GitHub/MotorolaMobilityLLC/kernel-slsi.git
15 years agodma-debug: change hash_bucket_find from first-fit to best-fit
Joerg Roedel [Fri, 5 Jun 2009 10:01:35 +0000 (12:01 +0200)]
dma-debug: change hash_bucket_find from first-fit to best-fit

Some device drivers map the same physical address multiple times to a
dma address. Without an IOMMU this results in the same dma address being
put into the dma-debug hash multiple times. With a first-fit match in
hash_bucket_find() this function may return the wrong dma_debug_entry.

This can result in false positive warnings. This patch fixes it by
changing the first-fit behavior of hash_bucket_find() into a best-fit
algorithm.

Reported-by: Torsten Kaiser <just.for.lkml@googlemail.com>
Reported-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Cc: lethal@linux-sh.org
Cc: just.for.lkml@googlemail.com
Cc: hancockrwd@gmail.com
Cc: jens.axboe@oracle.com
Cc: bharrosh@panasas.com
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: <stable@kernel.org>
LKML-Reference: <20090605104132.GE24836@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: enable GART-IOMMU only after setting up protection methods
Mark Langsdorf [Sun, 5 Jul 2009 20:50:52 +0000 (15:50 -0500)]
x86: enable GART-IOMMU only after setting up protection methods

The current code to set up the GART as an IOMMU enables GART
translations before it removes the aperture from the kernel memory
map, sets the GART PTEs to UC, sets up the guard and scratch
pages, or does a wbinvd().  This leaves the possibility of cache
aliasing open and can cause system crashes.

Re-order the code so as to enable the GART translations only
after all safeguards are in place and the tlb has been flushed.

AMD has tested this patch on both Istanbul systems and 1st
generation Opteron systems with APG enabled and seen no adverse
effects.  Istanbul systems with HT Assist enabled sometimes
see MCE errors due to cache artifacts with the unmodified
code.

Signed-off-by: Mark Langsdorf <mark.langsdorf@amd.com>
Cc: <stable@kernel.org>
Cc: Joerg Roedel <joerg.roedel@amd.com>
Cc: akpm@linux-foundation.org
Cc: jbarnes@virtuousgeek.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoMerge branches 'amd-iommu/fixes', 'amd-iommu/debug', 'amd-iommu/suspend-resume' and...
Joerg Roedel [Thu, 28 May 2009 16:23:56 +0000 (18:23 +0200)]
Merge branches 'amd-iommu/fixes', 'amd-iommu/debug', 'amd-iommu/suspend-resume' and 'amd-iommu/extended-allocator' into amd-iommu/2.6.31

Conflicts:
arch/x86/kernel/amd_iommu.c
arch/x86/kernel/amd_iommu_init.c

15 years agoamd-iommu: don't free dma adresses below 512MB with CONFIG_IOMMU_STRESS
Joerg Roedel [Fri, 22 May 2009 10:40:54 +0000 (12:40 +0200)]
amd-iommu: don't free dma adresses below 512MB with CONFIG_IOMMU_STRESS

This will test the automatic aperture enlargement code. This is
important because only very few devices will ever trigger this code
path. So force it under CONFIG_IOMMU_STRESS.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: don't preallocate page tables with CONFIG_IOMMU_STRESS
Joerg Roedel [Fri, 22 May 2009 10:31:53 +0000 (12:31 +0200)]
amd-iommu: don't preallocate page tables with CONFIG_IOMMU_STRESS

This forces testing of on-demand page table allocation code.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: disable round-robin allocator for CONFIG_IOMMU_STRESS
Joerg Roedel [Fri, 22 May 2009 10:27:53 +0000 (12:27 +0200)]
amd-iommu: disable round-robin allocator for CONFIG_IOMMU_STRESS

Disabling the round-robin allocator results in reusing the same
dma-addresses again very fast. This is a good test if the iotlb flushing
is working correctly.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: remove amd_iommu_size kernel parameter
Joerg Roedel [Tue, 19 May 2009 10:16:29 +0000 (12:16 +0200)]
amd-iommu: remove amd_iommu_size kernel parameter

This parameter is not longer necessary when aperture increases
dynamically.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: enlarge the aperture dynamically
Joerg Roedel [Tue, 19 May 2009 08:23:15 +0000 (10:23 +0200)]
amd-iommu: enlarge the aperture dynamically

By dynamically increasing the aperture the extended allocator is now
ready for use.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: handle exlusion ranges and unity mappings in alloc_new_range
Joerg Roedel [Tue, 19 May 2009 07:52:40 +0000 (09:52 +0200)]
amd-iommu: handle exlusion ranges and unity mappings in alloc_new_range

This patch makes sure no reserved addresses are allocated in an dma_ops
domain when the aperture is increased dynamically.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: move aperture_range allocation code to seperate function
Joerg Roedel [Mon, 18 May 2009 14:38:55 +0000 (16:38 +0200)]
amd-iommu: move aperture_range allocation code to seperate function

This patch prepares the dynamic increasement of dma_ops domain
apertures.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: change dma_dom->next_bit to dma_dom->next_address
Joerg Roedel [Mon, 18 May 2009 13:32:48 +0000 (15:32 +0200)]
amd-iommu: change dma_dom->next_bit to dma_dom->next_address

Simplify the code a little bit by using the same unit for all address
space related state in the dma_ops domain structure.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: make address allocator aware of multiple aperture ranges
Joerg Roedel [Fri, 15 May 2009 10:30:05 +0000 (12:30 +0200)]
amd-iommu: make address allocator aware of multiple aperture ranges

This patch changes the AMD IOMMU address allocator to allow up to 32
aperture ranges per dma_ops domain.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: handle page table allocation failures in dma_ops code
Joerg Roedel [Tue, 12 May 2009 10:17:38 +0000 (12:17 +0200)]
amd-iommu: handle page table allocation failures in dma_ops code

The code will be required when the aperture size increases dynamically
in the extended address allocator.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: move page table allocation code to seperate function
Joerg Roedel [Tue, 12 May 2009 10:02:46 +0000 (12:02 +0200)]
amd-iommu: move page table allocation code to seperate function

This patch makes page table allocation usable for dma_ops code.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: introduce aperture_range structure
Joerg Roedel [Tue, 12 May 2009 08:56:44 +0000 (10:56 +0200)]
amd-iommu: introduce aperture_range structure

This is a preperation for extended address allocator.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: implement suspend/resume
Joerg Roedel [Tue, 12 May 2009 07:56:12 +0000 (09:56 +0200)]
amd-iommu: implement suspend/resume

This patch puts everything together and enables suspend/resume support
in the AMD IOMMU driver.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd_iommu: un __init functions required for suspend/resume
Joerg Roedel [Tue, 12 May 2009 07:52:46 +0000 (09:52 +0200)]
amd_iommu: un __init functions required for suspend/resume

This patch makes sure that no function required for suspend/resume of
AMD IOMMU driver is thrown away after boot.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: add function to flush tlb for all devices
Joerg Roedel [Tue, 5 May 2009 13:48:10 +0000 (15:48 +0200)]
amd-iommu: add function to flush tlb for all devices

This function is required for suspend/resume support with AMD IOMMU
enabled.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: add function to flush tlb for all domains
Joerg Roedel [Tue, 5 May 2009 13:33:57 +0000 (15:33 +0200)]
amd-iommu: add function to flush tlb for all domains

This function is required for suspend/resume support with AMD IOMMU
enabled.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: add function to disable all iommus
Joerg Roedel [Tue, 19 May 2009 17:06:27 +0000 (19:06 +0200)]
amd-iommu: add function to disable all iommus

This function is required for suspend/resume support with AMD IOMMU
enabled.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: remove support for msi-x
Joerg Roedel [Mon, 4 May 2009 16:51:00 +0000 (18:51 +0200)]
amd-iommu: remove support for msi-x

Current hardware uses msi instead of msi-x so this code it not necessary
and can not be tested. The best thing is to drop this code.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: drop pointless iommu-loop in msi setup code
Joerg Roedel [Mon, 4 May 2009 16:46:34 +0000 (18:46 +0200)]
amd-iommu: drop pointless iommu-loop in msi setup code

It is not necessary to loop again over all IOMMUs in this code. So drop
the loop.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: consolidate hardware initialization to one function
Joerg Roedel [Mon, 4 May 2009 16:41:16 +0000 (18:41 +0200)]
amd-iommu: consolidate hardware initialization to one function

This patch restructures the AMD IOMMU initialization code to initialize
all hardware registers with one single function call.
This is helpful for suspend/resume support.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: introduce for_each_iommu* macros
Joerg Roedel [Mon, 4 May 2009 13:06:20 +0000 (15:06 +0200)]
amd-iommu: introduce for_each_iommu* macros

This patch introduces the for_each_iommu and for_each_iommu_safe macros
to simplify the developers life when having to iterate over all AMD
IOMMUs in the system.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd iommu: properly detach from protection domain on ->remove
Chris Wright [Thu, 21 May 2009 07:56:58 +0000 (00:56 -0700)]
amd iommu: properly detach from protection domain on ->remove

Some drivers may use the dma api during ->remove which will
cause a protection domain to get reattached to a device.  Delay the
detach until after the driver is completely unbound.

[ joro: added a little merge helper ]

[ Impact: fix too early device<->domain removal ]

Signed-off-by: Chris Wright <chrisw@redhat.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: make sure only ivmd entries are parsed
Joerg Roedel [Fri, 22 May 2009 10:48:05 +0000 (12:48 +0200)]
amd-iommu: make sure only ivmd entries are parsed

The bug never triggered. But it should be fixed to protect against
broken ACPI tables in the future.

[ Impact: protect against broken ivrs acpi table ]

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: fix the handling of device aliases in the AMD IOMMU driver.
Neil Turton [Thu, 14 May 2009 13:08:11 +0000 (14:08 +0100)]
amd-iommu: fix the handling of device aliases in the AMD IOMMU driver.

The devid parameter to set_dev_entry_from_acpi is the requester ID
rather than the device ID since it is used to index the IOMMU device
table.  The handling of IVHD_DEV_ALIAS used to pass the device ID.
This patch fixes it to pass the requester ID.

[ Impact: fix setting the wrong req-id in acpi-table parsing ]

Signed-off-by: Neil Turton <nturton@solarflare.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: fix an off-by-one error in the AMD IOMMU driver.
Neil Turton [Thu, 14 May 2009 13:00:35 +0000 (14:00 +0100)]
amd-iommu: fix an off-by-one error in the AMD IOMMU driver.

The variable amd_iommu_last_bdf holds the maximum bdf of any device
controlled by an IOMMU, so the number of device entries needed is
amd_iommu_last_bdf+1.  The function tbl_size used amd_iommu_last_bdf
instead.  This would be a problem if the last device were a large
enough power of 2.

[ Impact: fix amd_iommu_last_bdf off-by-one error ]

Signed-off-by: Neil Turton <nturton@solarflare.com>
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: disable device isolation with CONFIG_IOMMU_STRESS
Joerg Roedel [Fri, 22 May 2009 10:44:03 +0000 (12:44 +0200)]
amd-iommu: disable device isolation with CONFIG_IOMMU_STRESS

With device isolation disabled we can test better for race conditions in
dma_ops related code.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agox86/iommu: add IOMMU_STRESS Kconfig entry
Joerg Roedel [Fri, 22 May 2009 10:15:49 +0000 (12:15 +0200)]
x86/iommu: add IOMMU_STRESS Kconfig entry

This Kconfig option is intended to enable various code paths or
parameters in IOMMU implementations to stress test the code and/or the
hardware. This can also be done by disabling optimizations in the code
when this option is switched on.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
15 years agoamd-iommu: move protection domain printk to dump code
Joerg Roedel [Fri, 22 May 2009 10:02:48 +0000 (12:02 +0200)]
amd-iommu: move protection domain printk to dump code

This information is only helpful for debugging. Don't print it anymore
unless explicitly requested.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: print ivmd information to dmesg when requested
Joerg Roedel [Wed, 20 May 2009 14:24:21 +0000 (16:24 +0200)]
amd-iommu: print ivmd information to dmesg when requested

Add information about device memory mapping requirements for the IOMMU
as described in the IVRS ACPI table to the kernel log if amd_iommu_dump
was specified on the kernel command line.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: print ivhd information to dmesg when requested
Joerg Roedel [Wed, 20 May 2009 13:41:28 +0000 (15:41 +0200)]
amd-iommu: print ivhd information to dmesg when requested

Add information about devices belonging to an IOMMU as described in the
IVRS ACPI table to the kernel log if amd_iommu_dump was specified on the
kernel command line.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: add dump for iommus described in ivrs table
Joerg Roedel [Wed, 20 May 2009 11:53:57 +0000 (13:53 +0200)]
amd-iommu: add dump for iommus described in ivrs table

Add information about IOMMU devices described in the IVRS ACPI table to
the kernel log if amd_iommu_dump was specified on the kernel command
line.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoamd-iommu: add amd_iommu_dump parameter
Joerg Roedel [Wed, 20 May 2009 10:21:42 +0000 (12:21 +0200)]
amd-iommu: add amd_iommu_dump parameter

This kernel parameter will be useful to get some AMD IOMMU related
information in dmesg that is not necessary for the default user but may
be helpful in debug situations.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
15 years agoMerge commit 'v2.6.30-rc5' into core/iommu
Ingo Molnar [Mon, 11 May 2009 12:44:27 +0000 (14:44 +0200)]
Merge commit 'v2.6.30-rc5' into core/iommu

Merge reason: core/iommu was on an .30-rc1 base,
              update it to .30-rc5 to refresh.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoLinux 2.6.30-rc5
Linus Torvalds [Sat, 9 May 2009 00:14:14 +0000 (17:14 -0700)]
Linux 2.6.30-rc5

15 years agoMerge git://git.infradead.org/mtd-2.6
Linus Torvalds [Sat, 9 May 2009 00:00:41 +0000 (17:00 -0700)]
Merge git://git.infradead.org/mtd-2.6

* git://git.infradead.org/mtd-2.6:
  mtd: fix timeout in M25P80 driver
  mtd: Bug in m25p80.c during whole-chip erase
  mtd: expose subpage size via sysfs
  mtd: mtd in mtd_release is unused without CONFIG_MTD_CHAR

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 8 May 2009 23:59:23 +0000 (16:59 -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: MCE: make cmci_discover_lock irq-safe
  x86: xen, i386: reserve Xen pagetables
  x86, kexec: fix crashdump panic with CONFIG_KEXEC_JUMP
  x86-64: finish cleanup_highmaps()'s job wrt. _brk_end
  x86: fix boot hang in early_reserve_e820()
  x86: Fix a typo in a printk message
  x86, srat: do not register nodes beyond e820 map

15 years agoMerge branch 'hwmon-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6
Linus Torvalds [Fri, 8 May 2009 23:25:00 +0000 (16:25 -0700)]
Merge branch 'hwmon-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6

* 'hwmon-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6:
  hwmon: (w83781d) Fix W83782D support (NULL pointer dereference)
  hwmon: (asus_atk0110) Fix compiler warning

15 years agoMerge branch 'fixes-for-linus' of git://git.monstr.eu/linux-2.6-microblaze
Linus Torvalds [Fri, 8 May 2009 23:24:25 +0000 (16:24 -0700)]
Merge branch 'fixes-for-linus' of git://git.monstr.eu/linux-2.6-microblaze

* 'fixes-for-linus' of git://git.monstr.eu/linux-2.6-microblaze:
  microblaze: Fix return value for sys_ipc
  microblaze: Storage class should be before const qualifier

15 years agokprobes: fix to use text_mutex around arm/disarm kprobe
Masami Hiramatsu [Thu, 7 May 2009 20:31:26 +0000 (16:31 -0400)]
kprobes: fix to use text_mutex around arm/disarm kprobe

Fix kprobes to lock text_mutex around some arch_arm/disarm_kprobe() which
are newly added by commit de5bd88d5a5cce3cacea904d3503e5ebdb3852a2.

Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Cc: Jim Keniston <jkenisto@us.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agohwmon: (w83781d) Fix W83782D support (NULL pointer dereference)
Jean Delvare [Fri, 8 May 2009 18:27:28 +0000 (20:27 +0200)]
hwmon: (w83781d) Fix W83782D support (NULL pointer dereference)

Commit 360782dde00a2e6e7d9fd57535f90934707ab8a8 (hwmon: (w83781d) Stop
abusing struct i2c_client for ISA devices) broke W83782D support for
devices connected on the ISA bus. You will hit a NULL pointer
dereference as soon as you read any device attribute. Other devices,
and W83782D devices on the SMBus, aren't affected.

Reported-by: Michel Abraham
Signed-off-by: Jean Delvare <khali@linux-fr.org>
Tested-by: Michel Abraham
15 years agohwmon: (asus_atk0110) Fix compiler warning
Luca Tettamanti [Fri, 8 May 2009 18:27:28 +0000 (20:27 +0200)]
hwmon: (asus_atk0110) Fix compiler warning

atk_sensor_type is only used when DEBUG is defined.

Signed-off-by: Luca Tettamanti <kronos.it@gmail.com>
Signed-off-by: Jean Delvare <khali@linux-fr.org>
15 years agomtd: fix timeout in M25P80 driver
Peter Horton [Fri, 8 May 2009 12:51:53 +0000 (13:51 +0100)]
mtd: fix timeout in M25P80 driver

Extend erase timeout in M25P80 SPI Flash driver.

The M25P80 drivers fails erasing sectors on a M25P128 because the ready
wait timeout is too short. Change the timeout from a simple loop count to a
suitable number of seconds.

Signed-off-by: Peter Horton <zero@colonel-panic.org>
Tested-by: Martin Michlmayr <tbm@cyrius.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
15 years agox86: MCE: make cmci_discover_lock irq-safe
Hidetoshi Seto [Fri, 8 May 2009 08:28:40 +0000 (17:28 +0900)]
x86: MCE: make cmci_discover_lock irq-safe

Lockdep reports the warning below when Li tries to offline one cpu:

[  110.835487] =================================
[  110.835616] [ INFO: inconsistent lock state ]
[  110.835688] 2.6.30-rc4-00336-g8c9ed89 #52
[  110.835757] ---------------------------------
[  110.835828] inconsistent {HARDIRQ-ON-W} -> {IN-HARDIRQ-W} usage.
[  110.835908] swapper/0 [HC1[1]:SC0[0]:HE0:SE1] takes:
[  110.835982]  (cmci_discover_lock){?.+...}, at: [<ffffffff80236dc0>] cmci_clear+0x30/0x9b

cmci_clear() can be called via smp_call_function_single().

It is better to disable interrupt while holding cmci_discover_lock,
to turn it into an irq-safe lock - we can deadlock otherwise.

[ Impact: fix possible deadlock in the MCE code ]

Reported-by: Shaohua Li <shaohua.li@intel.com>
Signed-off-by: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
LKML-Reference: <4A03ED38.8000700@jp.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Reported-by: Shaohua Li<shaohua.li@intel.com>
15 years agox86: xen, i386: reserve Xen pagetables
Jeremy Fitzhardinge [Thu, 7 May 2009 18:56:44 +0000 (11:56 -0700)]
x86: xen, i386: reserve Xen pagetables

The Xen pagetables are no longer implicitly reserved as part of the other
i386_start_kernel reservations, so make sure we explicitly reserve them.
This prevents them from being released into the general kernel free page
pool and reused.

[ Impact: fix Xen guest crash ]

Also-Bisected-by: Bryan Donlan <bdonlan@gmail.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Xen-devel <xen-devel@lists.xensource.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
LKML-Reference: <4A032EEC.30509@goop.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, kexec: fix crashdump panic with CONFIG_KEXEC_JUMP
Huang Ying [Fri, 8 May 2009 02:51:41 +0000 (10:51 +0800)]
x86, kexec: fix crashdump panic with CONFIG_KEXEC_JUMP

Tim Starling reported that crashdump will panic with kernel compiled
with CONFIG_KEXEC_JUMP due to null pointer deference in
machine_kexec_32.c: machine_kexec(), when deferencing
kexec_image. Refering to:

http://bugzilla.kernel.org/show_bug.cgi?id=13265

This patch fixes the BUG via replacing global variable reference:
kexec_image in machine_kexec() with local variable reference: image,
which is more appropriate, and will not be null.

Same BUG is in machine_kexec_64.c too, so fixed too in the same way.

[ Impact: fix crash on kexec ]

Reported-by: Tim Starling <tstarling@wikimedia.org>
Signed-off-by: Huang Ying <ying.huang@intel.com>
LKML-Reference: <1241751101.6259.85.camel@yhuang-dev.sh.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
15 years agox86-64: finish cleanup_highmaps()'s job wrt. _brk_end
Jan Beulich [Wed, 6 May 2009 12:06:47 +0000 (13:06 +0100)]
x86-64: finish cleanup_highmaps()'s job wrt. _brk_end

With the introduction of the .brk section, special care must be taken
that no unused page table entries remain if _brk_end and _end are
separated by a 2M page boundary. cleanup_highmap() runs very early and
hence cannot take care of that, hence potential entries needing to be
removed past _brk_end must be cleared once the brk allocator has done
its job.

[ Impact: avoids undesirable TLB aliases ]

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
15 years agox86: fix boot hang in early_reserve_e820()
Jan Beulich [Wed, 6 May 2009 12:02:19 +0000 (13:02 +0100)]
x86: fix boot hang in early_reserve_e820()

If the first non-reserved (sub-)range doesn't fit the size requested,
an endless loop will be entered. If a range returned from
find_e820_area_size() turns out insufficient in size, the range must
be skipped before calling the function again.

[ Impact: fixes boot hang on some platforms ]

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6
Linus Torvalds [Fri, 8 May 2009 04:13:24 +0000 (21:13 -0700)]
Merge git://git./linux/kernel/git/sfrench/cifs-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/sfrench/cifs-2.6: (32 commits)
  [CIFS] Fix double list addition in cifs posix open code
  [CIFS] Allow raw ntlmssp code to be enabled with sec=ntlmssp
  [CIFS] Fix SMB uid in NTLMSSP authenticate request
  [CIFS] NTLMSSP reenabled after move from connect.c to sess.c
  [CIFS] Remove sparse warning
  [CIFS] remove checkpatch warning
  [CIFS] Fix final user of old string conversion code
  [CIFS] remove cifs_strfromUCS_le
  [CIFS] NTLMSSP support moving into new file, old dead code removed
  [CIFS] Fix endian conversion of vcnum field
  [CIFS] Remove trailing whitespace
  [CIFS] Remove sparse endian warnings
  [CIFS] Add remaining ntlmssp flags and standardize field names
  [CIFS] Fix build warning
  cifs: fix length handling in cifs_get_name_from_search_buf
  [CIFS] Remove unneeded QuerySymlink call and fix mapping for unmapped status
  [CIFS] rename cifs_strndup to cifs_strndup_from_ucs
  Added loop check when mounting DFS tree.
  Enable dfs submounts to handle remote referrals.
  [CIFS] Remove older session setup implementation
  ...

15 years ago[CIFS] Fix double list addition in cifs posix open code
Steve French [Fri, 8 May 2009 03:04:30 +0000 (03:04 +0000)]
[CIFS] Fix double list addition in cifs posix open code

Remove adding open file entry twice to lists in the file
Do not fill file info twice in case of posix opens and creates

Signed-off-by: Shirish Pargaonkar <shirishp@us.ibm.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
15 years agoNOMMU: Don't check vm_region::vm_start is page aligned in add_nommu_region()
David Howells [Thu, 7 May 2009 10:41:37 +0000 (11:41 +0100)]
NOMMU: Don't check vm_region::vm_start is page aligned in add_nommu_region()

Don't check vm_region::vm_start is page aligned in add_nommu_region() because
the region may reflect some non-page-aligned mapped file, such as could be
obtained from RomFS XIP.

Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Greg Ungerer <gerg@uclinux.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'for-linus' of git://neil.brown.name/md
Linus Torvalds [Thu, 7 May 2009 19:01:41 +0000 (12:01 -0700)]
Merge branch 'for-linus' of git://neil.brown.name/md

* 'for-linus' of git://neil.brown.name/md:
  md: remove rd%d links immediately after stopping an array.
  md: remove ability to explicit set an inactive array to 'clean'.
  md: constify VFTs
  md: tidy up status_resync to handle large arrays.
  md: fix some (more) errors with bitmaps on devices larger than 2TB.
  md/raid10: don't clear bitmap during recovery if array will still be degraded.
  md: fix loading of out-of-date bitmap.

15 years agorandom: make get_random_int() more random
Linus Torvalds [Tue, 5 May 2009 15:17:43 +0000 (08:17 -0700)]
random: make get_random_int() more random

It's a really simple patch that basically just open-codes the current
"secure_ip_id()" call, but when open-coding it we now use a _static_
hashing area, so that it gets updated every time.

And to make sure somebody can't just start from the same original seed of
all-zeroes, and then do the "half_md4_transform()" over and over until
they get the same sequence as the kernel has, each iteration also mixes in
the same old "current->pid + jiffies" we used - so we should now have a
regular strong pseudo-number generator, but we also have one that doesn't
have a single seed.

Note: the "pid + jiffies" is just meant to be a tiny tiny bit of noise. It
has no real meaning. It could be anything. I just picked the previous
seed, it's just that now we keep the state in between calls and that will
feed into the next result, and that should make all the difference.

I made that hash be a per-cpu data just to avoid cache-line ping-pong:
having multiple CPU's write to the same data would be fine for randomness,
and add yet another layer of chaos to it, but since get_random_int() is
supposed to be a fast interface I did it that way instead. I considered
using "__raw_get_cpu_var()" to avoid any preemption overhead while still
getting the hash be _mostly_ ping-pong free, but in the end good taste won
out.

Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Thu, 7 May 2009 17:54:32 +0000 (10:54 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm:
  [ARM] 5507/1: support R_ARM_MOVW_ABS_NC and MOVT_ABS relocation types
  [ARM] 5506/1: davinci: DMA_32BIT_MASK --> DMA_BIT_MASK(32)
  i.MX31: Disable CPU_32v6K in mx3_defconfig.
  mx3fb: Fix compilation with CONFIG_PM
  mx27ads: move PBC mapping out of vmalloc space
  MXC: remove BUG_ON in interrupt handler
  mx31: remove mx31moboard_defconfig
  ARM: ARCH_MXC should select HAVE_CLK
  mxc : BUG in imx_dma_request
  mxc : Clean up properly when imx_dma_free() used without imx_dma_disable()
  [ARM] mv78xx0: update defconfig
  [ARM] orion5x: update defconfig
  [ARM] Kirkwood: update defconfig
  [ARM] Kconfig typo fix:  "PXA930" -> "CPU_PXA930".
  [ARM] S3C2412: Add missing cache flush in suspend code
  [ARM] S3C: Add UDIVSLOT support for newer UARTS
  [ARM] S3C64XX: Add S3C64XX_PA_IIS{0,1} to <mach/map.h>

15 years ago[ARM] 5507/1: support R_ARM_MOVW_ABS_NC and MOVT_ABS relocation types
Paul Gortmaker [Thu, 7 May 2009 15:18:40 +0000 (16:18 +0100)]
[ARM] 5507/1: support R_ARM_MOVW_ABS_NC and MOVT_ABS relocation types

From: Bruce Ashfield <bruce.ashfield@windriver.com>

To fully support the armv7-a instruction set/optimizations, support
for the R_ARM_MOVW_ABS_NC and R_ARM_MOVT_ABS relocation types is
required.

The MOVW and MOVT are both load-immediate instructions, MOVW loads 16
bits into the bottom half of a register, and MOVT loads 16 bits into the
top half of a register.

The relocation information for these instructions has a full 32 bit
value, plus an addend which is stored in the 16 immediate bits in the
instruction itself.  The immediate bits in the instruction are not
contiguous (the register # splits it into a 4 bit and 12 bit value),
so the addend has to be extracted accordingly and added to the value.
The value is then split and put into the instruction; a MOVW uses the
bottom 16 bits of the value, and a MOVT uses the top 16 bits.

Signed-off-by: David Borman <david.borman@windriver.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years ago[ARM] 5506/1: davinci: DMA_32BIT_MASK --> DMA_BIT_MASK(32)
Kevin Hilman [Thu, 7 May 2009 13:25:48 +0000 (14:25 +0100)]
[ARM] 5506/1: davinci: DMA_32BIT_MASK --> DMA_BIT_MASK(32)

As per commit 284901a90a9e0b812ca3f5f852cbbfb60d10249d, use
DMA_BIT_MASK(n)

Signed-off-by: Kevin Hilman <khilman@deeprootsystems.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agomd: remove rd%d links immediately after stopping an array.
NeilBrown [Thu, 7 May 2009 02:51:06 +0000 (12:51 +1000)]
md: remove rd%d links immediately after stopping an array.

md maintains link in sys/mdXX/md/ to identify which device has
which role in the array. e.g.
   rd2 -> dev-sda

indicates that the device with role '2' in the array is sda.

These links are only present when the array is active.  They are
created immediately after ->run is called, and so should be removed
immediately after ->stop is called.
However they are currently removed a little bit later, and it is
possible for ->run to be called again, thus adding these links, before
they are removed.

So move the removal earlier so they are consistently only present when
the array is active.

Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: remove ability to explicit set an inactive array to 'clean'.
NeilBrown [Thu, 7 May 2009 02:50:57 +0000 (12:50 +1000)]
md: remove ability to explicit set an inactive array to 'clean'.

Being able to write 'clean' to an 'array_state' of an inactive array
to activate it in 'clean' mode is both unnecessary and inconvenient.

It is unnecessary because the same can be achieved by writing
'active'.  This activates and array, but it still remains 'clean'
until the first write.

It is inconvenient because writing 'clean' is more often used to
cause an 'active' array to revert to 'clean' mode (thus blocking
any writes until a 'write-pending' is promoted to 'active').

Allowing 'clean' to both activate an array and mark an active array as
clean can lead to races:  One program writes 'clean' to mark the
active array as clean at the same time as another program writes
'inactive' to deactivate (stop) and active array.  Depending on which
writes first, the array could be deactivated and immediately
reactivated which isn't what was desired.

So just disable the use of 'clean' to activate an array.

This avoids a race that can be triggered with mdadm-3.0 and external
metadata, so it suitable for -stable.

Reported-by: Rafal Marszewski <rafal.marszewski@intel.com>
Acked-by: Dan Williams <dan.j.williams@intel.com>
Cc: <stable@kernel.org>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: constify VFTs
Jan Engelhardt [Thu, 7 May 2009 02:49:37 +0000 (12:49 +1000)]
md: constify VFTs

Signed-off-by: Jan Engelhardt <jengelh@medozas.de>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: tidy up status_resync to handle large arrays.
NeilBrown [Thu, 7 May 2009 02:49:35 +0000 (12:49 +1000)]
md: tidy up status_resync to handle large arrays.

Two problems in status_resync.
1/ It still used Kilobytes as the basic block unit, while most code
   now uses sectors uniformly.
2/ It doesn't allow for the possibility that max_sectors exceeds
   the range of "unsigned long".

So
 - change "max_blocks" to "max_sectors", and store sector numbers
   in there and in 'resync'
 - Make 'rt' a 'sector_t' so it can temporarily hold the number of
   remaining sectors.
 - use sector_div rather than normal division.
 - change the magic '100' used to preserve precision to '32'.
   + making it a power of 2 makes division easier
   + it doesn't need to be as large as it was chosen when we averaged
     speed over the entire run.  Now we average speed over the last 30
     seconds or so.

Reported-by: "Mario 'BitKoenig' Holbe" <Mario.Holbe@TU-Ilmenau.DE>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: fix some (more) errors with bitmaps on devices larger than 2TB.
NeilBrown [Thu, 7 May 2009 02:49:06 +0000 (12:49 +1000)]
md: fix some (more) errors with bitmaps on devices larger than 2TB.

If a write intent bitmap covers more than 2TB, we sometimes work with
values beyond 32bit, so these need to be sector_t.  This patches
add the required casts to some unsigned longs that are being shifted
up.

This will affect any raid10 larger than 2TB, or any raid1/4/5/6 with
member devices that are larger than 2TB.

Signed-off-by: NeilBrown <neilb@suse.de>
Reported-by: "Mario 'BitKoenig' Holbe" <Mario.Holbe@TU-Ilmenau.DE>
Cc: stable@kernel.org
15 years agomd/raid10: don't clear bitmap during recovery if array will still be degraded.
NeilBrown [Thu, 7 May 2009 02:48:10 +0000 (12:48 +1000)]
md/raid10: don't clear bitmap during recovery if array will still be degraded.

If we have a raid10 with multiple missing devices, and we recover just
one of these to a spare, then we risk (depending on the bitmap and
array chunk size) clearing bits of the bitmap for which recovery isn't
complete (because a device is still missing).

This can lead to a subsequent "re-add" being recovered without
any IO happening, which would result in loss of data.

This patch takes the safe approach of not clearing bitmap bits
if the array will still be degraded.

This patch is suitable for all active -stable kernels.

Cc: stable@kernel.org
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agomd: fix loading of out-of-date bitmap.
NeilBrown [Thu, 7 May 2009 02:47:19 +0000 (12:47 +1000)]
md: fix loading of out-of-date bitmap.

When md is loading a bitmap which it knows is out of date, it fills
each page with 1s and writes it back out again.  However the
write_page call makes used of bitmap->file_pages and
bitmap->last_page_size which haven't been set correctly yet.  So this
can sometimes fail.

Move the setting of file_pages and last_page_size to before the call
to write_page.

This bug can cause the assembly on an array to fail, thus making the
data inaccessible.  Hence I think it is a suitable candidate for
-stable.

Cc: stable@kernel.org
Reported-by: Vojtech Pavlik <vojtech@suse.cz>
Signed-off-by: NeilBrown <neilb@suse.de>
15 years agodrivers/base/iommu.c: add missing includes
Andrew Morton [Wed, 6 May 2009 23:03:07 +0000 (16:03 -0700)]
drivers/base/iommu.c: add missing includes

Fix zillions of -mm x86_64 allmodconfig build errors - the file uses
EXPORT_SYMBOL() and kmalloc but misses the needed includes.

Cc: Greg Kroah-Hartman <gregkh@suse.de>
Cc: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoinitramfs: clean up messages related to initramfs unpacking
Eric Piel [Wed, 6 May 2009 23:03:06 +0000 (16:03 -0700)]
initramfs: clean up messages related to initramfs unpacking

With the removal of duplicate unpack_to_rootfs() (commit
df52092f3c97788592ef72501a43fb7ac6a3cfe0) the messages displayed do not
actually correspond to what the kernel is doing.  In addition, depending
if ramdisks are supported or not, the messages are not at all the same.

So keep the messages more in sync with what is really doing the kernel,
and only display a second message in case of failure.  This also ensure
that the printk message cannot be split by other printk's.

Signed-off-by: Eric Piel <eric.piel@tremplin-utc.net>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agonommu: make the initial mmap allocation excess behaviour Kconfig configurable
David Howells [Wed, 6 May 2009 23:03:05 +0000 (16:03 -0700)]
nommu: make the initial mmap allocation excess behaviour Kconfig configurable

NOMMU mmap() has an option controlled by a sysctl variable that determines
whether the allocations made by do_mmap_private() should have the excess
space trimmed off and returned to the allocator.  Make the initial setting
of this variable a Kconfig configuration option.

The reason there can be excess space is that the allocator only allocates
in power-of-2 size chunks, but mmap()'s can be made in sizes that aren't a
power of 2.

There are two alternatives:

 (1) Keep the excess as dead space.  The dead space then remains unused for the
     lifetime of the mapping.  Mappings of shared objects such as libc, ld.so
     or busybox's text segment may retain their dead space forever.

 (2) Return the excess to the allocator.  This means that the dead space is
     limited to less than a page per mapping, but it means that for a transient
     process, there's more chance of fragmentation as the excess space may be
     reused fairly quickly.

During the boot process, a lot of transient processes are created, and
this can cause a lot of fragmentation as the pagecache and various slabs
grow greatly during this time.

By turning off the trimming of excess space during boot and disabling
batching of frees, Coldfire can manage to boot.

A better way of doing things might be to have /sbin/init turn this option
off.  By that point libc, ld.so and init - which are all long-duration
processes - have all been loaded and trimmed.

Reported-by: Lanttor Guo <lanttor.guo@freescale.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Lanttor Guo <lanttor.guo@freescale.com>
Cc: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agonommu: clamp zone_batchsize() to 0 under NOMMU conditions
David Howells [Wed, 6 May 2009 23:03:03 +0000 (16:03 -0700)]
nommu: clamp zone_batchsize() to 0 under NOMMU conditions

Clamp zone_batchsize() to 0 under NOMMU conditions to stop
free_hot_cold_page() from queueing and batching frees.

The problem is that under NOMMU conditions it is really important to be
able to allocate large contiguous chunks of memory, but when munmap() or
exit_mmap() releases big stretches of memory, return of these to the buddy
allocator can be deferred, and when it does finally happen, it can be in
small chunks.

Whilst the fragmentation this incurs isn't so much of a problem under MMU
conditions as userspace VM is glued together from individual pages with
the aid of the MMU, it is a real problem if there isn't an MMU.

By clamping the page freeing queue size to 0, pages are returned to the
allocator immediately, and the buddy detector is more likely to be able to
glue them together into large chunks immediately, and fragmentation is
less likely to occur.

By disabling batching of frees, and by turning off the trimming of excess
space during boot, Coldfire can manage to boot.

Reported-by: Lanttor Guo <lanttor.guo@freescale.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Lanttor Guo <lanttor.guo@freescale.com>
Cc: Greg Ungerer <gerg@snapgear.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agomm: use roundown_pow_of_two() in zone_batchsize()
David Howells [Wed, 6 May 2009 23:03:02 +0000 (16:03 -0700)]
mm: use roundown_pow_of_two() in zone_batchsize()

Use roundown_pow_of_two(N) in zone_batchsize() rather than (1 <<
(fls(N)-1)) as they are equivalent, and with the former it is easier to
see what is going on.

Signed-off-by: David Howells <dhowells@redhat.com>
Tested-by: Lanttor Guo <lanttor.guo@freescale.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoisl29003: fix resume functionality
Daniel Mack [Wed, 6 May 2009 23:03:00 +0000 (16:03 -0700)]
isl29003: fix resume functionality

The isl29003 does not interpret the return value of
i2c_smbus_write_byte_data() correctly and hence causes an error on system
resume.

Also introduce power_state_before_suspend and restore the chip's power
state upon wakeup.

Signed-off-by: Daniel Mack <daniel@caiaq.de>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Jean Delvare <khali@linux-fr.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agofbdev: remove makefile reference to removed driver
Krzysztof Helt [Wed, 6 May 2009 23:03:00 +0000 (16:03 -0700)]
fbdev: remove makefile reference to removed driver

The cyblafb driver is removed so remove its last trace in the makefile.

Signed-off-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoalloc_vmap_area: fix memory leak
Ralph Wuerthner [Wed, 6 May 2009 23:02:59 +0000 (16:02 -0700)]
alloc_vmap_area: fix memory leak

If alloc_vmap_area() fails the allocated struct vmap_area has to be freed.

Signed-off-by: Ralph Wuerthner <ralphw@linux.vnet.ibm.com>
Reviewed-by: Christoph Lameter <cl@linux-foundation.org>
Reviewed-by: Minchan Kim <minchan.kim@gmail.com>
Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agodoc: small kernel-parameters updates
Randy Dunlap [Wed, 6 May 2009 23:02:58 +0000 (16:02 -0700)]
doc: small kernel-parameters updates

Change last "i386" to X86-32 as is used throughout the rest of the file.
Change combination of X86-32,X86-64 to just X86, as is done throughout the
rest of the file.

Add a note that hyphens and underscores are equivalent in parameter names,
with examples.

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Jan Engelhardt <jengelh@medozas.de>
Cc: Christopher Sylvain <chris.sylvain@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agofbdev: fix fillrect for 24bpp modes
Michal Januszewski [Wed, 6 May 2009 23:02:56 +0000 (16:02 -0700)]
fbdev: fix fillrect for 24bpp modes

The software fillrect routines do not work properly when the number of
pixels per machine word is not an integer.  To see that, run the following
command on a fbdev console with a 24bpp video mode, using a
non-accelerated driver such as (u)vesafb:

  reset ; echo -e '\e[41mtest\e[K'

The expected result is 'test' displayed on a line with red background.
Instead of that, 'test' has a red background, but the rest of the line
(rendered using fillrect()) contains a distored colorful pattern.

This patch fixes the problem by correctly computing rotation shifts.  It
has been tested in a 24bpp mode on 32- and 64-bit little-endian machines.

Signed-off-by: Michal Januszewski <spock@gentoo.org>
Acked-by: Krzysztof Helt <krzysztof.h1@wp.pl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agooom: prevent livelock when oom_kill_allocating_task is set
David Rientjes [Wed, 6 May 2009 23:02:55 +0000 (16:02 -0700)]
oom: prevent livelock when oom_kill_allocating_task is set

When /proc/sys/vm/oom_kill_allocating_task is set for large systems that
want to avoid the lengthy tasklist scan, it's possible to livelock if
current is ineligible for oom kill.  This normally happens when it is set
to OOM_DISABLE, but is also possible if any threads are sharing the same
->mm with a different tgid.

So change __out_of_memory() to fall back to the full task-list scan if it
was unable to kill `current'.

Cc: Nick Piggin <npiggin@suse.de>
Signed-off-by: David Rientjes <rientjes@google.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agofiemap: fix problem with setting FIEMAP_EXTENT_LAST
Josef Bacik [Wed, 6 May 2009 23:02:53 +0000 (16:02 -0700)]
fiemap: fix problem with setting FIEMAP_EXTENT_LAST

Fix a problem where the generic block based fiemap stuff would not
properly set FIEMAP_EXTENT_LAST on the last extent.  I've reworked things
to keep track if we go past the EOF, and mark the last extent properly.
The problem was reported by and tested by Eric Sandeen.

Tested-by: Eric Sandeen <sandeen@redhat.com>
Signed-off-by: Josef Bacik <jbacik@redhat.com>
Cc: <linux-ext4@vger.kernel.org>
Cc: <xfs-masters@oss.sgi.com>
Cc: <linux-btrfs@vger.kernel.org>
Cc: Steven Whitehouse <swhiteho@redhat.com>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <Joel.Becker@oracle.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoEliminate thousands of warnings with gcc 3.2 build
Andi Kleen [Wed, 6 May 2009 23:02:53 +0000 (16:02 -0700)]
Eliminate thousands of warnings with gcc 3.2 build

When building with gcc 3.2 I get thousands of warnings such as

include/linux/gfp.h: In function `allocflags_to_migratetype':
include/linux/gfp.h:105: warning: null format string

due to passing a NULL format string to warn_slowpath() in

#define __WARN() warn_slowpath(__FILE__, __LINE__, NULL)

Split this case out into a separate call.  This also shrinks the kernel
slightly:

          text    data     bss     dec     hex filename
       4802274  707668  712704 6222646  5ef336 vmlinux
          text    data     bss     dec     hex filename
       4799027  703572  712704 6215303  5ed687 vmlinux

due to removeing one argument from the commonly-called __WARN().

[akpm@linux-foundation.org: reduce scope of `empty']
Acked-by: Jesper Nilsson <jesper.nilsson@axis.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Acked-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agodoc: hashdist defaults on for 64bit
Hugh Dickins [Wed, 6 May 2009 23:02:51 +0000 (16:02 -0700)]
doc: hashdist defaults on for 64bit

kernel boot parameter `hashdist' now defaults on for all 64bit NUMA.

Signed-off-by: Hugh Dickins <hugh@veritas.com>
Acked-by: Mel Gorman <mel@csn.ul.ie>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoinotify: use GFP_NOFS in kernel_event() to work around a lockdep false-positive
Wu Fengguang [Wed, 6 May 2009 23:02:50 +0000 (16:02 -0700)]
inotify: use GFP_NOFS in kernel_event() to work around a lockdep false-positive

There is what we believe to be a false positive reported by lockdep.

inotify_inode_queue_event() => take inotify_mutex => kernel_event() =>
kmalloc() => SLOB => alloc_pages_node() => page reclaim => slab reclaim =>
dcache reclaim => inotify_inode_is_dead => take inotify_mutex => deadlock

The plan is to fix this via lockdep annotation, but that is proving to be
quite involved.

The patch flips the allocation over to GFP_NFS to shut the warning up, for
the 2.6.30 release.

Hopefully we will fix this for real in 2.6.31.  I'll queue a patch in -mm
to switch it back to GFP_KERNEL so we don't forget.

  =================================
  [ INFO: inconsistent lock state ]
  2.6.30-rc2-next-20090417 #203
  ---------------------------------
  inconsistent {RECLAIM_FS-ON-W} -> {IN-RECLAIM_FS-W} usage.
  kswapd0/380 [HC0[0]:SC0[0]:HE1:SE1] takes:
   (&inode->inotify_mutex){+.+.?.}, at: [<ffffffff8112f1b5>] inotify_inode_is_dead+0x35/0xb0
  {RECLAIM_FS-ON-W} state was registered at:
    [<ffffffff81079188>] mark_held_locks+0x68/0x90
    [<ffffffff810792a5>] lockdep_trace_alloc+0xf5/0x100
    [<ffffffff810f5261>] __kmalloc_node+0x31/0x1e0
    [<ffffffff81130652>] kernel_event+0xe2/0x190
    [<ffffffff81130826>] inotify_dev_queue_event+0x126/0x230
    [<ffffffff8112f096>] inotify_inode_queue_event+0xc6/0x110
    [<ffffffff8110444d>] vfs_create+0xcd/0x140
    [<ffffffff8110825d>] do_filp_open+0x88d/0xa20
    [<ffffffff810f6b68>] do_sys_open+0x98/0x140
    [<ffffffff810f6c50>] sys_open+0x20/0x30
    [<ffffffff8100c272>] system_call_fastpath+0x16/0x1b
    [<ffffffffffffffff>] 0xffffffffffffffff
  irq event stamp: 690455
  hardirqs last  enabled at (690455): [<ffffffff81564fe4>] _spin_unlock_irqrestore+0x44/0x80
  hardirqs last disabled at (690454): [<ffffffff81565372>] _spin_lock_irqsave+0x32/0xa0
  softirqs last  enabled at (690178): [<ffffffff81052282>] __do_softirq+0x202/0x220
  softirqs last disabled at (690157): [<ffffffff8100d50c>] call_softirq+0x1c/0x50

  other info that might help us debug this:
  2 locks held by kswapd0/380:
   #0:  (shrinker_rwsem){++++..}, at: [<ffffffff810d0bd7>] shrink_slab+0x37/0x180
   #1:  (&type->s_umount_key#17){++++..}, at: [<ffffffff8110cfbf>] shrink_dcache_memory+0x11f/0x1e0

  stack backtrace:
  Pid: 380, comm: kswapd0 Not tainted 2.6.30-rc2-next-20090417 #203
  Call Trace:
   [<ffffffff810789ef>] print_usage_bug+0x19f/0x200
   [<ffffffff81018bff>] ? save_stack_trace+0x2f/0x50
   [<ffffffff81078f0b>] mark_lock+0x4bb/0x6d0
   [<ffffffff810799e0>] ? check_usage_forwards+0x0/0xc0
   [<ffffffff8107b142>] __lock_acquire+0xc62/0x1ae0
   [<ffffffff810f478c>] ? slob_free+0x10c/0x370
   [<ffffffff8107c0a1>] lock_acquire+0xe1/0x120
   [<ffffffff8112f1b5>] ? inotify_inode_is_dead+0x35/0xb0
   [<ffffffff81562d43>] mutex_lock_nested+0x63/0x420
   [<ffffffff8112f1b5>] ? inotify_inode_is_dead+0x35/0xb0
   [<ffffffff8112f1b5>] ? inotify_inode_is_dead+0x35/0xb0
   [<ffffffff81012fe9>] ? sched_clock+0x9/0x10
   [<ffffffff81077165>] ? lock_release_holdtime+0x35/0x1c0
   [<ffffffff8112f1b5>] inotify_inode_is_dead+0x35/0xb0
   [<ffffffff8110c9dc>] dentry_iput+0xbc/0xe0
   [<ffffffff8110cb23>] d_kill+0x33/0x60
   [<ffffffff8110ce23>] __shrink_dcache_sb+0x2d3/0x350
   [<ffffffff8110cffa>] shrink_dcache_memory+0x15a/0x1e0
   [<ffffffff810d0cc5>] shrink_slab+0x125/0x180
   [<ffffffff810d1540>] kswapd+0x560/0x7a0
   [<ffffffff810ce160>] ? isolate_pages_global+0x0/0x2c0
   [<ffffffff81065a30>] ? autoremove_wake_function+0x0/0x40
   [<ffffffff8107953d>] ? trace_hardirqs_on+0xd/0x10
   [<ffffffff810d0fe0>] ? kswapd+0x0/0x7a0
   [<ffffffff8106555b>] kthread+0x5b/0xa0
   [<ffffffff8100d40a>] child_rip+0xa/0x20
   [<ffffffff8100cdd0>] ? restore_args+0x0/0x30
   [<ffffffff81065500>] ? kthread+0x0/0xa0
   [<ffffffff8100d400>] ? child_rip+0x0/0x20

[eparis@redhat.com: fix audit too]
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Matt Mackall <mpm@selenic.com>
Cc: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Wu Fengguang <fengguang.wu@intel.com>
Signed-off-by: Eric Paris <eparis@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agojsm: removing unused spinlock
Breno Leitao [Wed, 6 May 2009 16:17:57 +0000 (17:17 +0100)]
jsm: removing unused spinlock

This patch removes bd_lock spinlock (inside jsm_board structure).
The lock is initialized in the probe function and not used anymore.

Signed-off-by: Breno Leitao <leitao@linux.vnet.ibm.com>
Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agovt: Add a note on the historical abuse of CLOCK_TICK_RATE
Alan Cox [Wed, 6 May 2009 16:17:26 +0000 (17:17 +0100)]
vt: Add a note on the historical abuse of CLOCK_TICK_RATE

This is one area where we can't just magic away the bizarre use of
CLOCK_TICK_RATE as it leaks to user space APIs. It also means the visible
CLOCK_TICK_RATE is frozen for architectures which is horrible.

We need to fix this somehow

Signed-off-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agox86: Fix a typo in a printk message
Nikanth Karthikesan [Mon, 4 May 2009 03:38:26 +0000 (09:08 +0530)]
x86: Fix a typo in a printk message

[ Impact: printk message cleanup ]

Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
LKML-Reference: <200905040908.27299.knikanth@suse.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, srat: do not register nodes beyond e820 map
David Rientjes [Tue, 5 May 2009 19:50:02 +0000 (12:50 -0700)]
x86, srat: do not register nodes beyond e820 map

The mem= option will truncate the memory map at a specified address so
it's not possible to register nodes with memory beyond the e820 upper
bound.

unparse_node() is only called when then node had memory associated with
it, although with the mem= option it is no longer addressable.

[ Impact: fix boot hang on certain (large) systems ]

Reported-by: "Zhang, Yanmin" <yanmin_zhang@linux.intel.com>
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Jack Steiner <steiner@sgi.com>
LKML-Reference: <alpine.DEB.2.00.0905051248150.20021@chino.kir.corp.google.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years ago[CIFS] Allow raw ntlmssp code to be enabled with sec=ntlmssp
Steve French [Wed, 6 May 2009 04:16:04 +0000 (04:16 +0000)]
[CIFS] Allow raw ntlmssp code to be enabled with sec=ntlmssp

On mount, "sec=ntlmssp" can now be specified to allow
"rawntlmssp" security to be enabled during
CIFS session establishment/authentication (ntlmssp used to
require specifying krb5 which was counterintuitive).

Signed-off-by: Steve French <sfrench@us.ibm.com>
15 years ago[CIFS] Fix SMB uid in NTLMSSP authenticate request
Steve French [Wed, 6 May 2009 00:48:30 +0000 (00:48 +0000)]
[CIFS] Fix SMB uid in NTLMSSP authenticate request

We were not setting the SMB uid in NTLMSSP authenticate
request which could lead to INVALID_PARAMETER error
on 2nd session setup.

Signed-off-by: Steve French <sfrench@us.ibm.com>
15 years agoMerge branch 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Wed, 6 May 2009 00:02:05 +0000 (17:02 -0700)]
Merge branch 'drm-linus' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6:
  drm/r128: fix r128 ioremaps to use ioremap_wc.
  drm: cleanup properly in drm_get_dev() failure paths
  drm: clean the map list before destroying the hash table
  drm: remove unreachable code in drm_sysfs.c
  drm: add control node checks missing from kms merge
  drm/kms: don't try to shortcut drm mode set function
  drm/radeon: bump minor version for occlusion queries support

15 years agodrm/r128: fix r128 ioremaps to use ioremap_wc.
Dave Airlie [Tue, 5 May 2009 23:04:52 +0000 (09:04 +1000)]
drm/r128: fix r128 ioremaps to use ioremap_wc.

This should allow r128 to start working again since PAT changes.

taken from F-11 kernel.

Signed-off-by: Dave Airlie <airlied@redhat.com>
15 years agoMerge branch 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6
Linus Torvalds [Tue, 5 May 2009 22:48:03 +0000 (15:48 -0700)]
Merge branch 'release' of git://git./linux/kernel/git/aegl/linux-2.6

* 'release' of git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux-2.6:
  [IA64] xen_domu_defconfig: fix build issues/warnings

15 years agoIgnore madvise(MADV_WILLNEED) for hugetlbfs-backed regions
Mel Gorman [Tue, 5 May 2009 15:37:17 +0000 (16:37 +0100)]
Ignore madvise(MADV_WILLNEED) for hugetlbfs-backed regions

madvise(MADV_WILLNEED) forces page cache readahead on a range of memory
backed by a file.  The assumption is made that the page required is
order-0 and "normal" page cache.

On hugetlbfs, this assumption is not true and order-0 pages are
allocated and inserted into the hugetlbfs page cache.  This leaks
hugetlbfs page reservations and can cause BUGs to trigger related to
corrupted page tables.

This patch causes MADV_WILLNEED to be ignored for hugetlbfs-backed
regions.

Signed-off-by: Mel Gorman <mel@csn.ul.ie>
Cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'timers/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip...
Linus Torvalds [Tue, 5 May 2009 19:09:38 +0000 (12:09 -0700)]
Merge branch 'timers/urgent' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'timers/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  clockevents: prevent endless loop in tick_handle_periodic()

15 years agoMerge branch 'irq/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux...
Linus Torvalds [Tue, 5 May 2009 19:09:27 +0000 (12:09 -0700)]
Merge branch 'irq/urgent' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'irq/urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  Revert "genirq: assert that irq handlers are indeed running in hardirq context"

15 years agoMerge branch 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 5 May 2009 19:08:40 +0000 (12:08 -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: account system time properly

15 years agoMerge branch 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 5 May 2009 19:08:20 +0000 (12:08 -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:
  kernel/posix-cpu-timers.c: fix sparse warning
  dma-debug: remove broken dma memory leak detection for 2.6.30
  locking: Documentation: lockdep-design.txt, fix note of state bits

15 years agoMerge branch 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 5 May 2009 19:08:02 +0000 (12:08 -0700)]
Merge branch 'tracing-fixes-for-linus' of git://git./linux/kernel/git/tip/linux-2.6-tip

* 'tracing-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
  tracing: x86, mmiotrace: fix range test
  tracing: fix ref count in splice pages

15 years agoMerge branch 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Tue, 5 May 2009 19:07:21 +0000 (12:07 -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: show number of core_siblings instead of thread_siblings in /proc/cpuinfo
  amd-iommu: fix iommu flag masks
  x86: initialize io_bitmap_base on 32bit
  x86: gettimeofday() vDSO: fix segfault when tv == NULL

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes
Linus Torvalds [Tue, 5 May 2009 19:06:54 +0000 (12:06 -0700)]
Merge git://git./linux/kernel/git/sam/kbuild-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes:
  kbuild, modpost: fix unexpected non-allocatable warning with mips
  kbuild, modpost: fix "unexpected non-allocatable" warning with SUSE gcc
  kbuild, modpost: fix unexpected non-allocatable section when cross compiling

15 years ago[IA64] xen_domu_defconfig: fix build issues/warnings
Jan Beulich [Tue, 5 May 2009 12:57:52 +0000 (13:57 +0100)]
[IA64] xen_domu_defconfig: fix build issues/warnings

- drivers/xen/events.c did not compile
- xen_setup_hook caused a modpost section warning
- the use of u64 (instead of unsigned long long) together with a %llu
  in drivers/xen/balloon.c caused a compiler warning

Signed-off-by: Jan Beulich <jbeulich@novell.com>
Signed-off-by: Tony Luck <tony.luck@intel.com>
15 years agoMerge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6
Linus Torvalds [Tue, 5 May 2009 15:27:14 +0000 (08:27 -0700)]
Merge branch 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6

* 'i2c-for-linus' of git://jdelvare.pck.nerim.net/jdelvare-2.6:
  i2c-algo-pca: Let PCA9564 recover from unacked data byte (state 0x30)
  i2c-algo-bit: Fix timeout test
  i2c: Timeouts off by 1

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6
Linus Torvalds [Tue, 5 May 2009 15:26:10 +0000 (08:26 -0700)]
Merge git://git./linux/kernel/git/davem/net-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (30 commits)
  e1000: fix virtualization bug
  bonding: fix alb mode locking regression
  Bluetooth: Fix issue with sysfs handling for connections
  usbnet: CDC EEM support (v5)
  tcp: Fix tcp_prequeue() to get correct rto_min value
  ehea: fix invalid pointer access
  ne2k-pci: Do not register device until initialized.
  Subject: [PATCH] br2684: restore net_dev initialization
  net: Only store high 16 bits of kernel generated filter priorities
  virtio_net: Fix function name typo
  virtio_net: Cleanup command queue scatterlist usage
  bonding: correct the cleanup in bond_create()
  virtio: add missing include to virtio_net.h
  smsc95xx: add support for LAN9512 and LAN9514
  smsc95xx: configure LED outputs
  netconsole: take care of NETDEV_UNREGISTER event
  xt_socket: checks for the state of nf_conntrack
  bonding: bond_slave_info_query() fix
  cxgb3: fixing gcc 4.4 compiler warning: suggest parentheses around operand of ‘!’
  netfilter: use likely() in xt_info_rdlock_bh()
  ...