pandora-kernel.git
14 years agorcu: Need to update rnp->gpnum if preemptable RCU is to be reliable
Paul E. McKenney [Tue, 8 Sep 2009 22:54:36 +0000 (15:54 -0700)]
rcu: Need to update rnp->gpnum if preemptable RCU is to be reliable

Without this patch, tasks preempted in RCU read-side critical
sections can fail to block the grace period, given that
rnp->gpnum is used to determine which rnp->blocked_tasks[]
element the preempted task is enqueued on.

Before the patch, rnp->gpnum is always zero, so preempted tasks
are always enqueued on rnp->blocked_tasks[0], which is correct
only when the current CPU has not checked into the current
grace period and the grace-period number is even, or,
similarly, if the current CPU -has- checked into the current
grace period and the grace-period number is odd.

Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Cc: laijs@cn.fujitsu.com
Cc: dipankar@in.ibm.com
Cc: akpm@linux-foundation.org
Cc: mathieu.desnoyers@polymtl.ca
Cc: josht@linux.vnet.ibm.com
Cc: dvhltc@us.ibm.com
Cc: niv@us.ibm.com
Cc: peterz@infradead.org
LKML-Reference: <12524504771622-git-send-email->
Signed-off-by: Ingo Molnar <mingo@elte.hu>
14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Linus Torvalds [Thu, 17 Sep 2009 16:55:52 +0000 (09:55 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/bp/bp

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp:
  amd64_edac: check NB MCE bank enable on the current node properly
  amd64_edac: Rewrite unganged mode code of f10_early_channel_count
  amd64_edac: cleanup amd64_check_ecc_enabled
  x86, EDAC: Provide function to return NodeId of a CPU
  amd64_edac: build driver only on AMD hardware

14 years agoMerge branch 'for-linus' of git://oss.sgi.com/xfs/xfs
Linus Torvalds [Thu, 17 Sep 2009 16:54:37 +0000 (09:54 -0700)]
Merge branch 'for-linus' of git://oss.sgi.com/xfs/xfs

* 'for-linus' of git://oss.sgi.com/xfs/xfs: (39 commits)
  xfs: includecheck fix for fs/xfs/xfs_iops.c
  xfs: switch to seq_file
  xfs: Record new maintainer information
  xfs: use correct log reservation when handling ENOSPC in xfs_create
  xfs: xfs_showargs() reports group *and* project quotas enabled
  xfs: un-static xfs_inobt_lookup
  xfs: actually enable the swapext compat handler
  xfs: simplify xfs_trans_iget
  xfs: merge fsync and O_SYNC handling
  xfs: speed up free inode search
  xfs: rationalize xfs_inobt_lookup*
  xfs: untangle xfs_dialloc
  xfs: factor out debug checks from xfs_dialloc and xfs_difree
  xfs: improve xfs_inobt_update prototype
  xfs: improve xfs_inobt_get_rec prototype
  xfs: factor out inode initialisation
  fs/xfs: Correct redundant test
  xfs: remove XFS_INO64_OFFSET
  un-static xfs_read_agf
  xfs: add more statics & drop some unused functions
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier...
Linus Torvalds [Thu, 17 Sep 2009 16:53:04 +0000 (09:53 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/vapier/blackfin

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/vapier/blackfin: (75 commits)
  Blackfin: update ftrace for latest toolchain
  Blackfin: fix elf_fpregset_t definition
  Blackfin: unify cache init functions
  Blackfin: swap clocksource ratings for gptimer/cycles
  Blackfin: update ftrace_push_return_trace() breakage
  Blackfin: update cm board resources
  Blackfin: cm-bf537u: split board from cm-bf537e
  Blackfin: bf538-ezkit: add SPI IRQ resources
  Blackfin: increase default async timings for parallel flashes
  Blackfin: add ICPLB coverage for async banks
  Blackfin: use KERN_ALERT in all kgdb_test output
  Blackfin: fix BF54x SPI CS resources
  Blackfin: fix typo in isram_write()
  Blackfin: bf537-stamp: add adp5588 gpio resources
  Blackfin: add some isram-driver self tests
  Blackfin: workaround anomaly 05000283
  Blackfin: fix spelling in a few comments
  Blackfin: use raw_smp_processor_id() in exception code
  Blackfin: remove useless duplicated assignment in gpio code
  Blackfin: Fix link errors with binutils 2.19 and GCC 4.3
  ...

14 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Linus Torvalds [Thu, 17 Sep 2009 16:52:43 +0000 (09:52 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/gerg/m68knommu

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu: (53 commits)
  m68knommu: Make PAGE_SIZE available to assembly files.
  m68knommu: fix ColdFire definition of CLOCK_TICK_RATE
  m68knommu: set multi-function pins for ethernet when enabled
  m68knommu: remove special interrupt handling code for ne2k support
  m68knommu: relax IO_SPACE_LIMIT setting
  m68knommu: remove ColdFire direct interrupt register access
  m68knommu: create a speciailized ColdFire 5272 interrupt controller
  m68knommu: add support for second interrupt controller of ColdFire 5249
  m68knommu: clean up old ColdFire timer irq setup
  m68knommu: map ColdFire interrupts to correct masking bits
  m68knommu: clean up ColdFire 532x CPU timer setup
  m68knommu: simplify ColdFire "timers" clock initialization
  m68knommu: support code to mask external interrupts on old ColdFire CPU's
  m68knommu: merge old ColdFire interrupt controller masking macros
  m68knommu: remove duplicate ColdFire mcf_autovector() code
  m68knommu: move ColdFire INTC definitions to new include file
  m68knommu: mask off all interrupts in ColdFire intc-simr controller
  m68knommu: remove timer device interrupt setup for ColdFire 532x
  m68knommu: remove interrupt masking from ColdFire pit timer
  m68knommu: remove unecessary interrupt level setting in ColdFire 520x setup
  ...

14 years agoMerge branch 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelv...
Linus Torvalds [Thu, 17 Sep 2009 16:47:35 +0000 (09:47 -0700)]
Merge branch 'hwmon-for-linus' of git://git./linux/kernel/git/jdelvare/staging

* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  hwmon: (lm85) Don't bind to Winbond/Nuvoton WPCD377I
  hwmon: (pcf8591) Documentation clean-ups
  hwmon: Clearly mark ACPI drivers as such
  hwmon: Use resource_size
  hwmon: Include <linux/io.h> instead of <asm/io.h>
  hwmon: (tmp421) Add documentation
  hwmon: Add driver for Texas Instruments TMP421/422/423 sensor chips
  hwmon-vid: Ignore 6th VID pin of AMD family 0Fh processors
  hwmon: (asus_atk0110) Add maintainer information
  hwmon: (abituguru3) Support multiple DMI strings per chip ID

14 years agoMerge branch 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/dvrabel/uwb
Linus Torvalds [Thu, 17 Sep 2009 16:44:09 +0000 (09:44 -0700)]
Merge branch 'for-upstream' of git://git./linux/kernel/git/dvrabel/uwb

* 'for-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/dvrabel/uwb:
  uwb: avoid radio controller reset loops
  uwb: stop uwbd thread if rc->start() fails
  uwb: handle radio controller events with out-of-range IDs correctly

14 years agoBlackfin: update ftrace for latest toolchain
Yi Li [Tue, 15 Sep 2009 09:24:31 +0000 (09:24 +0000)]
Blackfin: update ftrace for latest toolchain

The mcount support that was finally added to the Blackfin gcc port isn't
exactly the same as what ftrace was developed against.  Now that the final
gcc version is in place, update the ftrace code to match.

While updating this, fix the swapped arguments to the tracer (signature is
(ip, parent_ip) while we were passing (parent_ip, ip)).

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix elf_fpregset_t definition
Mike Frysinger [Mon, 14 Sep 2009 21:22:06 +0000 (21:22 +0000)]
Blackfin: fix elf_fpregset_t definition

The elf_fpregset_t type relied on an empty struct in the asm/user.h, but
the transition to asm-generic/user.h dropped that empty struct.  Rather
than restore this useless struct, define the only user (elf_fpregset_t)
as an empty struct itself.  This fixes building when ELF dump support is
enabled.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: unify cache init functions
Mike Frysinger [Mon, 14 Sep 2009 19:42:26 +0000 (19:42 +0000)]
Blackfin: unify cache init functions

The CPLB implementations (mpu/nompu) had exact copies of the cacheinit
code.  Even the i/d cache functions are largely the same.  So unify them
both in the common kernel cache code.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: swap clocksource ratings for gptimer/cycles
Graf Yang [Mon, 14 Sep 2009 04:41:00 +0000 (04:41 +0000)]
Blackfin: swap clocksource ratings for gptimer/cycles

The cycles clocksource is a higher resolution than the gptimer one, so
make sure the ratings field reflects this.

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: update ftrace_push_return_trace() breakage
Mike Frysinger [Mon, 14 Sep 2009 01:50:22 +0000 (01:50 +0000)]
Blackfin: update ftrace_push_return_trace() breakage

Commit 71e308a239c updated ftrace_push_return_trace() prototype but didn't
update the Blackfin ftrace code, so things broke.  Since we don't support
the new stuff yet, call it with stub values.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: update cm board resources
Harald Krapfenbauer [Thu, 10 Sep 2009 15:30:03 +0000 (15:30 +0000)]
Blackfin: update cm board resources

Signed-off-by: Harald Krapfenbauer <Harald.Krapfenbauer@bluetechnix.at>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: cm-bf537u: split board from cm-bf537e
Harald Krapfenbauer [Thu, 10 Sep 2009 15:12:08 +0000 (15:12 +0000)]
Blackfin: cm-bf537u: split board from cm-bf537e

The cm-bf537u module, while similar to the cm-bf537e, is different enough
to warrant its own resources.  It has a USB controller but no PHY.

Signed-off-by: Harald Krapfenbauer <Harald.Krapfenbauer@bluetechnix.at>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf538-ezkit: add SPI IRQ resources
Barry Song [Thu, 10 Sep 2009 04:32:47 +0000 (04:32 +0000)]
Blackfin: bf538-ezkit: add SPI IRQ resources

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: increase default async timings for parallel flashes
Sonic Zhang [Mon, 7 Sep 2009 03:20:48 +0000 (03:20 +0000)]
Blackfin: increase default async timings for parallel flashes

The default async timings are a little too fast for the parallel flash
that is attached by default to the async banks.  So slow things down a bit
so accessing the hardware is stable.

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: add ICPLB coverage for async banks
Bernd Schmidt [Wed, 2 Sep 2009 08:14:05 +0000 (08:14 +0000)]
Blackfin: add ICPLB coverage for async banks

When doing XIP, we need to execute out of the async banks, so we need
ICPLBs to allow this.

Signed-off-by: Bernd Schmidt <bernds_cb1@t-online.de>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: use KERN_ALERT in all kgdb_test output
Mingquan Pan [Mon, 31 Aug 2009 04:56:06 +0000 (04:56 +0000)]
Blackfin: use KERN_ALERT in all kgdb_test output

Most messages are already using KERN_ALERT, so be consistent to make
things easier to check with test scripts.

Signed-off-by: Mingquan Pan <grace.pan@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix BF54x SPI CS resources
Yi Li [Sun, 30 Aug 2009 20:45:50 +0000 (20:45 +0000)]
Blackfin: fix BF54x SPI CS resources

The BF54x has three slave select signals for SPI0/SPI1, not eight.

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix typo in isram_write()
Robin Getz [Thu, 27 Aug 2009 14:39:31 +0000 (14:39 +0000)]
Blackfin: fix typo in isram_write()

The DTEST write bit is 2, not 1.  Improve comments in the related macro
while we're here.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: add adp5588 gpio resources
Michael Hennerich [Thu, 27 Aug 2009 04:09:32 +0000 (04:09 +0000)]
Blackfin: bf537-stamp: add adp5588 gpio resources

For the adp5588 GPIO daughter card.

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: add some isram-driver self tests
Mike Frysinger [Thu, 27 Aug 2009 01:46:48 +0000 (01:46 +0000)]
Blackfin: add some isram-driver self tests

Make it easy to figure out if code changes here are correct.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: workaround anomaly 05000283
Robin Getz [Wed, 26 Aug 2009 15:54:10 +0000 (15:54 +0000)]
Blackfin: workaround anomaly 05000283

Make sure our interrupt entry code with exact hardware errors handles
anomaly 05000283 (infinite stall in system MMR kill) so we don't stall
while under load.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix spelling in a few comments
Michael Hennerich [Fri, 21 Aug 2009 03:49:19 +0000 (03:49 +0000)]
Blackfin: fix spelling in a few comments

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: use raw_smp_processor_id() in exception code
Yi Li [Thu, 20 Aug 2009 04:17:47 +0000 (04:17 +0000)]
Blackfin: use raw_smp_processor_id() in exception code

When preempt debugging is enabled, smp_processor_id() may utilize the
"current" structure.  This may not be safe to access under all exceptions
due to it being in dynamically allocated memory.  So in exception code,
make sure we use raw_smp_processor_id() instead to get at the real value
directly.

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: remove useless duplicated assignment in gpio code
Michael Hennerich [Wed, 19 Aug 2009 03:15:56 +0000 (03:15 +0000)]
Blackfin: remove useless duplicated assignment in gpio code

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: Fix link errors with binutils 2.19 and GCC 4.3
Ingo Molnar [Tue, 18 Aug 2009 16:45:25 +0000 (16:45 +0000)]
Blackfin: Fix link errors with binutils 2.19 and GCC 4.3

Not sure whether this has been reported/fixed before.

Today I built a Blackfin tool-chain from scratch for -tip testing,
and it triggers:

 arch/blackfin/kernel/vmlinux.lds:1238: undefined section `.data_a_l1' referenced in expression

and:

 arch/blackfin/kernel/vmlinux.lds:1238: undefined section `.text_data_l1'
referenced in expression

Now i dont have any way to test this linker script, but it now at
least builds fine after fixing what appears to be typos in those
assert statements.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: handle the core timer interrupt with handle_percpu_irq on SMP
Graf Yang [Tue, 18 Aug 2009 04:29:33 +0000 (04:29 +0000)]
Blackfin: handle the core timer interrupt with handle_percpu_irq on SMP

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: optimize fixed code handling for the most common case
Mike Frysinger [Mon, 17 Aug 2009 19:05:07 +0000 (19:05 +0000)]
Blackfin: optimize fixed code handling for the most common case

The majority of the time we are returning to user space, it is not in the
fixed atomic code region.  So rather than branch to a function where we
check the PC and return, do the check inline and branch only when needed.

Also, tweak some of the fixed code handling based on assumptions we are
aware of but cannot be expressed in C.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: delete '-spi' suffix in ad1836/ad1938 driver name
Barry Song [Thu, 13 Aug 2009 21:07:37 +0000 (21:07 +0000)]
Blackfin: delete '-spi' suffix in ad1836/ad1938 driver name

The ASoC drivers have dropped the redundant "-spi" suffix in the driver
name, so update the board resources accordingly.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: punt dead cache locking code
Mike Frysinger [Thu, 13 Aug 2009 23:32:11 +0000 (19:32 -0400)]
Blackfin: punt dead cache locking code

No one uses these functions, and some are duplicate of existing C code.  So
just punt the whole thing.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: drop board resources for dead devices (pbx/ad9960)
Mike Frysinger [Thu, 13 Aug 2009 04:27:31 +0000 (04:27 +0000)]
Blackfin: drop board resources for dead devices (pbx/ad9960)

These hardware devices are dead and the drivers never cleaned up/merged,
so punt the useless board resource info.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: do not trace the exception handler
Yi Li [Wed, 12 Aug 2009 23:05:35 +0000 (23:05 +0000)]
Blackfin: do not trace the exception handler

Since the exception handler cannot cause exceptions, we cannot trace it
without easily causing double faults and crashing the system.

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix BF522/BF523 max VCO mixup
Mike Frysinger [Tue, 11 Aug 2009 21:27:09 +0000 (21:27 +0000)]
Blackfin: fix BF522/BF523 max VCO mixup

The BF522 and BF523 had their real max VCO values swapped.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: SMP port does not yet support cpu frequency scaling
Graf Yang [Fri, 7 Aug 2009 03:52:54 +0000 (03:52 +0000)]
Blackfin: SMP port does not yet support cpu frequency scaling

We have an item to get this fixed, but in the mean time, disable selection
via Kconfig dependencies.

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: cleanup sync handling when enabling/disabling cplbs
Yi Li [Fri, 7 Aug 2009 01:20:58 +0000 (01:20 +0000)]
Blackfin: cleanup sync handling when enabling/disabling cplbs

The handling of updating the [DI]MEM_CONTROL MMRs does not follow proper
sync procedures as laid out in the Blackfin programming manual.  So rather
than audit/fix every call location, create helper functions that do the
right things in order to safely update these MMRs.  Then convert all call
sites to use these new helper functions.

While we're fixing the code, drop the workaround for anomaly 05000125 as
that anomaly applies to old versions of silicon that we do not support.

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: update ad1836 resources
Barry Song [Thu, 6 Aug 2009 21:03:02 +0000 (21:03 +0000)]
Blackfin: bf537-stamp: update ad1836 resources

Update the ad1836 resources for the new ASoC driver.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: update anomaly lists
Yi Li [Wed, 5 Aug 2009 10:02:14 +0000 (10:02 +0000)]
Blackfin: update anomaly lists

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf538-ezkit: add resources for parallel flash
Barry Song [Mon, 3 Aug 2009 04:40:36 +0000 (04:40 +0000)]
Blackfin: bf538-ezkit: add resources for parallel flash

The board has some parallel flash hooked up to the async banks, so add
appropriate physmap resources for it.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix read buffer overflow
Roel Kluin [Sun, 2 Aug 2009 12:26:48 +0000 (14:26 +0200)]
Blackfin: fix read buffer overflow

Check whether index is within bounds before testing the element.

Signed-off-by: Roel Kluin <roel.kluin@gmail.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf548-ezkit: fix incorrect LCD size parameters
Stefan Pledl [Wed, 29 Jul 2009 08:09:45 +0000 (08:09 +0000)]
Blackfin: bf548-ezkit: fix incorrect LCD size parameters

I think we have to use the physical dimensions [mm] of the display for
.width and .heigth in struct bfin_bf54xfb_mach_info bf54x_lq043_data which
are copied to fbinfo->var.height/.width in bf54x-lq043fb.c.

linux/fb.h describes this values as 'height/weight of picture in mm'

Otherwise QT calcs the wrong dpi value and the displayed fonts are very
small.

Signed-off-by: Stefan Pledl <stefan.pledl@mesutronic.de>
Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: cm-bf527/bf537-stamp: fix dm9000 resources
Barry Song [Mon, 27 Jul 2009 06:42:50 +0000 (06:42 +0000)]
Blackfin: cm-bf527/bf537-stamp: fix dm9000 resources

The dm9000 driver expects two IORESOURCE_MEM to get at the device, so make
sure we declare things properly.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: drop unused MMR defines that only cause bad code to be written
Mike Frysinger [Mon, 27 Jul 2009 00:44:25 +0000 (00:44 +0000)]
Blackfin: drop unused MMR defines that only cause bad code to be written

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: convert boards to use platform data with smc91x
Michael Hennerich [Fri, 24 Jul 2009 08:48:31 +0000 (08:48 +0000)]
Blackfin: convert boards to use platform data with smc91x

Latest smc91x driver allows you to specify settings in board resources
rather than needing CONFIG_BLACKFIN in the drivers/net/smc91x.h header.

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: drop ATA interrupt probe workaround
Barry Song [Thu, 23 Jul 2009 04:20:48 +0000 (04:20 +0000)]
Blackfin: bf537-stamp: drop ATA interrupt probe workaround

The interrupt probe workaround doesn't work without hacks to common code,
and the add-on card only needs a simple resistor to fix the problem, so
drop the board-specific hack.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: improve double fault debug handling
Graf Yang [Wed, 22 Jul 2009 11:56:24 +0000 (11:56 +0000)]
Blackfin: improve double fault debug handling

Since the hardware only provides reporting for the last exception handled,
and the values are valid only when executing the exception handler, we
need to save the context for reporting at a later point.  While we do this
for one exception, it doesn't work properly when handling a second one as
the original exception is clobbered by the double fault.  So when double
fault debugging is enabled, create a dedicated shadow of these values and
save/restore out of there.  Now the crash report properly displays the
first exception as well as the second one.

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf533-stamp: fix typo in SPI frequency for ad1836 codec
Cliff Cai [Wed, 22 Jul 2009 06:34:55 +0000 (06:34 +0000)]
Blackfin: bf533-stamp: fix typo in SPI frequency for ad1836 codec

One too many zeros means we run way faster than the codec can handle.

Signed-off-by: Cliff Cai <cliff.cai@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix MPU handling of invalid memory accesses
Graf Yang [Tue, 21 Jul 2009 02:26:57 +0000 (02:26 +0000)]
Blackfin: fix MPU handling of invalid memory accesses

The protect_page() function was incorrectly setting up the hardware tables
based on possible access capabilities rather than the actual requested
values.  This means we would grant more access to mmap-ed pages than we
should have.  Once we fix this, we need to tweak the signal generated by
such accesses to aline ourselves with other ports.  This allows the LTP
mmap0{5,6,7} cases to run properly.

Signed-off-by: Graf Yang <graf.yang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: clean up duplicate I2C device type definitions
Jean Delvare [Sat, 18 Jul 2009 14:54:03 +0000 (16:54 +0200)]
Blackfin: clean up duplicate I2C device type definitions

I2C_BOARD_INFO() already sets .type, no need to set it again.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: update AD714x resources
Barry Song [Fri, 17 Jul 2009 07:04:55 +0000 (07:04 +0000)]
Blackfin: bf537-stamp: update AD714x resources

Push the device table to the board resources as data interpretation can be
changed on a per-board basis.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: do not try displaying the end of the stack
Sonic Zhang [Thu, 16 Jul 2009 10:36:35 +0000 (10:36 +0000)]
Blackfin: do not try displaying the end of the stack

The end of the stack may not be valid (and that could be OK), so do not
attempt to parse it.  If we do, we might use a bad pointer in kernel space
which makes things panic().

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf533-stamp: add resources for mmc_spi card
Yi Li [Thu, 16 Jul 2009 10:12:30 +0000 (10:12 +0000)]
Blackfin: bf533-stamp: add resources for mmc_spi card

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf538: add bfin_clear_PPI_STATUS() helper
Barry Song [Mon, 13 Jul 2009 04:16:42 +0000 (04:16 +0000)]
Blackfin: bf538: add bfin_clear_PPI_STATUS() helper

Add the bf538 version of bfin_clear_PPI_STATUS() to match all other ports.

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix BF548 UART0 DMA IRQ translation
Stefan Pledl [Sat, 11 Jul 2009 11:50:42 +0000 (13:50 +0200)]
Blackfin: fix BF548 UART0 DMA IRQ translation

The initial BF54x port included some defines to keep code simple across
different processors, but it just ended up causing the UART0 DMA IRQs to
be set to the UART1 channels.

Signed-off-by: Stefan Pledl <stefan.pledl@mesutronic.de>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: catch hardware errors earlier during booting
Robin Getz [Tue, 7 Jul 2009 20:17:09 +0000 (20:17 +0000)]
Blackfin: catch hardware errors earlier during booting

Allow hardware errors to be caught during early portions of booting, and
leave something in the shadow console that people can use to debug their
system with (to be printed out by the bootloader on next reset).

This enables the hardare error interrupts in head.S, allowing us to find
hardware errors when they happen (well, as much as you can with a hardware
error) and prints out the trace if it is enabled.  This will catch errors
(like booting the wrong image on a 533) which previously resulted in a
infinite loop/hang, as well as random hardware errors before before
setup_arch().

To disable this debug only feature - turn off EARLY_PRINTK.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: add an early shadow console
Robin Getz [Mon, 6 Jul 2009 14:53:19 +0000 (14:53 +0000)]
Blackfin: add an early shadow console

Add a memory based shadow console to keep a copy of the printk buffer in a
location which can be found externally.  This allows bootloaders to locate
and utilize the log buffer in case of silent (early/resume/etc...) crashes.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: add support for common FDPIC ptrace requests
Mike Frysinger [Fri, 3 Jul 2009 00:17:45 +0000 (00:17 +0000)]
Blackfin: add support for common FDPIC ptrace requests

The FDPIC arches support a standard set of ptrace requests so rather than
define our own custom API, hook up those requests for common code to
leverage.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: push SRAM locks down into related ifdefs
Mike Frysinger [Wed, 1 Jul 2009 15:42:13 +0000 (15:42 +0000)]
Blackfin: push SRAM locks down into related ifdefs

Rather than defining the locks and initializing them all the time, only do
so when we actually need them (i.e. the SRAM regions exist).  This avoids
dead data and code bloat during runtime.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: unify duplicated SMP checks in L2 cache kconfig
Sonic Zhang [Tue, 30 Jun 2009 09:48:03 +0000 (09:48 +0000)]
Blackfin: unify duplicated SMP checks in L2 cache kconfig

Signed-off-by: Sonic Zhang <sonic.zhang@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: rename PCF8574 driver config
Michael Hennerich [Tue, 30 Jun 2009 14:57:22 +0000 (14:57 +0000)]
Blackfin: rename PCF8574 driver config

The "TWI_KEYPAD" driver was renamed to "INPUT_PCF8574", so update the
defines in the board resources accordingly.

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: clean up early memory setup code
Robin Getz [Fri, 26 Jun 2009 12:52:46 +0000 (12:52 +0000)]
Blackfin: clean up early memory setup code

Remove code duplication, and only print out memory warnings when they are
an actual problem.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix module reloc handling for all memory regions
Mike Frysinger [Fri, 26 Jun 2009 00:49:51 +0000 (00:49 +0000)]
Blackfin: fix module reloc handling for all memory regions

The current module relocation code has spotty handling wrt different
memory regions (like L1 instruction).  Rather than try to fix each
little spot, use the new common memory functions to greatly simplify
everything and make sure it is always correct.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: cleanup module section checking
Mike Frysinger [Fri, 26 Jun 2009 00:48:33 +0000 (00:48 +0000)]
Blackfin: cleanup module section checking

The current module section handling code has a lot of verbose statements
copied and pasted throughout which makes it pretty hard to digest at a
glance.  By unifying all of these up front, it is a lot easier to quickly
get an idea of what is actually going on.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: convert malloc()/memset() to zalloc() in module code
Mike Frysinger [Fri, 26 Jun 2009 00:37:40 +0000 (00:37 +0000)]
Blackfin: convert malloc()/memset() to zalloc() in module code

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: cleanup printk() usage in module code
Mike Frysinger [Fri, 26 Jun 2009 00:35:24 +0000 (00:35 +0000)]
Blackfin: cleanup printk() usage in module code

Convert all printk() statements to use the common pr_xxx() funcs and use
the new pr_fmt() function to standardize all of the output.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: reject outdated/unused/wrong relocation types
Robin Getz [Thu, 25 Jun 2009 15:49:38 +0000 (15:49 +0000)]
Blackfin: reject outdated/unused/wrong relocation types

All kernel modules are required to be built with -mlong-calls and thus
should not generate any of these relocations.  If they do, it means the
module has not been compiled properly, so rather than trying to handle
them (and running into random run time errors) just error out on module
load to force the module to be compiled correctly.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: convert ptrace to new memory functions
Mike Frysinger [Thu, 25 Jun 2009 00:02:58 +0000 (20:02 -0400)]
Blackfin: convert ptrace to new memory functions

Now that we have a Blackfin memory function to figure out how to properly
access the different regions, drop the custom memory range checks in our
ptrace code and use that.  It makes the code nicer and fixes bugs where
the ptrace logic wasn't handling all the different regions.

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: cleanup traps decode_address() a bit
Mike Frysinger [Tue, 23 Jun 2009 20:17:21 +0000 (20:17 +0000)]
Blackfin: cleanup traps decode_address() a bit

Unify the address display to shrink the code, and add missing decoding of
a few special Blackfin-specific regions (L1 ROM and MMRs).

Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: inline I-pipe bypass code in ret_from_exception
Philippe Gerum [Mon, 22 Jun 2009 16:26:22 +0000 (18:26 +0200)]
Blackfin: inline I-pipe bypass code in ret_from_exception

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: sanitize manual control of IPEND[4]
Philippe Gerum [Mon, 22 Jun 2009 16:26:07 +0000 (18:26 +0200)]
Blackfin: sanitize manual control of IPEND[4]

Cleanup is performed in two ways:

- remove extraneous updates of IPEND[4] w/ CONFIG_IPIPE,
  and document remaining use.

- substitute pop-reg-from-stack instructions with plain SP fixups in
  all save-RETI-then-discard patterns.

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: document __ipipe_call_irqtail
Philippe Gerum [Mon, 22 Jun 2009 16:25:52 +0000 (18:25 +0200)]
Blackfin: document __ipipe_call_irqtail

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: allow EVT5 to preempt irqtail prologue (CONFIG_DEBUG_HWERR)
Philippe Gerum [Mon, 22 Jun 2009 16:24:18 +0000 (18:24 +0200)]
Blackfin: allow EVT5 to preempt irqtail prologue (CONFIG_DEBUG_HWERR)

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: reuse evt_evt14 handler to perform irqtail epilogue
Philippe Gerum [Mon, 22 Jun 2009 16:24:02 +0000 (18:24 +0200)]
Blackfin: reuse evt_evt14 handler to perform irqtail epilogue

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: use generic name for EVT14 handler
Philippe Gerum [Mon, 22 Jun 2009 16:23:48 +0000 (18:23 +0200)]
Blackfin: use generic name for EVT14 handler

The purpose of the EVT14 handler may depend on whether CONFIG_IPIPE is
enabled, albeit its implementation can be the same in both cases. When
the interrupt pipeline is enabled, EVT14 can be used to raise the core
priority level for the running code; when CONFIG_IPIPE is off, EVT14
can be used to lower this level before running softirq handlers.

Rename evt14_softirq to evt_evt14 to pick an identifier that fits
both, which allows to reuse the same vector setup code as well.

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: allow high priority domains to preempt schedule_tail()
Philippe Gerum [Mon, 22 Jun 2009 16:23:32 +0000 (18:23 +0200)]
Blackfin: allow high priority domains to preempt schedule_tail()

ret_from_fork is always entered with hw interrupts off, which prevents
real-time domains to preempt the Linux kernel during part of the
initial context switch to the new task, which could in turn raise the
worst-case latency figures.

To avoid this, stall the root domain stage in the interrupt pipeline
to keep the scheduling tail code free from Linux-handled IRQs, then
enable hardware interrupts again.

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: export show_stack() to modules
Philippe Gerum [Mon, 22 Jun 2009 16:23:12 +0000 (18:23 +0200)]
Blackfin: export show_stack() to modules

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: fix misnomer of some I-pipe helpers
Philippe Gerum [Mon, 22 Jun 2009 16:22:48 +0000 (18:22 +0200)]
Blackfin: fix misnomer of some I-pipe helpers

__ipipe_{stall, unstall}_root_raw() identifiers may leave the reader
under the impression that only the virtual state is affected by these
operations, which is wrong. Pick names following the convention used
throughout the interrupt pipeline code.

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: checkpatch --file arch/blackfin/kernel/ipipe.c
Philippe Gerum [Mon, 22 Jun 2009 16:22:25 +0000 (18:22 +0200)]
Blackfin: checkpatch --file arch/blackfin/kernel/ipipe.c

Signed-off-by: Philippe Gerum <rpm@xenomai.org>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: make EVT3->EVT5 lowering more robust wrt IPEND[4]
Robin Getz [Mon, 22 Jun 2009 02:02:16 +0000 (02:02 +0000)]
Blackfin: make EVT3->EVT5 lowering more robust wrt IPEND[4]

We handle many exceptions at EVT5 (hardware error level) so that we can
catch exceptions in our exception handling code.  Today - if the global
interrupt enable bit (IPEND[4]) is set (interrupts disabled) our trap
handling code goes into a infinite loop, since we need interrupts to be
on to defer things to EVT5.

Normal kernel code should not trigger this for any reason as IPEND[4] gets
cleared early (when doing an interrupt context save) and the kernel stack
there should be sane (or something much worse is happening in the system).
But there have been a few times where this has happened, so this change
makes sure we dump a proper crash message even when things have gone south.

Signed-off-by: Robin Getz <robin.getz@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: add resources for AD1938 audio card
Barry Song [Thu, 4 Jun 2009 10:14:17 +0000 (10:14 +0000)]
Blackfin: bf537-stamp: add resources for AD1938 audio card

Signed-off-by: Barry Song <barry.song@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: declare SPI IRQ resources
Yi Li [Wed, 3 Jun 2009 09:46:22 +0000 (09:46 +0000)]
Blackfin: bf537-stamp: declare SPI IRQ resources

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoBlackfin: bf537-stamp: update ADP5588 header name
Michael Hennerich [Tue, 2 Dec 2008 14:08:58 +0000 (14:08 +0000)]
Blackfin: bf537-stamp: update ADP5588 header name

Signed-off-by: Michael Hennerich <michael.hennerich@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
14 years agoFix ia64 build breakage in head.S
Anirban Sinha [Wed, 16 Sep 2009 17:16:18 +0000 (10:16 -0700)]
Fix ia64 build breakage in head.S

The "cleanup console_print()" patch in commit
353f6dd2dec992ddd34620a94b051b0f76227379 introduced an "extern"
declaration into an assembly language file.  Remove it.

Signed-off-by: Anirban Sinha <asinha@zeugmasystems.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6
Linus Torvalds [Wed, 16 Sep 2009 15:27:10 +0000 (08:27 -0700)]
Merge git://git./linux/kernel/git/gregkh/driver-core-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/driver-core-2.6:
  Driver Core: devtmpfs - kernel-maintained tmpfs-based /dev
  debugfs: Modify default debugfs directory for debugging pktcdvd.
  debugfs: Modified default dir of debugfs for debugging UHCI.
  debugfs: Change debugfs directory of IWMC3200
  debugfs: Change debuhgfs directory of trace-events-sample.h
  debugfs: Fix mount directory of debugfs by default in events.txt
  hpilo: add poll f_op
  hpilo: add interrupt handler
  hpilo: staging for interrupt handling
  driver core: platform_device_add_data(): use kmemdup()
  Driver core: Add support for compatibility classes
  uio: add generic driver for PCI 2.3 devices
  driver-core: move dma-coherent.c from kernel to driver/base
  mem_class: fix bug
  mem_class: use minor as index instead of searching the array
  driver model: constify attribute groups
  UIO: remove 'default n' from Kconfig
  Driver core: Add accessor for device platform data
  Driver core: move dev_get/set_drvdata to drivers/base/dd.c
  Driver core: add new device to bus's list before probing

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6
Linus Torvalds [Wed, 16 Sep 2009 15:11:54 +0000 (08:11 -0700)]
Merge git://git./linux/kernel/git/gregkh/staging-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6: (641 commits)
  Staging: remove sxg driver
  Staging: remove heci driver
  Staging: remove at76_usb wireless driver.
  Staging: rspiusb: remove the driver
  Staging: meilhaus: remove the drivers
  Staging: remove me4000 driver.
  Staging: line6: ffzb returns an unsigned integer
  Staging: line6: pod.c: style cleanups
  Staging: iio: introduce missing kfree
  Staging: dream: introduce missing kfree
  Staging: comedi: addi-data: NULL dereference of amcc in v_pci_card_list_init()
  Staging: vt665x: fix built-in compiling
  Staging: rt3090: enable NATIVE_WPA_SUPPLICANT_SUPPORT option
  Staging: rt3090: port changes in WPA_MIX_PAIR_CIPHER to rt3090
  Staging: rt3090: rename device from raX to wlanX
  Staging: rt3090: remove possible conflict with rt2860
  Staging: rt2860/rt2870/rt3070/rt3090: fix compiler warning on x86_64
  Staging: rt2860: add new device ids
  Staging: rt3090: add device id 1462:891a
  Staging: asus_oled: Cleaned up checkpatch issues.
  ...

14 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/pcmcia-2.6
Linus Torvalds [Wed, 16 Sep 2009 15:11:23 +0000 (08:11 -0700)]
Merge git://git./linux/kernel/git/gregkh/pcmcia-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/pcmcia-2.6:
  pcmcia: document return value of pcmcia_loop_config
  pcmcia: dtl1_cs: fix pcmcia_loop_config logic
  pcmcia: drop non-existant includes
  pcmcia: disable prefetch/burst for OZ6933
  pcmcia: fix incorrect argument order to list_add_tail()
  pcmcia: drivers/pcmcia/pcmcia_resource.c: Remove unnecessary semicolons
  pcmcia: Use phys_addr_t for physical addresses
  pcmcia: drivers/pcmcia: Make static

14 years agoMerge branch 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Wed, 16 Sep 2009 14:49:54 +0000 (07:49 -0700)]
Merge branch 'linux-next' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'linux-next' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6: (75 commits)
  PCI hotplug: clean up acpi_run_hpp()
  PCI hotplug: acpiphp: use generic pci_configure_slot()
  PCI hotplug: shpchp: use generic pci_configure_slot()
  PCI hotplug: pciehp: use generic pci_configure_slot()
  PCI hotplug: add pci_configure_slot()
  PCI hotplug: clean up acpi_get_hp_params_from_firmware() interface
  PCI hotplug: acpiphp: don't cache hotplug_params in acpiphp_bridge
  PCI hotplug: acpiphp: remove superfluous _HPP/_HPX evaluation
  PCI: Clear saved_state after the state has been restored
  PCI PM: Return error codes from pci_pm_resume()
  PCI: use dev_printk in quirk messages
  PCI / PCIe portdrv: Fix pcie_portdrv_slot_reset()
  PCI Hotplug: convert acpi_pci_detect_ejectable() to take an acpi_handle
  PCI Hotplug: acpiphp: find bridges the easy way
  PCI: pcie portdrv: remove unused variable
  PCI / ACPI PM: Propagate wake-up enable for devices w/o ACPI support
  ACPI PM: Replace wakeup.prepared with reference counter
  PCI PM: Introduce device flag wakeup_prepared
  PCI / ACPI PM: Rework some debug messages
  PCI PM: Simplify PCI wake-up code
  ...

Fixed up conflict in arch/powerpc/kernel/pci_64.c due to OF device tree
scanning having been moved and merged for the 32- and 64-bit cases.  The
'needs_freset' initialization added in 6e19314cc ("PCI/powerpc: support
PCIe fundamental reset") is now in arch/powerpc/kernel/pci_of_scan.c.

14 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Wed, 16 Sep 2009 14:46:34 +0000 (07:46 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  block: fix linkage problem with blk_iopoll and !CONFIG_BLOCK

14 years agoMerge branch 'writeback' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Wed, 16 Sep 2009 14:45:38 +0000 (07:45 -0700)]
Merge branch 'writeback' of git://git.kernel.dk/linux-2.6-block

* 'writeback' of git://git.kernel.dk/linux-2.6-block:
  writeback: fix possible bdi writeback refcounting problem
  writeback: Fix bdi use after free in wb_work_complete()
  writeback: improve scalability of bdi writeback work queues
  writeback: remove smp_mb(), it's not needed with list_add_tail_rcu()
  writeback: use schedule_timeout_interruptible()
  writeback: add comments to bdi_work structure
  writeback: splice dirty inode entries to default bdi on bdi_destroy()
  writeback: separate starting of sync vs opportunistic writeback
  writeback: inline allocation failure handling in bdi_alloc_queue_work()
  writeback: use RCU to protect bdi_list
  writeback: only use bdi_writeback_all() for WB_SYNC_NONE writeout
  fs: Assign bdi in super_block
  writeback: make wb_writeback() take an argument structure
  writeback: merely wakeup flusher thread if work allocation fails for WB_SYNC_NONE
  writeback: get rid of wbc->for_writepages
  fs: remove bdev->bd_inode_backing_dev_info

14 years agowriteback: fix possible bdi writeback refcounting problem
Nick Piggin [Tue, 15 Sep 2009 19:37:55 +0000 (21:37 +0200)]
writeback: fix possible bdi writeback refcounting problem

wb_clear_pending AFAIKS should not be called after the item has been
put on the list, except by the worker threads. It could lead to the
situation where the refcount is decremented below 0 and cause lots of
problems.

Presumably the !wb_has_dirty_io case is not a common one, so it can
be discovered when the thread wakes up to check?

Also add a comment in bdi_work_clear.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: Fix bdi use after free in wb_work_complete()
Nick Piggin [Tue, 15 Sep 2009 19:34:51 +0000 (21:34 +0200)]
writeback: Fix bdi use after free in wb_work_complete()

By the time bdi_work_on_stack gets evaluated again in bdi_work_free, it
can already have been deallocated and used for something else in the
!on stack case, giving a false positive in this test and causing
corruption.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: improve scalability of bdi writeback work queues
Nick Piggin [Tue, 15 Sep 2009 19:34:12 +0000 (21:34 +0200)]
writeback: improve scalability of bdi writeback work queues

If you're going to do an atomic RMW on each list entry, there's not much
point in all the RCU complexities of the list walking. This is only going
to help the multi-thread case I guess, but it doesn't hurt to do now.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: remove smp_mb(), it's not needed with list_add_tail_rcu()
Nick Piggin [Tue, 15 Sep 2009 19:32:58 +0000 (21:32 +0200)]
writeback: remove smp_mb(), it's not needed with list_add_tail_rcu()

list_add_tail_rcu contains required barriers.

Signed-off-by: Nick Piggin <npiggin@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: use schedule_timeout_interruptible()
Jens Axboe [Tue, 15 Sep 2009 19:27:40 +0000 (21:27 +0200)]
writeback: use schedule_timeout_interruptible()

Gets rid of a manual set_current_state().

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: add comments to bdi_work structure
Jens Axboe [Tue, 15 Sep 2009 18:04:57 +0000 (20:04 +0200)]
writeback: add comments to bdi_work structure

And document its retriever, get_next_work_item().

Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: splice dirty inode entries to default bdi on bdi_destroy()
Jens Axboe [Mon, 14 Sep 2009 10:57:56 +0000 (12:57 +0200)]
writeback: splice dirty inode entries to default bdi on bdi_destroy()

We cannot safely ensure that the inodes are all gone at this point
in time, and we must not destroy this bdi with inodes having off it.
So just splice our entries to the default bdi since that one will
always persist.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: separate starting of sync vs opportunistic writeback
Jens Axboe [Wed, 16 Sep 2009 13:13:54 +0000 (15:13 +0200)]
writeback: separate starting of sync vs opportunistic writeback

bdi_start_writeback() is currently split into two paths, one for
WB_SYNC_NONE and one for WB_SYNC_ALL. Add bdi_sync_writeback()
for WB_SYNC_ALL writeback and let bdi_start_writeback() handle
only WB_SYNC_NONE.

Push down the writeback_control allocation and only accept the
parameters that make sense for each function. This cleans up
the API considerably.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: inline allocation failure handling in bdi_alloc_queue_work()
Jens Axboe [Sun, 13 Sep 2009 18:07:36 +0000 (20:07 +0200)]
writeback: inline allocation failure handling in bdi_alloc_queue_work()

This gets rid of work == NULL in bdi_queue_work() and puts the
OOM handling where it belongs.

Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: use RCU to protect bdi_list
Jens Axboe [Mon, 14 Sep 2009 11:12:40 +0000 (13:12 +0200)]
writeback: use RCU to protect bdi_list

Now that bdi_writeback_all() no longer handles integrity writeback,
it doesn't have to block anymore. This means that we can switch
bdi_list reader side protection to RCU.

Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
14 years agowriteback: only use bdi_writeback_all() for WB_SYNC_NONE writeout
Jens Axboe [Tue, 15 Sep 2009 07:53:35 +0000 (09:53 +0200)]
writeback: only use bdi_writeback_all() for WB_SYNC_NONE writeout

Data integrity writeback must use bdi_start_writeback() and ensure
that wbc->sb and wbc->bdi are set.

Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>