy.  The system then resumes
correctly from STR.

[akpm@osdl.org: modular build fix]
Signed-off-by: Michal Schmidt <xschmi00@stud.feec.vutbr.cz>
Acked-by: Alan Cox <alan@lxorguk.ukuu.org.uk>
Cc: Bartlomiej Zolnierkiewicz <B.Zolnierkiewicz@elka.pw.edu.pl>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] invalidate_bdev() speedup
Andrew Morton [Sun, 30 Jul 2006 10:03:28 +0000]
[PATCH] invalidate_bdev() speedup

We can immediately bail from invalidate_bdev() if the blockdev has no
pagecache.

This solves the huge IPI storms which hald is causing on the big ia64
machines when it polls CDROM drives.

Acked-by: Jes Sorensen <jes@sgi.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] v850: call init_page_count() instead of set_page_count()
Miles Bader [Sun, 30 Jul 2006 10:03:28 +0000]
[PATCH] v850: call init_page_count() instead of set_page_count()

Signed-off-by: Miles Bader <miles@gnu.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] v850: Remove symbol exports which duplicate global ones
Miles Bader [Sun, 30 Jul 2006 10:03:27 +0000]
[PATCH] v850: Remove symbol exports which duplicate global ones

Signed-off-by: Miles Bader <miles@gnu.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] IA64: kprobe invalidate icache of jump buffer
bibo, mao [Sun, 30 Jul 2006 10:03:26 +0000]
[PATCH] IA64: kprobe invalidate icache of jump buffer

Kprobe inserts breakpoint instruction in probepoint and then jumps to
instruction slot when breakpoint is hit, the instruction slot icache must
be consistent with dcache.  Here is the patch which invalidates instruction
slot icache area.

Without this patch, in some machines there will be fault when executing
instruction slot where icache content is inconsistent with dcache.

Signed-off-by: bibo,mao <bibo.mao@intel.com>
Acked-by: "Luck, Tony" <tony.luck@intel.com>
Acked-by: Keshavamurthy Anil S <anil.s.keshavamurthy@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] arch/alpha: Use ARRAY_SIZE macro
Tobias Klauser [Sun, 30 Jul 2006 10:03:23 +0000]
[PATCH] arch/alpha: Use ARRAY_SIZE macro

Use ARRAY_SIZE macro instead of sizeof(x)/sizeof(x[0]) and remove a
duplicate of the macro.  Also remove some trailing whitespaces and needless
braces.

Signed-off-by: Tobias Klauser <tklauser@distanz.ch>
Cc: Richard Henderson <rth@twiddle.net>
Cc: Ivan Kokshaysky <ink@jurassic.park.msu.ru>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] i386: fix CONFIG_EFI help
Arthur Othieno [Sun, 30 Jul 2006 10:03:22 +0000]
[PATCH] i386: fix CONFIG_EFI help

It is described as being experimental, but doesn't actually depend on
EXPERIMENTAL.  Change the text.

Signed-off-by: Arthur Othieno <apgo@patchbomb.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] kprobe-booster: disable in preemptible kernel
Masami Hiramatsu [Sun, 30 Jul 2006 10:03:21 +0000]
[PATCH] kprobe-booster: disable in preemptible kernel

The kprobe-booster's safety check against preemption does not work well
now, because the preemption count has been modified by read_rcu_lock() in
atomic_notifier_call_chain() before we check it.  So, I'd like to prevent
boosting kprobe temporarily if the kernel is preemptable.

Now we are searching for the good solution.

Signed-off-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Prasanna S Panchamukhi <prasanna@in.ibm.com>
Cc: Anil S Keshavamurthy <anil.s.keshavamurthy@intel.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] machine_kexec.c: Fix the description of segment handling
Eric W. Biederman [Sun, 30 Jul 2006 10:03:20 +0000]
[PATCH] machine_kexec.c: Fix the description of segment handling

One of my original comments in machine_kexec was unclear
and this should fix it.

Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: Andi Kleen <ak@muc.de>
Acked-by: Horms <horms@verge.net.au>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] synchronize_tsc() fixes
Andrew Morton [Sun, 30 Jul 2006 10:03:19 +0000]
[PATCH] synchronize_tsc() fixes

- Move the tsc synchronisation variables into a struct, mark it __initdata

- local `realdelta' wants to be 64-bit

- Print the skew for negative skews, as well as for positive ones

- remove dead code

Cc: john stultz <johnstul@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] selinux: fix bug in security_compute_sid
Venkat Yekkirala [Sun, 30 Jul 2006 10:03:18 +0000]
[PATCH] selinux: fix bug in security_compute_sid

Initializes newcontext sooner to allow for its destruction in all cases.

Signed-off-by: Venkat Yekkirala <vyekkirala@TrustedCS.com>
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] selinux: fix memory leak
Darrel Goeddel [Sun, 30 Jul 2006 10:03:17 +0000]
[PATCH] selinux: fix memory leak

This patch fixes a memory leak when a policydb structure is destroyed.

Signed-off-by: Darrel Goeddel <dgoeddel@trustedcs.com>
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Acked-by: James Morris <jmorris@namei.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] knfsd: Fix stale file handle problem with subtree_checking.
NeilBrown [Sun, 30 Jul 2006 10:03:16 +0000]
[PATCH] knfsd: Fix stale file handle problem with subtree_checking.

A recent commit (7fc90ec93a5eb71f4b08403baf5ba7176b3ec6b1) moved the
call to nfsd_setuser out of the 'find a dentry for a filehandle' branch
of fh_verify so that it would always be called.

This had the unfortunately side-effect of moving *after* the call to
decode_fh, so the prober fsuid was not set when nfsd_acceptable was called,
the 'permission' check did the wrong thing.

This patch moves the nfsd_setuser call back where it was, and add as call
in the other branch of the if.

Cc: "J. Bruce Fields" <bfields@fieldses.org>
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] mce section fix
Andrew Morton [Sun, 30 Jul 2006 10:03:16 +0000]
[PATCH] mce section fix

mce_disabled cannot be __initdata - we access it during APM resume.

Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Fix ppc32 zImage inflate
Peter Korsgaard [Sun, 30 Jul 2006 10:03:12 +0000]
[PATCH] Fix ppc32 zImage inflate

The recent zlib update (commit 4f3865fb57a04db7cca068fed1c15badc064a302)
broke ppc32 zImage decompression as it tries to decompress to address zero
and the updated zlib_inflate checks that strm->next_out isn't a null
pointer.

This little patch fixes it.

[rpurdie@rpsys.net: add comment]
Signed-off-by: Peter Korsgaard <jacmet@sunsite.dk>
Acked-by: Tom Rini <trini@kernel.crashing.org>
Signed-off-by: Richard Purdie <rpurdie@rpsys.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] delay accounting: temporarily enable by default
Shailabh Nagar [Sun, 30 Jul 2006 10:03:11 +0000]
[PATCH] delay accounting: temporarily enable by default

Enable delay accounting by default so that feature gets coverage testing
without requiring special measures.

Earlier, it was off by default and had to be enabled via a boot time param.
 This patch reverses the default behaviour to improve coverage testing.  It
can be removed late in the kernel development cycle if its believed users
shouldn't have to incur any cost if they don't want delay accounting.  Or
it can be retained forever if the utility of the stats is deemed common
enough to warrant keeping the feature on.

Signed-off-by: Shailabh Nagar <nagar@watson.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] taskstats: free skb, avoid returns in send_cpu_listeners
Shailabh Nagar [Sun, 30 Jul 2006 10:03:11 +0000]
[PATCH] taskstats: free skb, avoid returns in send_cpu_listeners

Add a missing freeing of skb in the case there are no listeners at all.
Also remove the returning of error values by the function as it is unused
by the sole caller.

Signed-off-by: Shailabh Nagar <nagar@watson.ibm.com>
Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] make taskstats sending completely independent of delay accounting on/off...
Shailabh Nagar [Sun, 30 Jul 2006 10:03:10 +0000]
[PATCH] make taskstats sending completely independent of delay accounting on/off status

Complete the separation of delay accounting and taskstats by ignoring the
return value of delay accounting functions that fill in parts of taskstats
before it is sent out (either in response to a command or as part of a task
exit).

Also make delayacct_add_tsk return silently when delay accounting is turned
off rather than treat it as an error.

Signed-off-by: Shailabh Nagar <nagar@watson.ibm.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] genirq: {en,dis}able_irq_wake() need refcounting too
David Brownell [Sun, 30 Jul 2006 10:03:08 +0000]
[PATCH] genirq: {en,dis}able_irq_wake() need refcounting too

IRQs need refcounting and a state flag to track whether the the IRQ should
be enabled or disabled as a "normal IRQ" source after a series of calls to
{en,dis}able_irq().  For shared IRQs, the IRQ must be enabled so long as at
least one driver needs it active.

Likewise, IRQs need the same support to track whether the IRQ should be
enabled or disabled as a "wakeup event" source after a series of calls to
{en,dis}able_irq_wake().  For shared IRQs, the IRQ must be enabled as a
wakeup source during sleep so long as at least one driver needs it.  But
right now they _don't have_ that refcounting ...  which means sharing a
wakeup-capable IRQ can't work correctly in some configurations.

This patch adds the refcount and flag mechanisms to set_irq_wake() -- which
is what {en,dis}able_irq_wake() call -- and minimal documentation of what
the irq wake mechanism does.

Drivers relying on the older (broken) "toggle" semantics will trigger a
warning; that'll be a handful of drivers on ARM systems.

Signed-off-by: David Brownell <dbrownell@users.sourceforge.net>
Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Russell King <rmk@arm.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] drivers/char/pc8736x_gpio.c: unexport a static struct
Adrian Bunk [Sun, 30 Jul 2006 10:03:07 +0000]
[PATCH] drivers/char/pc8736x_gpio.c: unexport a static struct

A static struct mustn't be exported.

Signed-off-by: Adrian Bunk <bunk@stusta.de>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] Process Events: Fix biarch compatibility issue. use __u64 timestamp
Chandra Seetharaman [Sun, 30 Jul 2006 10:03:04 +0000]
[PATCH] Process Events: Fix biarch compatibility issue. use __u64 timestamp

Events sent by Process Events Connector from a 64-bit kernel are not binary
compatible with a 32-bit userspace program because the "timestamp" field
(struct timespec) is not arch independent.  This affects the fields that
follow "timestamp" as they will be be off by 8 bytes.

This is a problem for 32-bit userspace programs running with 64-bit kernels
on ppc64, s390, x86-64..  any "biarch" system.

Matt had submitted a different solution to lkml as an RFC earlier.  We have
since switched to a solution recommended by Evgeniy Polyakov.

This patch fixes the problem by changing the timestamp to be a __u64, which
stores the number of nanoseconds.

Tested on a x86_64 system with both 32 bit application and 64 bit
application and on a i386 system.

Signed-off-by: Chandra Seetharaman <sekharan@us.ibm.com>
Signed-off-by: Matt Helsley <matthltc@us.ibm.com>
Cc: Evgeniy Polyakov <johnpol@2ka.mipt.ru>
Cc: Guillaume Thouvenin <guillaume.thouvenin@bull.net>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] ext3: avoid triggering ext3_error on bad NFS file handle
Neil Brown [Sun, 30 Jul 2006 10:03:01 +0000]
[PATCH] ext3: avoid triggering ext3_error on bad NFS file handle

The inode number out of an NFS file handle gets passed eventually to
ext3_get_inode_block() without any checking.  If ext3_get_inode_block()
allows it to trigger an error, then bad filehandles can have unpleasant
effect - ext3_error() will usually cause a forced read-only remount, or a
panic if `errors=panic' was used.

So remove the call to ext3_error there and put a matching check in
ext3/namei.c where inode numbers are read off storage.

[akpm@osdl.org: fix off-by-one error]
Signed-off-by: Neil Brown <neilb@suse.de>
Signed-off-by: Jan Kara <jack@suse.cz>
Cc: Marcel Holtmann <marcel@holtmann.org>
Cc: <stable@kernel.org>
Cc: "Stephen C. Tweedie" <sct@redhat.com>
Cc: Eric Sandeen <esandeen@redhat.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[PATCH] sched: build_sched_domains() fix
Siddha, Suresh B [Sun, 30 Jul 2006 10:02:59 +0000]
[PATCH] sched: build_sched_domains() fix

Use the correct groups while initializing sched groups power for
allnodes_domain.  This fixes the crash observed while creating exclusive
cpusets.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Reported-and-tested-by: Paul Jackson <pj@sgi.com>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>

13 years ago[POWERPC] Instrument Hypervisor Calls: merge headers
Mike Kravetz [Fri, 21 Jul 2006 06:39:51 +0000]
[POWERPC] Instrument Hypervisor Calls: merge headers

Move all the Hypervisor call definitions to to a single header file.

Signed-off-by: Mike Kravetz <kravetz@us.ibm.com>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] Simplify dma_ops bug conditions
Jeremy Kerr [Thu, 13 Jul 2006 06:32:52 +0000]
[POWERPC] Simplify dma_ops bug conditions

Use BUG_ON rather than BUG to simplify the dma_ops handing,
and remove the now-unnecessary return cases.

Booted on pseries.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] Make get_property() return a const void *
Jeremy Kerr [Wed, 12 Jul 2006 05:42:06 +0000]
[POWERPC] Make get_property() return a const void *

Previous changes have treated the return values of get_property as
const, so now we can make the actual change to get_property(). There
shouldn't be a need to cast the return values anymore.

We will now get compiler warnings when property values are assigned to
a non-const variable.

If properties need to be updated, there's still the of_find_property
function.

Built for cell_defconfig, chrp32_defconfig, g5_defconfig,
iseries_defconfig, maple_defconfig, pmac32_defconfig, ppc64_defconfig
and pseries_defconfig.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] tsi108: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:41:52 +0000]
[POWERPC] tsi108: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

tsi108 driver changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] sata_svw: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:41:41 +0000]
[POWERPC] sata_svw: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

sata_svw changes

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] tmp_atmel: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:41:30 +0000]
[POWERPC] tmp_atmel: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

tpm_atmel changes

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] sound: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:41:18 +0000]
[POWERPC] sound: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

powerpc-specific sound driver changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] netdevices: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:41:03 +0000]
[POWERPC] netdevices: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

powerpc-specific network device driver changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] scsi: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:40:51 +0000]
[POWERPC] scsi: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

powerpc-specific scsi driver changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] video & agp: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:40:40 +0000]
[POWERPC] video & agp: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

powerpc-specific video & agp driver changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] powermac: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:40:29 +0000]
[POWERPC] powermac: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

powermac platform & macintosh driver changes.

Built for pmac32_defconfig, g5_defconfig

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] maple: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:40:17 +0000]
[POWERPC] maple: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

maple platform changes.

Built for maple_defconfig

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] chrp: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:40:05 +0000]
[POWERPC] chrp: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

chrp platform changes.

Built for chrp32_defconfig

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] cell: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:39:54 +0000]
[POWERPC] cell: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

cell platform changes.

Built for cell_defconfig

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] mpc: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:39:42 +0000]
[POWERPC] mpc: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

mpc* platform changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] iseries: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:39:42 +0000]
[POWERPC] iseries: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

iseries platform changes.

Built for iseries_defconfig

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] pseries: Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:39:43 +0000]
[POWERPC] pseries: Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

pseries platform changes.

Built for pseries_defconfig

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] Constify & voidify get_property()
Jeremy Kerr [Wed, 12 Jul 2006 05:35:54 +0000]
[POWERPC] Constify & voidify get_property()

Now that get_property() returns a void *, there's no need to cast its
return value. Also, treat the return value as const, so we can
constify get_property later.

powerpc core changes.

Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>

13 years ago[POWERPC] briq_panel Kconfig fix
Andrew Morton [Sun, 9 Jul 2006 05:38:56 +0000]
[POWERPC] briq_panel Kconfig fix

drivers/char/briq_panel.c:28:22: error: asm/prom.h: No such file or directory

Cc: Jeremy Kerr <jk@ozlabs.org>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Paul Mackerras <paulus@samba.org>