GitHub/moto-9609/android_kernel_motorola_exynos9610.git
16 years ago[ARM] mm: move validation of membanks to one place
Russell King [Tue, 30 Sep 2008 18:29:25 +0000 (19:29 +0100)]
[ARM] mm: move validation of membanks to one place

The newly introduced sanity_check_meminfo() function should be
used to collect all validation of the meminfo array, which we
have in bootmem_init().  Move it there.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5272/1: remove conditional compilation in show_pte()
Nicolas Pitre [Tue, 30 Sep 2008 15:10:11 +0000 (16:10 +0100)]
[ARM] 5272/1: remove conditional compilation in show_pte()

The PTRS_PER_PMD != 1 condition can be evaluated with C code and
optimized at compile time.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5271/1: get rid of pages_to_mb()
Nicolas Pitre [Tue, 30 Sep 2008 15:05:09 +0000 (16:05 +0100)]
[ARM] 5271/1: get rid of pages_to_mb()

There is no use of this in the whole tree.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5269/1: ARMv7: Use -march=armv7-a as compiler flag
Catalin Marinas [Mon, 29 Sep 2008 10:06:09 +0000 (11:06 +0100)]
[ARM] 5269/1: ARMv7: Use -march=armv7-a as compiler flag

The current -march=armv7a is not supported by mainline gcc.

Cc: Paul Brook <paul@codesourcery.com>
Cc: Wei Zhong <weizhong@broadcom.com>
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: fix some comments in dma-mapping.h
Russell King [Tue, 30 Sep 2008 10:30:24 +0000 (11:30 +0100)]
[ARM] dma: fix some comments in dma-mapping.h

... to prevent people being mislead.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: don't touch cache on dma_*_for_cpu()
Russell King [Mon, 29 Sep 2008 18:50:59 +0000 (19:50 +0100)]
[ARM] dma: don't touch cache on dma_*_for_cpu()

As per the dma_unmap_* calls, we don't touch the cache when a DMA
buffer transitions from device to CPU ownership.  Presently, no
problems have been identified with speculative cache prefetching
which in itself is a new feature in later architectures.  We may
have to revisit the DMA API later for these architectures anyway.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: add validation of DMA params
Russell King [Mon, 29 Sep 2008 12:48:17 +0000 (13:48 +0100)]
[ARM] dma: add validation of DMA params

Validate the direction argument like x86 does.  In addition,
validate the dma_unmap_* parameters against those passed to
dma_map_* when using the DMA bounce code.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: coding style cleanups
Russell King [Thu, 25 Sep 2008 21:23:31 +0000 (22:23 +0100)]
[ARM] dma: coding style cleanups

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: fix dmabounce dma_sync_xxx() implementations
Russell King [Thu, 25 Sep 2008 21:16:22 +0000 (22:16 +0100)]
[ARM] dma: fix dmabounce dma_sync_xxx() implementations

The dmabounce dma_sync_xxx() implementation have been broken for
quite some time; they all copy data between the DMA buffer and
the CPU visible buffer no irrespective of the change of ownership.
(IOW, a DMA_FROM_DEVICE mapping copies data from the DMA buffer
to the CPU buffer during a call to dma_sync_single_for_device().)

Fix it by getting rid of sync_single(), moving the contents into
the recently created dmabounce_sync_for_xxx() functions and adjusting
appropriately.

This also makes it possible to properly support the DMA range sync
functions.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: use new dmabounce_sync_for_xxx() for dma_sync_single_xxx()
Russell King [Thu, 25 Sep 2008 20:52:49 +0000 (21:52 +0100)]
[ARM] dma: use new dmabounce_sync_for_xxx() for dma_sync_single_xxx()

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: Reduce to one dma_sync_sg_* implementation
Russell King [Thu, 25 Sep 2008 20:38:41 +0000 (21:38 +0100)]
[ARM] dma: Reduce to one dma_sync_sg_* implementation

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: Reduce to one dma_map_sg()/dma_unmap_sg() implementation
Russell King [Thu, 25 Sep 2008 20:05:02 +0000 (21:05 +0100)]
[ARM] dma: Reduce to one dma_map_sg()/dma_unmap_sg() implementation

No point having two of these; dma_map_page() can do all the work
for us.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: provide a better dma_map_page() implementation
Russell King [Thu, 25 Sep 2008 19:59:12 +0000 (20:59 +0100)]
[ARM] dma: provide a better dma_map_page() implementation

We can translate a struct page directly to a DMA address using
page_to_dma().  No need to use page_address() followed by
virt_to_dma().

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Update dma_map_sg()/dma_unmap_sg() API
Russell King [Thu, 25 Sep 2008 15:30:57 +0000 (16:30 +0100)]
[ARM] Update dma_map_sg()/dma_unmap_sg() API

Update the ARM DMA scatter gather APIs for the scatterlist changes.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] dma: rename consistent.c to dma-mapping.c
Russell King [Thu, 25 Sep 2008 14:59:19 +0000 (15:59 +0100)]
[ARM] dma: rename consistent.c to dma-mapping.c

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Print details relevant to how we handle the cache
Russell King [Thu, 25 Sep 2008 14:39:20 +0000 (15:39 +0100)]
[ARM] Print details relevant to how we handle the cache

This replaces the original cache type decoding printks.  We now
indicate how we're treating the cache which we found, rather
than what we found.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Introduce new bitmask based cache type macros
Russell King [Thu, 25 Sep 2008 14:35:28 +0000 (15:35 +0100)]
[ARM] Introduce new bitmask based cache type macros

Rather than trying to (inaccurately) decode the cache type from the
registers each time we need to decide what type of cache we have,
use a bitmask initialized early during boot.

Since the setup is a one-off initialization, we can be a little more
clever and take account of the CPU architecture as well.

Note that we continue to achieve the compactness on optimised kernels
by forcing tests to always-false or always-true as appropriate, thereby
allowing the compiler to do build-time code elimination.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Remove cache type printks
Russell King [Thu, 25 Sep 2008 13:45:02 +0000 (14:45 +0100)]
[ARM] Remove cache type printks

The cache type register found in ARMv5 and later CPUs changes format
and meaning depending on the CPU architecture version.  Currently,
this code:
a) doesn't work for everything - Xscale's are identified as
   'unknown 5'.
b) is not able to tell whether the caches are VIVT or VIPT from the
   cache type.
c) prints rubbish on some ARMv6 and ARMv7+ CPUs.

The two solutions to this are:
1. Add yet more code to decode and print the various different register
   formats.
2. Remove the code altogther.

The code only exists to decode and print the cache parameters.
Increasing the complexity of it just for the sake of a few prinks
isn't worth it.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Fix IOP13xx build warnings
Russell King [Wed, 17 Sep 2008 19:36:49 +0000 (20:36 +0100)]
[ARM] Fix IOP13xx build warnings

http://armlinux.simtec.co.uk/kautobuild/2.6.27-rc5/iop13xx_defconfig/zimage.log

Occurrences  Warning text
339  arch/arm/include/asm/dma-mapping.h:40: warning: return makes pointer from integer without a cast
203  arch/arm/include/asm/dma-mapping.h:45: warning: return makes integer from pointer without a cast

Acked-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Add -march=all to assembly file build in arch/arm/boot/compressed
Russell King [Tue, 9 Sep 2008 12:56:45 +0000 (13:56 +0100)]
[ARM] Add -march=all to assembly file build in arch/arm/boot/compressed

This allows assembly files to be crafted to cover all ARM CPU types
rather than erroring out on instructions only in later CPUs.  We
are careful in these files to only execute CPU specific code when
the CPU ID says we can.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Convert asm/bitops.h to linux/bitops.h
Russell King [Sat, 6 Sep 2008 11:13:59 +0000 (12:13 +0100)]
[ARM] Convert asm/bitops.h to linux/bitops.h

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Convert asm/delay.h to linux/delay.h
Russell King [Sat, 6 Sep 2008 11:11:37 +0000 (12:11 +0100)]
[ARM] Convert asm/delay.h to linux/delay.h

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Convert asm/io.h to linux/io.h
Russell King [Sat, 6 Sep 2008 11:10:45 +0000 (12:10 +0100)]
[ARM] Convert asm/io.h to linux/io.h

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Convert asm/uaccess.h to linux/uaccess.h
Russell King [Sat, 6 Sep 2008 10:35:55 +0000 (11:35 +0100)]
[ARM] Convert asm/uaccess.h to linux/uaccess.h

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] clean up a load of old declarations
Russell King [Sat, 6 Sep 2008 10:23:30 +0000 (11:23 +0100)]
[ARM] clean up a load of old declarations

... some of which are now in linux/*.h headers.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] move initrd code from kernel/setup.c to mm/init.c
Russell King [Sat, 6 Sep 2008 09:57:03 +0000 (10:57 +0100)]
[ARM] move initrd code from kernel/setup.c to mm/init.c

This quietens some sparse warnings about phys_initrd_start and
phys_initrd_size.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] sparse: quieten arch/arm/kernel/irq.c
Russell King [Sat, 6 Sep 2008 09:56:27 +0000 (10:56 +0100)]
[ARM] sparse: quieten arch/arm/kernel/irq.c

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] remove pc_pointer()
Russell King [Sat, 6 Sep 2008 09:14:24 +0000 (10:14 +0100)]
[ARM] remove pc_pointer()

pc_pointer() was a function to mask the PC for 26-bit ARMs, which
we no longer support.  Remove it.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] sparse: fix several warnings
Russell King [Fri, 5 Sep 2008 13:08:44 +0000 (14:08 +0100)]
[ARM] sparse: fix several warnings

arch/arm/kernel/process.c:270:6: warning: symbol 'show_fpregs' was not declared. Should it be static?

This function isn't used, so can be removed.

arch/arm/kernel/setup.c:532:9: warning: symbol 'len' shadows an earlier one
arch/arm/kernel/setup.c:524:6: originally declared here

A function containing two 'len's.

arch/arm/mm/fault-armv.c:188:13: warning: symbol 'check_writebuffer_bugs' was not declared. Should it be static?
arch/arm/mm/mmap.c:122:5: warning: symbol 'valid_phys_addr_range' was not declared. Should it be static?
arch/arm/mm/mmap.c:137:5: warning: symbol 'valid_mmap_phys_addr_range' was not declared. Should it be static?

Missing includes.

arch/arm/kernel/traps.c:71:77: warning: Using plain integer as NULL pointer
arch/arm/mm/ioremap.c:355:46: error: incompatible types in comparison expression (different address spaces)

Sillies.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5231/1: Do not save the frame pointer in the csum_partial_copy_* functions
Catalin Marinas [Fri, 29 Aug 2008 17:31:27 +0000 (18:31 +0100)]
[ARM] 5231/1: Do not save the frame pointer in the csum_partial_copy_* functions

Since the other assembly functions do not seem to save the frame
pointer onto the stack, this patch changes the csum_partial_copy_*
functions to behave in the same way.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5230/1: Replace post-indexed LDRT/STRT in uaccess.h
Catalin Marinas [Fri, 29 Aug 2008 17:31:15 +0000 (18:31 +0100)]
[ARM] 5230/1: Replace post-indexed LDRT/STRT in uaccess.h

The post-index immediate value is optional if it is 0 and this patch
removes it. The reason is to allow such instructions to compile to
Thumb-2 where only pre-indexed LDRT/STRT instructions are allowed.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5232/1: Do not post-index STRT instruction in clear_user.S
Catalin Marinas [Fri, 29 Aug 2008 17:31:21 +0000 (18:31 +0100)]
[ARM] 5232/1: Do not post-index STRT instruction in clear_user.S

The last strnebt instruction has a post-index of 1 but the address
register is set to 0 in the next instruction, so no need for
post-indexing.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5227/1: Add the ENDPROC declarations to the .S files
Catalin Marinas [Thu, 28 Aug 2008 10:22:32 +0000 (11:22 +0100)]
[ARM] 5227/1: Add the ENDPROC declarations to the .S files

This declaration specifies the "function" type and size for various
assembly functions, mainly needed for generating the correct branch
instructions in Thumb-2.

Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5222/1: Allow configuring user:kernel split via Kconfig
Lennert Buytenhek [Mon, 25 Aug 2008 20:03:32 +0000 (21:03 +0100)]
[ARM] 5222/1: Allow configuring user:kernel split via Kconfig

This patch adds a config option (CONFIG_VMSPLIT_*) to allow choosing
between 3:1, 2:2 and 1:3 user:kernel memory splits.

Tested-by: Riku Voipio <riku.voipio@iki.fi>
Signed-off-by: Lennert Buytenhek <buytenh@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5221/1: fix ldm/stm emulation for kprobes
Nicolas Pitre [Thu, 21 Aug 2008 22:22:49 +0000 (23:22 +0100)]
[ARM] 5221/1: fix ldm/stm emulation for kprobes

Logic for the p bit was reversed.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5211/2: fix a couple warnings from BUG() usage
Nicolas Pitre [Thu, 21 Aug 2008 22:09:48 +0000 (23:09 +0100)]
[ARM] 5211/2: fix a couple warnings from BUG() usage

When CONFIG_DEBUG_BUGVERBOSE is not set, we get warnings such as:

arch/arm/mm/ioremap.c: In function â€˜remap_area_pte’:
arch/arm/mm/ioremap.c:67: warning: control reaches end of non-void function
mm/bootmem.c: In function â€˜mark_bootmem’:
mm/bootmem.c:321: warning: control reaches end of non-void function
fs/dcache.c: In function â€˜d_materialise_unique’:
fs/dcache.c:1875: warning: control reaches end of non-void function
fs/nfs/client.c: In function â€˜nfs_sockaddr_match_ipaddr’:
fs/nfs/client.c:251: warning: control reaches end of non-void function
block/cfq-iosched.c: In function â€˜cfq_async_queue_prio’:
block/cfq-iosched.c:1501: warning: control reaches end of non-void function

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5218/1: arm: improved futex support
Mikael Pettersson [Wed, 20 Aug 2008 08:36:07 +0000 (09:36 +0100)]
[ARM] 5218/1: arm: improved futex support

Linux/ARM currently doesn't support robust or PI futexes.
The problem is that the kernel wants to perform certain ops
(cmpxchg, set, add, or, andn, xor) atomically on user-space
addresses, and ARM's futex.h doesn't support that.

This patch adds that support, but only for uniprocessor machines.
For UP it's enough to disable preemption to ensure mutual exclusion
with other software agents (futexes don't need to care about other
hardware agents, fortunately).

This patch is based on one posted by Khem Raj on 2007-08-01
<http://marc.info/?l=linux-arm-kernel&m=118599407413016&w=2>.
(That patch is included in the -RT kernel patches.)
My changes since that version include:
* corrected implementation of FUTEX_OP_ANDN (must complement oparg)
* added missing memory clobber to futex_atomic_cmpxchg_inatomic()
* removed spinlock because it's unnecessary for UP and insufficient
  for SMP, instead the code is restricted to UP and relies on the
  fact that pagefault_disable() also disables preemption
* coding style cleanups

Tested on ARMv5 XScales with the glibc-2.6 nptl test suite.

Tested-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5206/1: remove kprobe_trap_handler() hack
Nicolas Pitre [Tue, 19 Aug 2008 03:15:23 +0000 (04:15 +0100)]
[ARM] 5206/1: remove kprobe_trap_handler() hack

As mentioned in commit 796969104cab0d454dbc792ad0d12a4f365a8564,
and because of commit b03a5b7559563dafdbe52f8b5d8e453a914db941,
the direct calling of kprobe_trap_handler() can be removed.

Signed-off-by: Nicolas Pitre <nico@marvell.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] Update arch/arm/Kconfig for drivers/Kconfig changes, add cpuidle
Russell King [Mon, 18 Aug 2008 16:26:00 +0000 (17:26 +0100)]
[ARM] Update arch/arm/Kconfig for drivers/Kconfig changes, add cpuidle

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] 5195/1: ARMv7 Oprofile support
Jean PIHET [Tue, 12 Aug 2008 18:07:39 +0000 (19:07 +0100)]
[ARM] 5195/1: ARMv7 Oprofile support

Add Oprofile kernel support for ARMv7.
Tested on OMAP3430 and OMAP3530 chipsets (Cortex-A8).

Signed-off-by: Jean Pihet <jpihet@mvista.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] cachetype: move definitions to separate header
Russell King [Sun, 10 Aug 2008 17:10:19 +0000 (18:10 +0100)]
[ARM] cachetype: move definitions to separate header

Rather than pollute asm/cacheflush.h with the cache type definitions,
move them to asm/cachetype.h, and include this new header where
necessary.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years ago[ARM] cputype: separate definitions, use them
Russell King [Sun, 10 Aug 2008 17:08:10 +0000 (18:08 +0100)]
[ARM] cputype: separate definitions, use them

Add asm/cputype.h, moving functions and definitions from asm/system.h
there.  Convert all users of 'processor_id' to the more efficient
read_cpuid_id() function.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 26 Aug 2008 00:48:07 +0000 (17:48 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  bnx2x: Version update
  bnx2x: Multi Queue
  bnx2x: NAPI and interrupts enable/disable
  bnx2x: NIC load failure cleanup
  bnx2x: Initialization structure
  bnx2x: HW lock timeout
  bnx2x: Minimize lock time
  bnx2x: Fan failure mechanism on additional design
  bnx2x: Rx work check
  ipv6: sysctl fixes
  ipv4: sysctl fixes
  sctp: add verification checks to SCTP_AUTH_KEY option

16 years agoALSA: fix for CA0106 on MSI K8N Diamond PLUS Motherboard
Stephen Rothwell [Tue, 26 Aug 2008 00:33:32 +0000 (10:33 +1000)]
ALSA: fix for CA0106 on MSI K8N Diamond PLUS Motherboard

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agobnx2x: Version update
Eilon Greenstein [Mon, 25 Aug 2008 22:30:39 +0000 (15:30 -0700)]
bnx2x: Version update

Version update

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Multi Queue
Yitchak Gertner [Mon, 25 Aug 2008 22:27:06 +0000 (15:27 -0700)]
bnx2x: Multi Queue

The multi queue support is still disabled by default for the bnx2x
(needs some more testing and validation), but there are 2 obvious bug in
it which are fixed in this patch

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: NAPI and interrupts enable/disable
Yitchak Gertner [Mon, 25 Aug 2008 22:26:24 +0000 (15:26 -0700)]
bnx2x: NAPI and interrupts enable/disable

Fixing the order of enabling and disabling NAPI and the interrupts

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: NIC load failure cleanup
Yitchak Gertner [Mon, 25 Aug 2008 22:25:45 +0000 (15:25 -0700)]
bnx2x: NIC load failure cleanup

Load failures were not handled correctly

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Initialization structure
Yitchak Gertner [Mon, 25 Aug 2008 22:24:21 +0000 (15:24 -0700)]
bnx2x: Initialization structure

The TPA initialization is part of the FW internal memory initialization
and so it is moved to the appropriate function

Signed-off-by: Yitchak Gertner <gertner@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: HW lock timeout
Eilon Greenstein [Mon, 25 Aug 2008 22:23:30 +0000 (15:23 -0700)]
bnx2x: HW lock timeout

Increasing the lock timeout to 5 seconds instead of 1 second to minimize
the chance of failures due to timeout

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Minimize lock time
Eilon Greenstein [Mon, 25 Aug 2008 22:22:46 +0000 (15:22 -0700)]
bnx2x: Minimize lock time

After iSCSI boot, the HW lock should only protect the flag so only the
first function will reset the chip and not then entire chip reset
process

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Fan failure mechanism on additional design
Eilon Greenstein [Mon, 25 Aug 2008 22:20:48 +0000 (15:20 -0700)]
bnx2x: Fan failure mechanism on additional design

The A1021G board is also using the fan failure mechanism in the same way
the A1022G board does

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agobnx2x: Rx work check
Eilon Greenstein [Mon, 25 Aug 2008 22:19:17 +0000 (15:19 -0700)]
bnx2x: Rx work check

The has Rx work check was wrong: when the FW was at the end of the page,
the driver was already at the beginning of the next page. Since the
check only validated that both driver and FW are pointing to the same
place, it concluded that there is still work to be done. This caused
some serious issues including long latency results on ping-pong test and
lockups while unloading the driver in that condition.

Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoipv6: sysctl fixes
Al Viro [Mon, 25 Aug 2008 22:18:15 +0000 (15:18 -0700)]
ipv6: sysctl fixes

Braino: net.ipv6 in ipv6 skeleton has no business in rotable
class

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoipv4: sysctl fixes
Al Viro [Mon, 25 Aug 2008 22:17:44 +0000 (15:17 -0700)]
ipv4: sysctl fixes

net.ipv4.neigh should be a part of skeleton to avoid ordering problems

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agosctp: add verification checks to SCTP_AUTH_KEY option
Vlad Yasevich [Mon, 25 Aug 2008 22:16:19 +0000 (15:16 -0700)]
sctp: add verification checks to SCTP_AUTH_KEY option

The structure used for SCTP_AUTH_KEY option contains a
length that needs to be verfied to prevent buffer overflow
conditions.  Spoted by Eugene Teo <eteo@redhat.com>.

Signed-off-by: Vlad Yasevich <vladislav.yasevich@hp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years ago[x86] Clean up MAXSMP Kconfig, and limit NR_CPUS to 512
Linus Torvalds [Mon, 25 Aug 2008 21:15:38 +0000 (14:15 -0700)]
[x86] Clean up MAXSMP Kconfig, and limit NR_CPUS to 512

This fixes a regression that was indirectly caused by commit
1184dc2ffe2c8fb9afb766d870850f2c3165ef25 ("x86: modify Kconfig to allow
up to 4096 cpus").

Allowing 4k CPU's is not practical at this time, because we still have a
number of places that have several 'cpumask_t's on the stack, and a
4k-bit cpumask is 512 bytes of stack-space for each such variable.  This
literally caused functions like 'smp_call_function_mask' to have a 2.5kB
stack frame, and several functions to have 2kB stackframes.

With an 8kB stack total, smashing the stack was simply much too likely.
At least bugzilla entry

http://bugzilla.kernel.org/show_bug.cgi?id=11342

was due to this.

The earlier commit to not inline load_module() into sys_init_module()
fixed the particular symptoms of this that Alan Brunelle saw in that
bugzilla entry, but the huge stack waste by cpumask_t's was the more
direct cause.

Some day we'll have allocation helpers that allocate large CPU masks
dynamically, but in the meantime we simply cannot allow cpumasks this
large.

Cc: Alan D. Brunelle <Alan.Brunelle@hp.com>
Cc: Mike Travis <travis@sgi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge branch 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 25 Aug 2008 18:26:56 +0000 (11:26 -0700)]
Merge branch 'timers-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  rtc: fix deadlock

16 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 25 Aug 2008 18:26:33 +0000 (11:26 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: add X86_FEATURE_XMM4_2 definitions
  x86: fix cpufreq + sched_clock() regression
  x86: fix HPET regression in 2.6.26 versus 2.6.25, check hpet against BAR, v3
  x86: do not enable TSC notifier if we don't need it
  x86 MCE: Fix CPU hotplug problem with multiple multicore AMD CPUs
  x86: fix: make PCI ECS for AMD CPUs hotplug capable
  x86: fix: do not run code in amd_bus.c on non-AMD CPUs

16 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Mon, 25 Aug 2008 18:26:02 +0000 (11:26 -0700)]
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched_clock: fix cpu_clock()

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Mon, 25 Aug 2008 18:25:10 +0000 (11:25 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jbarnes/pci-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes/pci-2.6:
  PCI: fix reference leak in pci_get_dev_by_id()
  PCI: shpchp: Rename duplicate slot name N as N-1, N-2, N-M...
  PCI: pciehp: Rename duplicate slot name N as N-1, N-2, N-M...

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Mon, 25 Aug 2008 18:24:51 +0000 (11:24 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: ASoC: Fix double free and memory leak in many codec drivers
  ALSA: CA0106 on MSI K8N Diamond PLUS Motherboard

16 years agoMerge branch 'kvm-updates-2.6.27' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 25 Aug 2008 18:19:53 +0000 (11:19 -0700)]
Merge branch 'kvm-updates-2.6.27' of git://git./linux/kernel/git/avi/kvm

* 'kvm-updates-2.6.27' of git://git.kernel.org/pub/scm/linux/kernel/git/avi/kvm:
  KVM: fix userspace ABI breakage
  KVM: MMU: Fix torn shadow pte
  KVM: Use .fixup instead of .text.fixup on __kvm_handle_fault_on_reboot

16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus
Linus Torvalds [Mon, 25 Aug 2008 18:18:49 +0000 (11:18 -0700)]
Merge git://git./linux/kernel/git/rusty/linux-2.6-for-linus

* git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux-2.6-for-linus:
  lguest: update commentry
  stop_machine: Remove deprecated stop_machine_run
  stop_machine: wean Xen off stop_machine_run
  virtio_balloon: fix towards_target when deflating balloon

16 years ago[module] Don't let gcc inline load_module()
Linus Torvalds [Mon, 25 Aug 2008 18:10:26 +0000 (11:10 -0700)]
[module] Don't let gcc inline load_module()

'load_module()' is a complex function that contains all the ELF section
logic, and inlining it is utterly insane.  But gcc will do it, simply
because there is only one call-site.  As a result, all the stack space
that is allocated for all the work to load the module will still be
active when we actually call the module init sequence, and the deep call
chain makes stack overflows happen.

And stack overflows are really hard to debug, because they not only
corrupt random pages below the stack, but also corrupt the thread_info
structure that is allocated under the stack.

In this case, Alan Brunelle reported some crazy oopses at bootup, after
loading the processor module that ends up doing complex ACPI stuff and
has quite a deep callchain.  This should fix it, and is the sane thing
to do regardless.

Cc: Alan D. Brunelle <Alan.Brunelle@hp.com>
Cc: Arjan van de Ven <arjan@linux.intel.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agosched_clock: fix cpu_clock()
Peter Zijlstra [Mon, 25 Aug 2008 15:15:34 +0000 (17:15 +0200)]
sched_clock: fix cpu_clock()

This patch fixes 3 issues:

a) it removes the dependency on jiffies, because jiffies are incremented
   by a single CPU, and the tick is not synchronized between CPUs. Therefore
   relying on it to calculate a window to clip whacky TSC values doesn't work
   as it can drift around.

   So instead use [GTOD, GTOD+TICK_NSEC) as the window.

b) __update_sched_clock() did (roughly speaking):

   delta = sched_clock() - scd->tick_raw;
   clock += delta;

   Which gives exponential growth, instead of linear.

c) allows the sched_clock_cpu() value to warp the u64 without breaking.

the results are more reliable sched_clock() deltas:

           before       after   sched_clock

cpu_clock: 15750        51312   51488
cpu_clock: 59719        51052   50947
cpu_clock: 15879        51249   51061
cpu_clock: 1            50933   51198
cpu_clock: 1            50931   51039
cpu_clock: 1            51093   50981
cpu_clock: 1            51043   51040
cpu_clock: 1            50959   50938
cpu_clock: 1            50981   51011
cpu_clock: 1            51364   51212
cpu_clock: 1            51219   51273
cpu_clock: 1            51389   51048
cpu_clock: 1            51285   51611
cpu_clock: 1            50964   51137
cpu_clock: 1            50973   50968
cpu_clock: 1            50967   50972
cpu_clock: 1            58910   58485
cpu_clock: 1            51082   51025
cpu_clock: 1            50957   50958
cpu_clock: 1            50958   50957
cpu_clock: 1006128      51128   50971
cpu_clock: 1            51107   51155
cpu_clock: 1            51371   51081
cpu_clock: 1            51104   51365
cpu_clock: 1            51363   51309
cpu_clock: 1            51107   51160
cpu_clock: 1            51139   51100
cpu_clock: 1            51216   51136
cpu_clock: 1            51207   51215
cpu_clock: 1            51087   51263
cpu_clock: 1            51249   51177
cpu_clock: 1            51519   51412
cpu_clock: 1            51416   51255
cpu_clock: 1            51591   51594
cpu_clock: 1            50966   51374
cpu_clock: 1            50966   50966
cpu_clock: 1            51291   50948
cpu_clock: 1            50973   50867
cpu_clock: 1            50970   50970
cpu_clock: 998306       50970   50971
cpu_clock: 1            50971   50970
cpu_clock: 1            50970   50970
cpu_clock: 1            50971   50971
cpu_clock: 1            50970   50970
cpu_clock: 1            51351   50970
cpu_clock: 1            50970   51352
cpu_clock: 1            50971   50970
cpu_clock: 1            50970   50970
cpu_clock: 1            51321   50971
cpu_clock: 1            50974   51324

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: add X86_FEATURE_XMM4_2 definitions
Austin Zhang [Mon, 25 Aug 2008 15:14:51 +0000 (11:14 -0400)]
x86: add X86_FEATURE_XMM4_2 definitions

Added Intel processor SSE4.2 feature flag.

No in-tree user at the moment, but makes the tree-merging life easier
for the crypto tree.

Signed-off-by: Austin Zhang <austin.zhang@intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoKVM: fix userspace ABI breakage
Adrian Bunk [Tue, 5 Aug 2008 22:36:09 +0000 (01:36 +0300)]
KVM: fix userspace ABI breakage

The following part of commit 9ef621d3be56e1188300476a8102ff54f7b6793f
(KVM: Support mixed endian machines) changed on the size of a struct
that is exported to userspace:

include/linux/kvm.h:

@@ -318,14 +318,14 @@ struct kvm_trace_rec {
  __u32 vcpu_id;
  union {
  struct {
- __u32 cycle_lo, cycle_hi;
+ __u64 cycle_u64;
  __u32 extra_u32[KVM_TRC_EXTRA_MAX];
  } cycle;
  struct {
  __u32 extra_u32[KVM_TRC_EXTRA_MAX];
  } nocycle;
  } u;
-};
+} __attribute__((packed));

Packing a struct was the correct idea, but it packed the wrong struct.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: MMU: Fix torn shadow pte
Avi Kivity [Fri, 22 Aug 2008 16:19:38 +0000 (19:19 +0300)]
KVM: MMU: Fix torn shadow pte

The shadow code assigns a pte directly in one place, which is nonatomic on
i386 can can cause random memory references.  Fix by using an atomic setter.

Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agoKVM: Use .fixup instead of .text.fixup on __kvm_handle_fault_on_reboot
Eduardo Habkost [Tue, 19 Aug 2008 23:00:08 +0000 (20:00 -0300)]
KVM: Use .fixup instead of .text.fixup on __kvm_handle_fault_on_reboot

vmlinux.lds expects the fixup code to be on a section named .fixup. The
.text.fixup section is not mentioned on vmlinux.lds, and is included on
the resulting vmlinux (just after .text) only because of ld heuristics on
placing orphan sections.

However, placing .text.fixup outside .text breaks the definition of
_etext, making it exclude the .text.fixup contents. That makes .text.fixup
be ignored by the kernel initialization code that needs to know about
section locations, such as the code setting page protection bits.

Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
16 years agolguest: update commentry
Rusty Russell [Tue, 26 Aug 2008 05:19:27 +0000 (00:19 -0500)]
lguest: update commentry

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
16 years agostop_machine: Remove deprecated stop_machine_run
Rusty Russell [Tue, 26 Aug 2008 05:19:27 +0000 (00:19 -0500)]
stop_machine: Remove deprecated stop_machine_run

Everyone should be using stop_machine() now.  The staged API
transition helped life in linux-next.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
16 years agostop_machine: wean Xen off stop_machine_run
Rusty Russell [Tue, 26 Aug 2008 05:19:26 +0000 (00:19 -0500)]
stop_machine: wean Xen off stop_machine_run

This is the last use of (the deprecated) stop_machine_run in the tree.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
16 years agovirtio_balloon: fix towards_target when deflating balloon
Anthony Liguori [Mon, 18 Aug 2008 22:15:31 +0000 (17:15 -0500)]
virtio_balloon: fix towards_target when deflating balloon

Both v and vb->num_pages are u32 and unsigned int respectively.  If v is less
than vb->num_pages (and it is, when deflating the balloon), the result is a
very large 32-bit number.  Since we're returning a s64, instead of getting the
same negative number we desire, we get a very large positive number.

This handles the case where v < vb->num_pages and ensures we get a small,
negative, s64 as the result.

Rusty: please push this for 2.6.27-rc4.  It's probably appropriate for the
stable tree too as it will cause an unexpected OOM when ballooning.

Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> (simplified)
16 years agox86: fix cpufreq + sched_clock() regression
Peter Zijlstra [Mon, 25 Aug 2008 11:35:06 +0000 (13:35 +0200)]
x86: fix cpufreq + sched_clock() regression

I noticed that my sched_clock() was slow on a number of machine, so I
started looking at cpufreq.

The below seems to fix the problem for me.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoMerge branch 'linus' into x86/urgent
Ingo Molnar [Mon, 25 Aug 2008 12:39:12 +0000 (14:39 +0200)]
Merge branch 'linus' into x86/urgent

16 years agoALSA: ASoC: Fix double free and memory leak in many codec drivers
Jean Delvare [Mon, 25 Aug 2008 10:49:20 +0000 (11:49 +0100)]
ALSA: ASoC: Fix double free and memory leak in many codec drivers

Many SoC audio codec drivers have improper freeing of memory in error
paths.

* codec is allocated in the platform device probe function, but is not
  freed there in case of error. Instead it is freed in the i2c device
  probe function's error path. However the success or failure of both
  functions is not linked, so this could result in a double free (if
  the platform device is successfully probed, the i2c device probing
  fails and then the platform driver is unregistered.)

* codec->private_data is allocated in many platform device probe
  functions but not freed in their error paths.

This patch hopefully solves all these problems.

Signed-off-by: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agox86: fix HPET regression in 2.6.26 versus 2.6.25, check hpet against BAR, v3
Yinghai Lu [Mon, 25 Aug 2008 07:56:08 +0000 (00:56 -0700)]
x86: fix HPET regression in 2.6.26 versus 2.6.25, check hpet against BAR, v3

David Witbrodt tracked down (and bisected) a hpet bootup hang on his
system to the following problem: a BIOS bug made the hpet device
visible as a generic PCI device. If e820 reserved entries happen to
be registered first in the resource tree [which v2.6.26 started doing],
then the PCI code will reallocate that device's BAR to some other
address - breaking timer IRQs and hanging the system.

( Normally hpet devices are hidden by the BIOS from the OS's PCI
  discovery via chipset magic. Sometimes the hpet is not a PCI device
  at all. )

Solve this fundamental fragility by making non-PCI platform drivers
insert resources into the resource tree even if it overlaps the e820
reserved entry, to keep the resource manager from updating the BAR.

Also do these checks for the ioapic and mmconfig addresses, and emit
a warning if this happens.

Bisected-by: David Witbrodt <dawitbro@sbcglobal.net>
Signed-off-by: Yinghai Lu <yhlu.kernel@gmail.com>
Tested-by: David Witbrodt <dawitbro@sbcglobal.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoALSA: CA0106 on MSI K8N Diamond PLUS Motherboard
Travis Place [Mon, 25 Aug 2008 06:11:50 +0000 (08:11 +0200)]
ALSA: CA0106 on MSI K8N Diamond PLUS Motherboard

Correct a previous patch for the ca0106 onboard the MSI K8N Diamond PLUS
motherboard. Confirmed to have Line/Mic/Aux working for input, and sound
output working as expected.

Signed-off-by: Travis Place <wishie@wishie.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
16 years agox86: do not enable TSC notifier if we don't need it
Linus Torvalds [Sun, 24 Aug 2008 18:52:06 +0000 (11:52 -0700)]
x86: do not enable TSC notifier if we don't need it

Impact: crash on non-TSC-equipped CPUs

Don't enable the TSC notifier if we *either*:

1. don't have a CPU, or
2. have a CPU with constant TSC.

In either of those cases, the notifier is either damaging (1) or useless(2).

From: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
16 years agoFix oops in acer_wmi driver (acer_wmi_init)
Arjan van de Ven [Sun, 24 Aug 2008 04:45:21 +0000 (21:45 -0700)]
Fix oops in acer_wmi driver (acer_wmi_init)

The acer_wmi driver does a DMI scan for quirks, and then sets flags into the
"interface" datastructure for some cases. However, the quirks happen real early
before "interface" is per se initialized from NULL.

The patch below 1) adds a NULL pointer check and 2) (re)runs the quirks at the
end, when "interface" has it's final value.

Reported-by: kerneloops.org
Acked-by: Carlos Corbacho <carlos@strangeworlds.co.uk>
CC: stable@vger.kernel.org
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Sat, 23 Aug 2008 19:14:42 +0000 (12:14 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6:
  ipv6: protocol for address routes
  icmp: icmp_sk() should not use smp_processor_id() in preemptible code
  pkt_sched: Fix qdisc list locking
  pkt_sched: Fix qdisc_watchdog() vs. dev_deactivate() race
  sctp: fix potential panics in the SCTP-AUTH API.

16 years agoremoved unused #include <linux/version.h>'s
Adrian Bunk [Sun, 17 Aug 2008 14:36:59 +0000 (17:36 +0300)]
removed unused #include <linux/version.h>'s

This patch lets the files using linux/version.h match the files that
#include it.

Signed-off-by: Adrian Bunk <bunk@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
16 years agortc: fix deadlock
Ingo Molnar [Sat, 23 Aug 2008 15:59:07 +0000 (17:59 +0200)]
rtc: fix deadlock

if get_rtc_time() is _ever_ called with IRQs off, we deadlock badly
in it, waiting for jiffies to increment.

So make the code more robust by doing an explicit mdelay(20).

This solves a very hard to reproduce/debug hard lockup reported
by Mikael Pettersson.

Reported-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86 MCE: Fix CPU hotplug problem with multiple multicore AMD CPUs
Rafael J. Wysocki [Fri, 22 Aug 2008 20:23:09 +0000 (22:23 +0200)]
x86 MCE: Fix CPU hotplug problem with multiple multicore AMD CPUs

During CPU hot-remove the sysfs directory created by
threshold_create_bank(), defined in
arch/x86/kernel/cpu/mcheck/mce_amd_64.c, has to be removed before
its parent directory, created by mce_create_device(), defined in
arch/x86/kernel/cpu/mcheck/mce_64.c .  Moreover, when the CPU in
question is hotplugged again, obviously the latter has to be created
before the former.  At present, the right ordering is not enforced,
because all of these operations are carried out by CPU hotplug
notifiers which are not appropriately ordered with respect to each
other.  This leads to serious problems on systems with two or more
multicore AMD CPUs, among other things during suspend and hibernation.

Fix the problem by placing threshold bank CPU hotplug callbacks in
mce_cpu_callback(), so that they are invoked at the right places,
if defined.  Additionally, use kobject_del() to remove the sysfs
directory associated with the kobject created by
kobject_create_and_add() in threshold_create_bank(), to prevent the
kernel from crashing during CPU hotplug operations on systems with
two or more multicore AMD CPUs.

This patch fixes bug #11337.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Acked-by: Andi Kleen <andi@firstfloor.org>
Tested-by: Mark Langsdorf <mark.langsdorf@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: fix: make PCI ECS for AMD CPUs hotplug capable
Robert Richter [Fri, 22 Aug 2008 18:23:38 +0000 (20:23 +0200)]
x86: fix: make PCI ECS for AMD CPUs hotplug capable

Until now, PCI ECS setup was performed at boot time only and for cpus
that are enabled then. This patch fixes this and adds cpu hotplug.

Tests sequence (check if ECS bit is set when bringing cpu online again):

 # ( perl -e 'sysseek(STDIN, 0xC001001F, 0)'; hexdump -n 8 -e '2/4 "%08x " "\n"' )   < /dev/cpu/1/msr
 00000008 00404010
 # ( perl -e 'sysseek(STDOUT, 0xC001001F, 0); print pack "l*", 8, 0x00400010' ) > /dev/cpu/1/msr
 # ( perl -e 'sysseek(STDIN, 0xC001001F, 0)'; hexdump -n 8 -e '2/4 "%08x " "\n"' )   < /dev/cpu/1/msr
 00000008 00400010
 # echo 0 > /sys/devices/system/cpu/cpu1/online
 # echo 1 > /sys/devices/system/cpu/cpu1/online
 # ( perl -e 'sysseek(STDIN, 0xC001001F, 0)'; hexdump -n 8 -e '2/4 "%08x " "\n"' )   < /dev/cpu/1/msr
 00000008 00404010

Reported-by: Yinghai Lu <yhlu.kernel@gmail.com>
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agox86: fix: do not run code in amd_bus.c on non-AMD CPUs
Robert Richter [Fri, 22 Aug 2008 18:23:37 +0000 (20:23 +0200)]
x86: fix: do not run code in amd_bus.c on non-AMD CPUs

Jan Beulich wrote:

> Even worse - this would even try to access the MSR on non-AMD CPUs
> (currently probably prevented just by the fact that only AMD ones use
> family values of 0x10 or higher).

This patch adds cpu vendor check to the postcore_initcalls.

Reported-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Robert Richter <robert.richter@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
16 years agoipv6: protocol for address routes
Stephen Hemminger [Sat, 23 Aug 2008 12:16:46 +0000 (05:16 -0700)]
ipv6: protocol for address routes

This fixes a problem spotted with zebra, but not sure if it is
necessary a kernel problem.  With IPV6 when an address is added to an
interface, Zebra creates a duplicate RIB entry, one as a connected
route, and other as a kernel route.

When an address is added to an interface the RTN_NEWADDR message
causes Zebra to create a connected route. In IPV4 when an address is
added to an interface a RTN_NEWROUTE message is set to user space with
the protocol RTPROT_KERNEL. Zebra ignores these messages, because it
already has the connected route.

The problem is that route created in IPV6 has route protocol ==
RTPROT_BOOT.  Was this a design decision or a bug? This fixes it. Same
patch applies to both net-2.6 and stable.

Signed-off-by: Stephen Hemminger <shemminger@vyatta.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoicmp: icmp_sk() should not use smp_processor_id() in preemptible code
Denis V. Lunev [Sat, 23 Aug 2008 11:43:33 +0000 (04:43 -0700)]
icmp: icmp_sk() should not use smp_processor_id() in preemptible code

Pass namespace into icmp_xmit_lock, obtain socket inside and return
it as a result for caller.

Thanks Alexey Dobryan for this report:

Steps to reproduce:

CONFIG_PREEMPT=y
CONFIG_DEBUG_PREEMPT=y
tracepath <something>

BUG: using smp_processor_id() in preemptible [00000000] code: tracepath/3205
caller is icmp_sk+0x15/0x30
Pid: 3205, comm: tracepath Not tainted 2.6.27-rc4 #1

Call Trace:
 [<ffffffff8031af14>] debug_smp_processor_id+0xe4/0xf0
 [<ffffffff80409405>] icmp_sk+0x15/0x30
 [<ffffffff8040a17b>] icmp_send+0x4b/0x3f0
 [<ffffffff8025a415>] ? trace_hardirqs_on_caller+0xd5/0x160
 [<ffffffff8025a4ad>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8023a475>] ? local_bh_enable_ip+0x95/0x110
 [<ffffffff804285b9>] ? _spin_unlock_bh+0x39/0x40
 [<ffffffff8025a26c>] ? mark_held_locks+0x4c/0x90
 [<ffffffff8025a4ad>] ? trace_hardirqs_on+0xd/0x10
 [<ffffffff8025a415>] ? trace_hardirqs_on_caller+0xd5/0x160
 [<ffffffff803e91b4>] ip_fragment+0x8d4/0x900
 [<ffffffff803e7030>] ? ip_finish_output2+0x0/0x290
 [<ffffffff803e91e0>] ? ip_finish_output+0x0/0x60
 [<ffffffff803e6650>] ? dst_output+0x0/0x10
 [<ffffffff803e922c>] ip_finish_output+0x4c/0x60
 [<ffffffff803e92e3>] ip_output+0xa3/0xf0
 [<ffffffff803e68d0>] ip_local_out+0x20/0x30
 [<ffffffff803e753f>] ip_push_pending_frames+0x27f/0x400
 [<ffffffff80406313>] udp_push_pending_frames+0x233/0x3d0
 [<ffffffff804067d1>] udp_sendmsg+0x321/0x6f0
 [<ffffffff8040d155>] inet_sendmsg+0x45/0x80
 [<ffffffff803b967f>] sock_sendmsg+0xdf/0x110
 [<ffffffff8024a100>] ? autoremove_wake_function+0x0/0x40
 [<ffffffff80257ce5>] ? validate_chain+0x415/0x1010
 [<ffffffff8027dc10>] ? __do_fault+0x140/0x450
 [<ffffffff802597d0>] ? __lock_acquire+0x260/0x590
 [<ffffffff803b9e55>] ? sockfd_lookup_light+0x45/0x80
 [<ffffffff803ba50a>] sys_sendto+0xea/0x120
 [<ffffffff80428e42>] ? _spin_unlock_irqrestore+0x42/0x80
 [<ffffffff803134bc>] ? __up_read+0x4c/0xb0
 [<ffffffff8024e0c6>] ? up_read+0x26/0x30
 [<ffffffff8020b8bb>] system_call_fastpath+0x16/0x1b

icmp6_sk() is similar.

Signed-off-by: Denis V. Lunev <den@openvz.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
16 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6
Linus Torvalds [Fri, 22 Aug 2008 15:38:00 +0000 (08:38 -0700)]
Merge git://git./linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6:
  crypto: authenc - Avoid using clobbered request pointer

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Fri, 22 Aug 2008 15:37:45 +0000 (08:37 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tiwai/sound-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6:
  ALSA: hda - Fix call of alc888_coef_init()
  ALSA: hda_intel: enable snoop for nvidia HDA controller

16 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4
Linus Torvalds [Fri, 22 Aug 2008 15:37:07 +0000 (08:37 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: Update documentation to remind users to update mke2fs.conf
  ext4: Fix small file fragmentation
  ext4: Initialize writeback_index to 0 when allocating a new inode
  ext4: make sure ext4_has_free_blocks returns 0 for ENOSPC
  ext4: journal credit fix for the delayed allocation's writepages() function
  ext4: Rework the ext4_da_writepages() function
  ext4: journal credits reservation fixes for DIO, fallocate
  ext4: journal credits reservation fixes for extent file writepage
  ext4: journal credits calulation cleanup and fix for non-extent writepage
  ext4: Fix bug where we return ENOSPC even though we have plenty of inodes
  ext4: don't try to resize if there are no reserved gdt blocks left
  ext4: Use ext4_discard_reservations instead of mballoc-specific call
  ext4: Fix ext4_dx_readdir hash collision handling
  ext4: Fix delalloc release block reservation for truncate
  ext4: Fix potential truncate BUG due to i_prealloc_list being non-empty
  ext4: Handle unwritten extent properly with delayed allocation

16 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 22 Aug 2008 15:36:55 +0000 (08:36 -0700)]
Merge branch 'sched-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  sched: enable LB_BIAS by default

16 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 22 Aug 2008 15:36:42 +0000 (08:36 -0700)]
Merge branch 'core-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  rcu: fix synchronize_rcu() so that kernel-doc works

16 years agoMerge branch 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 22 Aug 2008 15:36:20 +0000 (08:36 -0700)]
Merge branch 'timers-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  nohz: fix wrong event handler after online an offlined cpu

16 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 22 Aug 2008 15:23:53 +0000 (08:23 -0700)]
Merge branch 'x86-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  x86: work around MTRR mask setting, v2
  x86: fix section mismatch warning - uv_cpu_init
  x86: fix VMI for early params
  x86: fix two modpost warnings in mm/init_64.c
  x86: fix 1:1 mapping init on 64-bit (memory hotplug case)
  x86: work around MTRR mask setting
  x86: PAT Update validate_pat_support for intel CPUs
  devmem, x86: PAT Change /dev/mem mmap with O_SYNC to use UC_MINUS
  x86: PAT proper tracking of set_memory_uc and friends
  x86: fix BUG: unable to handle kernel paging request (numaq_tsc_disable)
  x86: export pv_lock_ops non-GPL
  x86, mmiotrace: silence section mismatch warning - leave_uniprocessor
  x86: use WARN() in arch/x86/kernel
  x86: use WARN() in arch/x86/mm/ioremap.c
  werror: fix pci calgary
  x86: fix oprofile + hibernation badness
  x86, SGI UV: hardcode the TLB flush interrupt system vector
  x86: fix Xorg startup/shutdown slowdown with PAT
  x86: fix "kernel won't boot on a Cyrix MediaGXm (Geode)"
  x86 iommu: remove unneeded parenthesis

16 years agoMerge branch 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzi...
Linus Torvalds [Fri, 22 Aug 2008 15:22:33 +0000 (08:22 -0700)]
Merge branch 'upstream-linus' of git://git./linux/kernel/git/jgarzik/libata-dev

* 'upstream-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jgarzik/libata-dev:
  [libata] pata_it821x: fix warning
  libata: Fix a large collection of DMA mode mismatches
  ahci: sis controllers actually can do PMP
  pata_via: clean up recent tf_load changes
  libata: restore SControl on detach
  libata: use ata_link_printk() when printing SError
  libata: always do follow-up SRST if hardreset returned -EAGAIN
  libata: fix EH action overwriting in ata_eh_reset()
  sata_mv: add the Gen IIE flag to the SoC devices.
  ata_piix: IDE Mode SATA patch for Intel Ibex Peak DeviceIDs
  ahci: RAID mode SATA patch for Intel Ibex Peak DeviceIDs
  sata_mv: don't issue two DMA commands concurrently
  libata: implement no[hs]rst force params

16 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland...
Linus Torvalds [Fri, 22 Aug 2008 15:21:48 +0000 (08:21 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/roland/infiniband

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/roland/infiniband:
  IPoIB: Fix deadlock on RTNL in ipoib_stop()
  IB/ipath: Fix incorrect check for max physical address in TID
  IB/ipath: Fix lost UD send work request

16 years agocrypto: authenc - Avoid using clobbered request pointer
Herbert Xu [Fri, 22 Aug 2008 15:04:06 +0000 (01:04 +1000)]
crypto: authenc - Avoid using clobbered request pointer

Authenc works in two stages for encryption, it first encrypts and
then computes an ICV.  The context memory of the request is used
by both operations.  The problem is that when an asynchronous
encryption completes, we will compute the ICV and then reread the
context memory of the encryption to get the original request.

It just happens that we have a buffer of 16 bytes in front of the
request pointer, so ICVs of 16 bytes (such as SHA1) do not trigger
the bug.  However, any attempt to uses a larger ICV instantly kills
the machine when the first asynchronous encryption is completed.

This patch fixes this by saving the request pointer before we start
the ICV computation.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
16 years agoALSA: hda - Fix call of alc888_coef_init()
Takashi Iwai [Fri, 22 Aug 2008 13:24:22 +0000 (15:24 +0200)]
ALSA: hda - Fix call of alc888_coef_init()

Using init_hook to call alc888_coef_init() is problematic for configurations
that already set another init_hook.  Better to put it in alc_init() as is
(although it looks a bit hackish).

Signed-off-by: Takashi Iwai <tiwai@suse.de>