openwrt/staging/blogic.git
15 years agox86, ioapic: Document another case when level irq is seen as an edge
Suresh Siddha [Tue, 1 Dec 2009 23:31:17 +0000 (15:31 -0800)]
x86, ioapic: Document another case when level irq is seen as an edge

In the case when cpu goes offline, fixup_irqs() will forward any
unhandled interrupt on the offlined cpu to the new cpu
destination that is handling the corresponding interrupt. This
interrupt forwarding is done via IPI's. Hence, in this case also
level-triggered io-apic interrupt will be seen as an edge
interrupt in the cpu's APIC IRR.

Document this scenario in the code which handles this case by doing
an explicit EOI to the io-apic to clear remote IRR of the io-apic RTE.

Requested-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
Cc: ebiederm@xmission.com
Cc: garyhade@us.ibm.com
LKML-Reference: <20091201233335.143970505@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, ioapic: Fix the EOI register detection mechanism
Suresh Siddha [Tue, 1 Dec 2009 23:31:16 +0000 (15:31 -0800)]
x86, ioapic: Fix the EOI register detection mechanism

Maciej W. Rozycki reported:

82093AA I/O APIC has its version set to 0x11 and it
> does not support the EOI register.  Similarly I/O APICs
> integrated into the 82379AB south bridge and the 82374EB/SB
> EISA component.

IO-APIC versions below 0x20 don't support EOI register.

Some of the Intel ICH Specs (ICH2 to ICH5) documents the io-apic
version as 0x2. This is an error with documentation and these
ICH chips use io-apic's of version 0x20 and indeed has a working
EOI register for the io-apic.

Fix the EOI register detection mechanism to check for version
0x20 and beyond.

And also, a platform can potentially  have io-apic's with
different versions. Make the EOI register check per io-apic.

Reported-by: Maciej W. Rozycki <macro@linux-mips.org>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: ebiederm@xmission.com
Cc: garyhade@us.ibm.com
LKML-Reference: <20091201233335.065361533@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, io-apic: Move the effort of clearing remoteIRR explicitly before migrating the irq
Maciej W. Rozycki [Tue, 1 Dec 2009 23:31:15 +0000 (15:31 -0800)]
x86, io-apic: Move the effort of clearing remoteIRR explicitly before migrating the irq

When the level-triggered interrupt is seen as an edge interrupt,
we try to clear the remoteIRR explicitly (using either an
io-apic eoi register when present or through the idea of
changing trigger mode of the io-apic RTE to edge and then back
to level). But this explicit try also needs to happen before we
try to migrate the irq. Otherwise irq migration attempt will
fail anyhow, as it postpones the irq migration to a later
attempt when it sees the remoteIRR in the io-apic RTE still set.

Signed-off-by: "Maciej W. Rozycki" <macro@linux-mips.org>
Reviewed-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: ebiederm@xmission.com
Cc: garyhade@us.ibm.com
LKML-Reference: <20091201233334.975416130@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: SGI UV: Map low MMR ranges
Jack Steiner [Wed, 25 Nov 2009 16:20:19 +0000 (10:20 -0600)]
x86: SGI UV: Map low MMR ranges

Explicitly mmap the UV chipset MMR address ranges used to
access blade-local registers. Although these same MMRs are also
mmaped at higher addresses, the low range is more
convenient when accessing blade-local registers.

The low range addresses always alias to the local blade
regardless of the blade id.

Signed-off-by: Jack Steiner <steiner@sgi.com>
LKML-Reference: <20091125162018.GA25445@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: apic: Print out SRAT table APIC id in hex
Yinghai Lu [Sat, 21 Nov 2009 08:23:37 +0000 (00:23 -0800)]
x86: apic: Print out SRAT table APIC id in hex

Make it consistent with APIC MADT print out,
for big systems APIC id in hex is more readable.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4B07A739.3030104@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Re-get cfg_new in case reuse/move irq_desc
Yinghai Lu [Sat, 21 Nov 2009 08:23:37 +0000 (00:23 -0800)]
x86: Re-get cfg_new in case reuse/move irq_desc

When irq_desc is moved, we need to make sure to use the right cfg_new.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4B07A739.3030104@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: apic: Remove not needed #ifdef
Yinghai Lu [Sat, 21 Nov 2009 08:23:37 +0000 (00:23 -0800)]
x86: apic: Remove not needed #ifdef

Suresh made dmar_table_init() already have that protection.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <4B07A739.3030104@kernel.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: io-apic: IO-APIC MMIO should not fail on resource insertion
Cyrill Gorcunov [Mon, 16 Nov 2009 15:14:26 +0000 (18:14 +0300)]
x86: io-apic: IO-APIC MMIO should not fail on resource insertion

If IO-APIC base address is 1K aligned we should not fail
on resourse insertion procedure. For this sake we define
IO_APIC_SLOT_SIZE constant which should cover all IO-APIC
direct accessible registers.

An example of a such configuration is there

http://marc.info/?l=linux-kernel&m=118114792006520

 |
 | Quoting the message
 |
 | IOAPIC[0]: apic_id 2, version 32, address 0xfec00000, GSI 0-23
 | IOAPIC[1]: apic_id 3, version 32, address 0xfec80000, GSI 24-47
 | IOAPIC[2]: apic_id 4, version 32, address 0xfec80400, GSI 48-71
 | IOAPIC[3]: apic_id 5, version 32, address 0xfec84000, GSI 72-95
 | IOAPIC[4]: apic_id 8, version 32, address 0xfec84400, GSI 96-119
 |

Reported-by: "Maciej W. Rozycki" <macro@linux-mips.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <20091116151426.GC5653@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Remove asm/apicnum.h
Dimitri Sivanich [Tue, 10 Nov 2009 19:58:35 +0000 (13:58 -0600)]
x86: Remove asm/apicnum.h

arch/x86/include/asm/apicnum.h is not referenced anywhere
anymore. Its definitions appear in apicdef.h. Remove it.

Signed-off-by: Dimitri Sivanich <sivanich@sgi.com>
Acked-by: Cyrill Gorcunov <gorcunov@openvz.org>
Acked-by: Mike Travis <travis@sgi.com>
LKML-Reference: <20091110195835.GA4393@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: apic: Do not use stacked physid_mask_t
Cyrill Gorcunov [Mon, 9 Nov 2009 22:06:59 +0000 (01:06 +0300)]
x86: apic: Do not use stacked physid_mask_t

We should not use physid_mask_t as a stack based
variable in apic code. This type depends on MAX_APICS
parameter which may be huge enough.

Especially it became a problem with apic NOOP driver which
is portable between 32 bit and 64 bit environment
(where we have really huge MAX_APICS).

So apic driver should operate with pointers and a caller
in turn should aware of allocation physid_mask_t variable.

As a side (but positive) effect -- we may use already
implemented physid_set_mask_of_physid function eliminating
default_apicid_to_cpu_present completely.

Note that physids_coerce and physids_promote turned into static
inline from macro (since macro hides the fact that parameter is
being interpreted as unsigned long, make it explicit).

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
LKML-Reference: <20091109220659.GA5568@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Get rid of apicid_to_cpu_present assign on 64-bit
Cyrill Gorcunov [Sun, 8 Nov 2009 13:16:45 +0000 (16:16 +0300)]
x86, apic: Get rid of apicid_to_cpu_present assign on 64-bit

In fact it's never get used on x86-64 (for 64 bit platform
we use differ technique to enumerate io-units).

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Peter Zijlstra <peterz@infradead.org>
LKML-Reference: <20091108131645.GD5300@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, ioapic: Use snrpintf while set names for IO-APIC resourses
Cyrill Gorcunov [Sun, 8 Nov 2009 15:54:31 +0000 (18:54 +0300)]
x86, ioapic: Use snrpintf while set names for IO-APIC resourses

We should be ready that one day MAX_IO_APICS may raise its
number. To prevent memory overwrite we're to use safe
snprintf while set IO-APIC resourse name.

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <20091108155431.GC25940@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Use PAGE_SIZE instead of numbers
Cyrill Gorcunov [Sun, 8 Nov 2009 15:53:56 +0000 (18:53 +0300)]
x86, apic: Use PAGE_SIZE instead of numbers

The whole page is reserved for IO-APIC fixmap
due to non-cacheable requirement. So lets note
this explicitly instead of playing with numbers.

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
LKML-Reference: <20091108155356.GB25940@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Remove local_irq_enable()/local_irq_disable() in fixup_irqs()
Suresh Siddha [Mon, 26 Oct 2009 22:24:36 +0000 (14:24 -0800)]
x86: Remove local_irq_enable()/local_irq_disable() in fixup_irqs()

To ensure that we handle all the pending interrupts (destined
for this cpu that is going down) in the interrupt subsystem
before the cpu goes offline, fixup_irqs() does:

local_irq_enable();
mdelay(1);
local_irq_disable();

Enabling interrupts is not a good thing as this cpu is already
offline. So this patch replaces that logic with,

mdelay(1);
check APIC_IRR bits
Retrigger the irq at the new destination if any interrupt has arrived
via IPI.

For IO-APIC level triggered interrupts, this retrigger IPI will
appear as an edge interrupt. ack_apic_level() will detect this
condition and IO-APIC RTE's remoteIRR is cleared using directed
EOI(using IO-APIC EOI register) on Intel platforms and for
others it uses the existing mask+edge logic followed by
unmask+level.

We can also remove mdelay() and then send spuriuous interrupts
to new cpu targets for all the irqs that were handled previously
by this cpu that is going offline. While it works, I have seen
spurious interrupt messages (nothing wrong but still annoying
messages during cpu offline, which can be seen during
suspend/resume etc)

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Gary Hade <garyhade@us.ibm.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
LKML-Reference: <20091026230002.043281924@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Use EOI register in io-apic on intel platforms
Suresh Siddha [Mon, 26 Oct 2009 22:24:35 +0000 (14:24 -0800)]
x86: Use EOI register in io-apic on intel platforms

IO-APIC's in intel chipsets support EOI register starting from
IO-APIC version 2. Use that when ever we need to clear the
IO-APIC RTE's RemoteIRR bit explicitly.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Gary Hade <garyhade@us.ibm.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
LKML-Reference: <20091026230001.947855317@sbs-t61.sc.intel.com>
[ Marked use_eio_reg as __read_mostly, fixed small details ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Force irq complete move during cpu offline
Suresh Siddha [Mon, 26 Oct 2009 22:24:34 +0000 (14:24 -0800)]
x86: Force irq complete move during cpu offline

When a cpu goes offline, fixup_irqs() try to move irq's
currently destined to the offline cpu to a new cpu. But this
attempt will fail if the irq is recently moved to this cpu and
the irq still hasn't arrived at this cpu (for non intr-remapping
platforms this is when we free the vector allocation at the
previous destination) that is about to go offline.

This will endup with the interrupt subsystem still pointing the
irq to the offline cpu, causing that irq to not work any more.

Fix this by forcing the irq to complete its move (its been a
long time we moved the irq to this cpu which we are offlining
now) and then move this irq to a new cpu before this cpu goes
offline.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Gary Hade <garyhade@us.ibm.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
LKML-Reference: <20091026230001.848830905@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Remove move_cleanup_count from irq_cfg
Suresh Siddha [Mon, 26 Oct 2009 22:24:33 +0000 (14:24 -0800)]
x86: Remove move_cleanup_count from irq_cfg

move_cleanup_count for each irq in irq_cfg is keeping track of
the total number of cpus that need to free the corresponding
vectors associated with the irq which has now been migrated to
new destination. As long as this move_cleanup_count is non-zero
(i.e., as long as we have n't freed the vector allocations on
the old destinations) we were preventing the irq's further
migration.

This cleanup count is unnecessary and it is enough to not allow
the irq migration till we send the cleanup vector to the
previous irq destination, for which we already have irq_cfg's
move_in_progress.  All we need to make sure is that we free the
vector at the old desintation but we don't need to wait till
that gets freed.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Gary Hade <garyhade@us.ibm.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
LKML-Reference: <20091026230001.752968906@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, intr-remap: Avoid irq_chip mask/unmask in fixup_irqs() for intr-remapping
Suresh Siddha [Mon, 26 Oct 2009 22:24:32 +0000 (14:24 -0800)]
x86, intr-remap: Avoid irq_chip mask/unmask in fixup_irqs() for intr-remapping

In the presence of interrupt-remapping, irqs will be migrated in
the process context and we don't do (and there is no need to)
irq_chip mask/unmask while migrating the interrupt.

Similarly fix the fixup_irqs() that get called during cpu
offline and avoid calling irq_chip mask/unmask for irqs that are
ok to be migrated in the process context.

While we didn't observe any race condition with the existing
code, this change takes complete advantage of
interrupt-remapping in the newer generation platforms and avoids
any potential HW lockup's (that often worry Eric :)

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Acked-by: Eric W. Biederman <ebiederm@xmission.com>
Cc: garyhade@us.ibm.com
LKML-Reference: <20091026230001.661423939@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: Unify fixup_irqs() for 32-bit and 64-bit kernels
Suresh Siddha [Mon, 26 Oct 2009 22:24:31 +0000 (14:24 -0800)]
x86: Unify fixup_irqs() for 32-bit and 64-bit kernels

There is no reason to have different fixup_irqs() for 32-bit and
64-bit kernels. Unify by using the superior 64-bit version for
both the kernels.

Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Gary Hade <garyhade@us.ibm.com>
Cc: Eric W. Biederman <ebiederm@xmission.com>
LKML-Reference: <20091026230001.562512739@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Clear APIC Timer Initial Count Register on shutdown
Andreas Herrmann [Tue, 27 Oct 2009 10:01:38 +0000 (11:01 +0100)]
x86, apic: Clear APIC Timer Initial Count Register on shutdown

Commit a98f8fd24fb24fcb9a359553e64dd6aac5cf4279 (x86: apic reset
counter on shutdown) set the counter to max to avoid spurious
interrupts when the timer is re-enabled.

(In theory) you'll still get a spurious interrupt if spending
more than 344 seconds with this interrupt disabled and then
unmasking it.

The right thing to do is to clear the register. This disables
the interrupt from happening (at least it does on AMD hardware).

Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20091027100138.GB30802@alberich.amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: apic: Allow noop operations to be called almost at any time
Cyrill Gorcunov [Thu, 15 Oct 2009 15:04:16 +0000 (19:04 +0400)]
x86: apic: Allow noop operations to be called almost at any time

As only apic noop is used we allow to use almost any operation
caller wants (and which of them noop driver supports of
course).

Initially it was reported by Ingo Molnar that apic noop
issue a warning for pkg id (which is actually false positive
and should be eliminated).

So we save checking (and warning issue) for read/write
operations while allow any other ops to be freely used.

Also:
 - fix noop_cpu_to_logical_apicid, it should be 0.
 - rename noop_default_phys_pkg_id to noop_phys_pkg_id
   (we use default_ prefix for more general routines
    in apic subsystem).

Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Maciej W. Rozycki <macro@linux-mips.org>
LKML-Reference: <20091015150416.GC5331@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Explain show_lapic= in kernel parameters list
Cyrill Gorcunov [Wed, 14 Oct 2009 15:09:04 +0000 (19:09 +0400)]
x86, apic: Explain show_lapic= in kernel parameters list

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: yinghai@kernel.org
Cc: macro@linux-mips.org
LKML-Reference: <20091014150904.GA5259@lenovo>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Fix prototype in hw_irq.h
Ingo Molnar [Wed, 14 Oct 2009 13:06:42 +0000 (15:06 +0200)]
x86, apic: Fix prototype in hw_irq.h

This warning:

 In file included from arch/x86/include/asm/ipi.h:23,
                  from arch/x86/kernel/apic/apic_noop.c:27:
 arch/x86/include/asm/hw_irq.h:105: warning: â€˜struct irq_desc’ declared inside parameter list
 arch/x86/include/asm/hw_irq.h:105: warning: its scope is only this definition or declaration, which is probably not what you want

triggers because irq_desc is defined after hw_irq.h is included
in irq.h. Since it's pointer reference only, a forward declaration
of the type will solve the problem.

LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Move SGI UV functionality out of generic IO-APIC code
Dimitri Sivanich [Tue, 13 Oct 2009 20:32:36 +0000 (15:32 -0500)]
x86, apic: Move SGI UV functionality out of generic IO-APIC code

Move UV specific functionality out of the generic IO-APIC code.

Signed-off-by: Dimitri Sivanich <sivanich@sgi.com>
LKML-Reference: <20091013203236.GD20543@sgi.com>
[ Cleaned up the code some more in their new places. ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86: SGI UV: Fix irq affinity for hub based interrupts
Dimitri Sivanich [Wed, 30 Sep 2009 16:02:59 +0000 (11:02 -0500)]
x86: SGI UV: Fix irq affinity for hub based interrupts

This patch fixes handling of uv hub irq affinity.  IRQs with ALL or
NODE affinity can be routed to cpus other than their originally
assigned cpu.  Those with CPU affinity cannot be rerouted.

Signed-off-by: Dimitri Sivanich <sivanich@sgi.com>
LKML-Reference: <20090930160259.GA7822@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Limit apic dumping, introduce new show_lapic= setup option
Cyrill Gorcunov [Tue, 13 Oct 2009 20:07:05 +0000 (00:07 +0400)]
x86, apic: Limit apic dumping, introduce new show_lapic= setup option

In case if a system has a large number of cpus printing apics
contents may consume a long time period.

We limit such an output by 1 apic by default. But to have an
ability to see all apics or some part of them we introduce
"show_lapic" setup option which allow us to limit/unlimit the
number of APICs being dumped.

Example: apic=debug show_lapic=5, or apic=debug show_lapic=all

Also move apic_verbosity checking upper that way so helper routines
do not need to inspect it at all.

Suggested-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: yinghai@kernel.org
Cc: macro@linux-mips.org
LKML-Reference: <20091013201022.926793122@openvz.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Use apic noop driver
Cyrill Gorcunov [Tue, 13 Oct 2009 20:07:04 +0000 (00:07 +0400)]
x86, apic: Use apic noop driver

In case if apic were disabled we may use the whole apic NOOP driver
instead of sparse poking the some functions in apic driver.

Also NOOP would catch any inappropriate apic operation calls (not
just read/write).

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: yinghai@kernel.org
Cc: macro@linux-mips.org
LKML-Reference: <20091013201022.747817361@openvz.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agox86, apic: Introduce the NOOP apic driver
Cyrill Gorcunov [Tue, 13 Oct 2009 20:07:03 +0000 (00:07 +0400)]
x86, apic: Introduce the NOOP apic driver

Introduce NOOP APIC driver. We should use it in case if apic was
disabled due to hardware of software/firmware problems (including
user requested to disable it case).

The driver is attempting to catch any inappropriate apic operation
call with warning issue.

Also it is possible to use some apic operation like IPI calls,
read/write without checking for apic presence which should make
callers code easier.

Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: yinghai@kernel.org
Cc: macro@linux-mips.org
LKML-Reference: <20091013201022.534682104@openvz.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
15 years agoMerge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block
Linus Torvalds [Tue, 13 Oct 2009 17:21:33 +0000 (10:21 -0700)]
Merge branch 'for-linus' of git://git.kernel.dk/linux-2.6-block

* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
  cciss: Add cciss_allow_hpsa module parameter
  cciss: Fix multiple calls to pci_release_regions
  blk-settings: fix function parameter kernel-doc notation
  writeback: kill space in debugfs item name
  writeback: account IO throttling wait as iowait
  elv_iosched_store(): fix strstrip() misuse
  cfq-iosched: avoid probable slice overrun when idling
  cfq-iosched: apply bool value where we return 0/1
  cfq-iosched: fix think time allowed for seekers
  cfq-iosched: fix the slice residual sign
  cfq-iosched: abstract out the 'may this cfqq dispatch' logic
  block: use proper BLK_RW_ASYNC in blk_queue_start_tag()
  block: Seperate read and write statistics of in_flight requests v2
  block: get rid of kblock_schedule_delayed_work()
  cfq-iosched: fix possible problem with jiffies wraparound
  cfq-iosched: fix issue with rq-rq merging and fifo list ordering

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu
Linus Torvalds [Tue, 13 Oct 2009 17:21:12 +0000 (10:21 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/tj/percpu

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/percpu:
  percpu: fix compile warnings

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes
Linus Torvalds [Tue, 13 Oct 2009 17:20:53 +0000 (10:20 -0700)]
Merge git://git./linux/kernel/git/sam/kbuild-fixes

* git://git.kernel.org/pub/scm/linux/kernel/git/sam/kbuild-fixes:
  kbuild: revert "save ARCH & CROSS_COMPILE ..."
  warn about use of uninstalled kernel headers
  kbuild: mkcompile_h: trivial cleanups
  kbuild: fix warning when domainname is not available
  kbuild: Fix size_append issue for bzip2/lzma kernel
  kbuild,scripts: use non-builtin echo for '-e'
  kbuild: fix the binrpm-pkg target to work with KBUILD_OUTPUT set

15 years agoheaders: Fix build after <linux/sched.h> removal
Ingo Molnar [Mon, 12 Oct 2009 14:22:46 +0000 (16:22 +0200)]
headers: Fix build after <linux/sched.h> removal

Commit d43c36dc6b357fa1806800f18aa30123c747a6d1 ("headers: remove
sched.h from interrupt.h") left some build errors in some configurations
due to drivers having depended on getting header files "accidentally".

Signed-off-by: Ingo Molnar <mingo@elte.hu>
[ Combined several one-liners from Ingo into one single patch  - Linus ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid
Linus Torvalds [Tue, 13 Oct 2009 17:10:33 +0000 (10:10 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/jikos/hid

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jikos/hid:
  HID: fix possible deadlock in hidraw_read
  HID: fix kerneldoc comment for hid_input_report()
  HID: add __init/__exit macros to twinhan.c

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound-2.6
Linus Torvalds [Tue, 13 Oct 2009 17:10:18 +0000 (10:10 -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 - Allow all formats as default for Nvidia HDMI
  ALSA: aaci: ARM1176 aaci-pl041 AC97 register read timeout
  ALSA: hda - Fix volume-knob setup for Dell laptops with STAC9228
  ALSA: hda - Fix mute sound with STAC9227/9228 codecs
  ALSA: bt87x - Add a whitelist for Pinnacle PCTV (11bd:0012)
  ALSA: hda - Fix overflow of spec->init_verbs in patch_realtek.c
  ALSA: ice1724 - Make call to set hw params succeed on ESI Juli@
  ALSA: ice1724: Fix surround on Chaintech AV-710
  ALSA: hda - Add full rates/formats support for Nvidia HDMI

15 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Tue, 13 Oct 2009 17:05:36 +0000 (10:05 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6:
  ext3: Update documentation about ext3 quota mount options
  ext3: Don't update superblock write time when filesystem is read-only

15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sameo/mfd-2.6
Linus Torvalds [Tue, 13 Oct 2009 17:05:04 +0000 (10:05 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/sameo/mfd-2.6

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/sameo/mfd-2.6:
  mfd: Fix twl4030 boot with twl4030 usb transceiver enabled

15 years agoMerge git://git.infradead.org/~dwmw2/iommu-2.6.32
Linus Torvalds [Tue, 13 Oct 2009 17:04:40 +0000 (10:04 -0700)]
Merge git://git.infradead.org/~dwmw2/iommu-2.6.32

* git://git.infradead.org/~dwmw2/iommu-2.6.32:
  x86: Move pci_iommu_init to rootfs_initcall()
  Run pci_apply_final_quirks() sooner.
  Mark pci_apply_final_quirks() __init rather than __devinit
  Rename pci_init() to pci_apply_final_quirks(), move it to quirks.c
  intel-iommu: Yet another BIOS workaround: Isoch DMAR unit with no TLB space
  intel-iommu: Decode (and ignore) RHSA entries
  intel-iommu: Make "Unknown DMAR structure" message more informative

15 years agoMerge branch 'fix/hda' into for-linus
Takashi Iwai [Tue, 13 Oct 2009 14:09:56 +0000 (16:09 +0200)]
Merge branch 'fix/hda' into for-linus

15 years agoALSA: hda - Allow all formats as default for Nvidia HDMI
Takashi Iwai [Tue, 13 Oct 2009 14:07:59 +0000 (16:07 +0200)]
ALSA: hda - Allow all formats as default for Nvidia HDMI

In the commit f0613d5752d8f7d1d02e6d40947f38877fdf9c90
    ALSA: hda - Add full rates/formats support for Nvidia HDMI
the flag LIMITIED_RATE_FMT_SUPPORT was set as default, as I forgot
to clear before commit.

Let's enable all formats/rates as default.

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: aaci: ARM1176 aaci-pl041 AC97 register read timeout
Philby John [Tue, 13 Oct 2009 11:00:22 +0000 (16:30 +0530)]
ALSA: aaci: ARM1176 aaci-pl041 AC97 register read timeout

After a reboot on an ARM1176 which amounts to a softreset, it has been
noted that the ALSA driver does not get registered and the probe fails
with the error "aaci-pl041 fpga:04: ac97 read back fail". In the process
of reading from a register the SL1TxBusy bit is set indicating that the
transceiver is busy and remains so until the default timeout occurs.
Set the Power down register 0x26 to an arbitrary value as specified in
the PL041 manual (page: 3-18) so that AACISL1TX/AACISL2TX registers take
their default state.

Signed-off-by: Philby John <pjohn@in.mvista.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: hda - Fix volume-knob setup for Dell laptops with STAC9228
Takashi Iwai [Tue, 13 Oct 2009 13:32:21 +0000 (15:32 +0200)]
ALSA: hda - Fix volume-knob setup for Dell laptops with STAC9228

The volume-knob widget needs to be set with 0x7f instead of 0xff
for Dell laptops with STAC9228 codec, too, like the previous commit.

Reference: Novell bnc#545013
http://bugzilla.novell.com/show_bug.cgi?id=545013

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: hda - Fix mute sound with STAC9227/9228 codecs
Takashi Iwai [Sun, 11 Oct 2009 15:38:29 +0000 (17:38 +0200)]
ALSA: hda - Fix mute sound with STAC9227/9228 codecs

On FSC laptops, the sound gets muted gradually when the volume is chnaged.
This is due to the wrong volume-knob widget setup.  The delta bit (bit 7)
shouldn't be set for these devices.

This patch adds a new quirk to set the value 0x7f to the widget 0x24
instead of 0xff.

Reference: Novell bnc#546006
http://bugzilla.novell.com/show_bug.cgi?id=546006

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoALSA: bt87x - Add a whitelist for Pinnacle PCTV (11bd:0012)
Takashi Iwai [Tue, 13 Oct 2009 07:34:28 +0000 (09:34 +0200)]
ALSA: bt87x - Add a whitelist for Pinnacle PCTV (11bd:0012)

Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agocciss: Add cciss_allow_hpsa module parameter
Stephen M. Cameron [Tue, 13 Oct 2009 07:18:22 +0000 (09:18 +0200)]
cciss: Add cciss_allow_hpsa module parameter

Add cciss_allow_hpsa module parameter.  This parameter causes
the cciss driver to ignore any Smart Array devices known to be
supported by the hpsa driver.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agocciss: Fix multiple calls to pci_release_regions
Stephen M. Cameron [Tue, 13 Oct 2009 07:18:22 +0000 (09:18 +0200)]
cciss: Fix multiple calls to pci_release_regions

Fix multiple calls to pci_release_regions.  If cciss_pci_init
fails, it already does any necessary call to pci_release_regions,
so this does not need to be done again in cciss_init_one in that
case.

Signed-off-by: Stephen M. Cameron <scameron@beardog.cce.hp.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agoALSA: hda - Fix overflow of spec->init_verbs in patch_realtek.c
Takashi Iwai [Tue, 13 Oct 2009 06:06:55 +0000 (08:06 +0200)]
ALSA: hda - Fix overflow of spec->init_verbs in patch_realtek.c

ALC861-VD lenovo model causes overflow of spec->init_verbs entries due to
the recent changes.  Simply increase the array size to avoid the overflow.

Reported-by: Luca Tettamanti <kronos.it@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoext3: Update documentation about ext3 quota mount options
Jan Kara [Thu, 17 Sep 2009 15:37:12 +0000 (17:37 +0200)]
ext3: Update documentation about ext3 quota mount options

Signed-off-by: Jan Kara <jack@suse.cz>
15 years agoext3: Don't update superblock write time when filesystem is read-only
Theodore Ts'o [Wed, 16 Sep 2009 15:21:13 +0000 (11:21 -0400)]
ext3: Don't update superblock write time when filesystem is read-only

This avoids updating the superblock write time when we are mounting
the root file system read/only but we need to replay the journal; at
that point, for people who are east of GMT and who make their clock
tick in localtime for Windows bug-for-bug compatibility, and this will
cause e2fsck to complain and force a full file system check.

Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Signed-off-by: Jan Kara <jack@suse.cz>
15 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jbarnes...
Linus Torvalds [Mon, 12 Oct 2009 21:38:34 +0000 (14:38 -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: Prevent AER driver from being loaded on non-root port PCIE devices
  PCI: get larger bridge ranges when space is available
  PCI: pci.c: fix kernel-doc notation
  PCI quirk: TI XIO200a erroneously reports support for fast b2b transfers
  PCI PM: Read device power state from register after updating it
  PCI: remove pci_assign_resource_fixed()
  PCI: PCIe portdrv: remove "-driver" from driver name

15 years agoMerge master.kernel.org:/home/rmk/linux-2.6-arm
Linus Torvalds [Mon, 12 Oct 2009 21:37:49 +0000 (14:37 -0700)]
Merge master.kernel.org:/home/rmk/linux-2.6-arm

* master.kernel.org:/home/rmk/linux-2.6-arm: (24 commits)
  ARM: force dcache flush if dcache_dirty bit set
  [ARM] pxa: workaround errata #37 by not using half turbo switching
  [ARM] pxamci: fix printing gpio numbers in pxamci_probe
  [ARM] pxa/csb726: adjust duplicate structure field initialization
  ARM: Add kmap_atomic type debugging
  ARM: boolean bit testing
  ARM: update die() output
  ARM: Dump code/mem oops lines with the appropriate log level
  ARM: Dump memory and backtrace as one printk per line
  ARM: 5756/1: ep93xx: introduce clk parent
  ARM: 5754/1: ep93xx: update i2c support
  ARM: 5753/1: ep93xx: remove old EP93XX_GPIO_* defines
  ARM: 5729/1: ep93xx: define EP93XX_*_PHYS_BASE with macros
  ARM: 5751/1: ep93xx/micro9: Add Micro9-Slim
  ARM: 5750/1: ep93xx/micro9: Update platform code
  ARM: 5749/1: ep93xx/micro9: Update maintainer
  ARM: 5752/1: SA1100: fix building of h3100
  ARM: 5748/1: bcmring: fix build warning messages
  ARM: 5747/1: Fix the start_pg value in free_memmap()
  ARM: 5746/1: Handle possible translation errors in ARMv6/v7 coherent_user_range
  ...

15 years agoUSB: musb: invert arch depend string
Mike Frysinger [Mon, 12 Oct 2009 13:49:56 +0000 (09:49 -0400)]
USB: musb: invert arch depend string

The MUSB code relies on platform implementations that currently only
exists for Arm and Blackfin processors, so have the MUSB Kconfig depend
upon those arches.

This should prevent other arches from building MUSB via randconfig.

Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoNFS: suppress a build warning
Stefan Richter [Mon, 12 Oct 2009 15:26:12 +0000 (11:26 -0400)]
NFS: suppress a build warning

struct sockaddr_storage * can safely be used as struct sockaddr *.
Suppress an "incompatible pointer type" warning.

Signed-off-by: Stefan Richter <stefanr@s5r6.in-berlin.de>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agotty, serial: Fix race and NULL check in uart_close()
Linus Torvalds [Mon, 12 Oct 2009 17:13:54 +0000 (19:13 +0200)]
tty, serial: Fix race and NULL check in uart_close()

Commit 46d57a449aa1 ("serial: use tty_port pointers in the core code")
contained two bugs that causes (rare) crashes:

 - the rename typoed one site

 - a NULL check was missed

Reported-and-tested-by: Ingo Molnar <mingo@elte.hu>
Cc: Greg KH <gregkh@suse.de>
Cc: Alan Cox <alan@lxorguk.ukuu.org.uk>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoARM: force dcache flush if dcache_dirty bit set
Nitin Gupta [Mon, 12 Oct 2009 08:50:23 +0000 (14:20 +0530)]
ARM: force dcache flush if dcache_dirty bit set

On ARM, update_mmu_cache() does dcache flush for a page only if
it has a kernel mapping (page_mapping(page) != NULL). The correct
behavior would be to force the flush based on dcache_dirty bit only.

One of the cases where present logic would be a problem is when
a RAM based block device[1] is used as a swap disk. In this case,
we would have in-memory data corruption as shown in steps below:

do_swap_page()
{
    - Allocate a new page (if not already in swap cache)
    - Issue read from swap disk
        - Block driver issues flush_dcache_page()
        - flush_dcache_page() simply sets PG_dcache_dirty bit and does not
          actually issue a flush since this page has no user space mapping yet.
    - Now, if swap disk is almost full, this newly read page is removed
      from swap cache and corrsponding swap slot is freed.
    - Map this page anonymously in user space.
    - update_mmu_cache()
        - Since this page does not have kernel mapping (its not in page/swap
          cache and is mapped anonymously), it does not issue dcache flush
          even if dcache_dirty bit is set by flush_dcache_page() above.

    <user now gets stale data since dcache was never flushed>
}

Same problem exists on mips too.

[1] example:
 - brd (RAM based block device)
 - ramzswap (RAM based compressed swap device)

Signed-off-by: Nitin Gupta <ngupta@vflare.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agox86: Move pci_iommu_init to rootfs_initcall()
David Woodhouse [Mon, 12 Oct 2009 11:59:29 +0000 (12:59 +0100)]
x86: Move pci_iommu_init to rootfs_initcall()

We want this to happen after the PCI quirks, which are now running at
the very end of the fs_initcalls.

This works around the BIOS problems which were originally addressed by
commit db8be50c4307dac2b37305fc59c8dc0f978d09ea ('USB: Work around BIOS
bugs by quiescing USB controllers earlier'), which was reverted in
commit d93a8f829fe1d2f3002f2c6ddb553d12db420412.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
15 years agoRun pci_apply_final_quirks() sooner.
David Woodhouse [Mon, 12 Oct 2009 11:51:22 +0000 (12:51 +0100)]
Run pci_apply_final_quirks() sooner.

Having this as a device_initcall() means that some real device drivers
can actually initialise _before_ the quirks are run, which is wrong.

We want it to run _before_ device_initcall(), but _after_ fs_initcall(),
since some arch-specific PCI initialisation like pcibios_assign_resources()
is done at fs_initcall().

We could use rootfs_initcall() but I actually want to use that for the
IOMMU initialisation, which has to come after the quirks, but still
before the real devices. So use fs_initcall_sync() instead -- since this
is entirely synchronous, it doesn't hurt that it'll escape the
synchronisation.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
15 years agoMark pci_apply_final_quirks() __init rather than __devinit
David Woodhouse [Mon, 12 Oct 2009 11:50:34 +0000 (12:50 +0100)]
Mark pci_apply_final_quirks() __init rather than __devinit

It doesn't get invoked on hotplug; it can be thrown away after init.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
15 years agoRename pci_init() to pci_apply_final_quirks(), move it to quirks.c
David Woodhouse [Mon, 12 Oct 2009 11:48:43 +0000 (12:48 +0100)]
Rename pci_init() to pci_apply_final_quirks(), move it to quirks.c

This function may have done more in the past, but all it does now is
apply the PCI_FIXUP_FINAL quirks. So name it sensibly and put it where
it belongs.

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
15 years agoMerge branch 'fix' of git://git.kernel.org/pub/scm/linux/kernel/git/ycmiao/pxa-linux-2.6
Russell King [Mon, 12 Oct 2009 13:38:08 +0000 (14:38 +0100)]
Merge branch 'fix' of git://git./linux/kernel/git/ycmiao/pxa-linux-2.6

15 years agomfd: Fix twl4030 boot with twl4030 usb transceiver enabled
Roger Quadros [Wed, 30 Sep 2009 15:26:37 +0000 (18:26 +0300)]
mfd: Fix twl4030 boot with twl4030 usb transceiver enabled

The usb regulator supplies (usb1v5, usb1v8 & usb3v1) must be available
before adding the twl4030_usb child, else twl4030_usb_ldo_init() will
always fail thus causing boot lock-up.

This patch fixes boot on OMAP systems using the twl4030 usb transceiver.
CONFIG_TWL4030_USB=y

Signed-off-by: Roger Quadros <ext-roger.quadros@nokia.com>
Signed-off-by: Samuel Ortiz <sameo@linux.intel.com>
15 years agoHID: fix possible deadlock in hidraw_read
Jiri Kosina [Mon, 12 Oct 2009 09:25:56 +0000 (11:25 +0200)]
HID: fix possible deadlock in hidraw_read

If the loop in hidraw_read() loops more than once, then we might
end up trying to acquire already locked mutex, casuing a deadlock.

Reported-by: iceberg <iceberg@ispras.ru>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
15 years agopercpu: fix compile warnings
Tejun Heo [Sun, 4 Oct 2009 00:31:05 +0000 (09:31 +0900)]
percpu: fix compile warnings

Fix the following two compile warnings which show up on i386.

mm/percpu.c:1873: warning: comparison of distinct pointer types lacks a cast
mm/percpu.c:1879: warning: format '%lx' expects type 'long unsigned int', but argument 2 has type 'size_t'

Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Randy Dunlap <randy.dunlap@oracle.com>
15 years ago[ARM] pxa: workaround errata #37 by not using half turbo switching
Dennis O'Brien [Sat, 10 Oct 2009 07:08:52 +0000 (15:08 +0800)]
[ARM] pxa: workaround errata #37 by not using half turbo switching

PXA27x Errata #37 implies system will hang when switching into or out of
half turbo (HT bit in CLKCFG) mode, workaround this by not using it.

Signed-off-by: Dennis O'Brien <dennis.obrien@eqware.net>
Cc: stable-2.6.31 <stable@kernel.org>
Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
15 years ago[ARM] pxamci: fix printing gpio numbers in pxamci_probe
Antonio Ospite [Fri, 2 Oct 2009 14:24:02 +0000 (16:24 +0200)]
[ARM] pxamci: fix printing gpio numbers in pxamci_probe

Signed-off-by: Antonio Ospite <ospite@studenti.unina.it>
Reviewed-by: Robert Jarzmik <robert.jarzmik@free.fr>
Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
15 years ago[ARM] pxa/csb726: adjust duplicate structure field initialization
Julia Lawall [Wed, 16 Sep 2009 09:06:52 +0000 (11:06 +0200)]
[ARM] pxa/csb726: adjust duplicate structure field initialization

Currently the irq_type field of the csb726_lan_config structure is
initialized twice.  The value in the first case,
SMSC911X_IRQ_POLARITY_ACTIVE_LOW, is normally stored in the irq_polarity
field, so I have renamed the field in the first initialization to that.

Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: Eric Miao <eric.y.miao@gmail.com>
15 years agoblk-settings: fix function parameter kernel-doc notation
Randy Dunlap [Mon, 12 Oct 2009 06:20:47 +0000 (08:20 +0200)]
blk-settings: fix function parameter kernel-doc notation

Fix kernel-doc notation in blk-settings.c::blk_queue_max_discard_sectors().

Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
15 years agoRevert "USB: Work around BIOS bugs by quiescing USB controllers earlier"
Linus Torvalds [Sun, 11 Oct 2009 22:57:57 +0000 (15:57 -0700)]
Revert "USB: Work around BIOS bugs by quiescing USB controllers earlier"

This reverts commit db8be50c4307dac2b37305fc59c8dc0f978d09ea, as per

http://bugzilla.kernel.org/show_bug.cgi?id=14374
http://marc.info/?l=linux-kernel&m=125446885705223&w=4

We simply can't do the USB handoff at FIXUP_HEADER time, since it will
often require us to have valid IO mappings etc.  But that in turn
requires a whole different approach, not this trivial one-liner.

Maybe we could teach all the USB quirk handoff handlers to only do the
quirk if the device has all its registers set up (since if it isn't
initialized, it's unlikely to be active), but regardless that will need
a whole lot more code than just saying "let's do it really early".

The proper fix is almost certainly to just leave the legacy IOMMU
mappings active until after all devices have been initialized.

Reported-by: Nick Piggin <npiggin@suse.de>
Cc: David Woodhouse <David.Woodhouse@intel.com>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoLinux 2.6.32-rc4
Linus Torvalds [Sun, 11 Oct 2009 21:43:56 +0000 (14:43 -0700)]
Linux 2.6.32-rc4

15 years agopci: increase alignment to make more space for hidden code
Yinghai Lu [Sun, 11 Oct 2009 21:17:16 +0000 (14:17 -0700)]
pci: increase alignment to make more space for hidden code

As reported in

http://bugzilla.kernel.org/show_bug.cgi?id=13940

on some system when acpi are enabled, acpi clears some BAR for some
devices without reason, and kernel will need to allocate devices for
them.  It then apparently hits some undocumented resource conflict,
resulting in non-working devices.

Try to increase alignment to get more safe range for unassigned devices.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agokbuild: revert "save ARCH & CROSS_COMPILE ..."
Sam Ravnborg [Sun, 11 Oct 2009 21:22:58 +0000 (23:22 +0200)]
kbuild: revert "save ARCH & CROSS_COMPILE ..."

Revert commit 575543347b5baed0ca927cb90ba8807396fe9cc9

It caused following issues:

- On architectures where ARCH= setting is used to select between
  32 and 64 bit this was no longer possible without "make mrproper"
- If ARCH was changed then kbuild refused to run "make mrproper"
  because ARCH had changed
- When CROSS_COMPILE was changed people were asked to run "make mrproper"
  but kbuild refused to run "make mrproper" because CROSS_COMPILE changed.
- Spaces in CROSS_COMPILE was not 'supported'
- If an non-existing ARCH= was used kbuild could get stuck

Lessons learned:
. Despite being simple and straghtforward people uses very different
  approaches when building the kernel.

. CROSS_COMPILE is sometimes used for ccache despite cache being
  only a CC frontend so one would have expected CC to be
  used for this purpose.

. And obviously this was not tested widely enough.

Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
Cc: Pavel Machek <pavel@ucw.cz>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
15 years agowarn about use of uninstalled kernel headers
Arnd Bergmann [Mon, 21 Sep 2009 14:37:12 +0000 (16:37 +0200)]
warn about use of uninstalled kernel headers

User applications frequently hit problems when they try to use
the kernel headers directly, rather than the exported headers.

This adds an explicit warning for this case, and points to
a URL holding an explanation of why this is wrong and what
to do about it.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
15 years agokbuild: mkcompile_h: trivial cleanups
Felipe Contreras [Wed, 16 Sep 2009 21:38:40 +0000 (00:38 +0300)]
kbuild: mkcompile_h: trivial cleanups

UTS_TRUNCATTE is simpler this way, and now editors idetify this as a
shell script.

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
15 years agokbuild: fix warning when domainname is not available
Felipe Contreras [Wed, 16 Sep 2009 21:38:39 +0000 (00:38 +0300)]
kbuild: fix warning when domainname is not available

Otherwise we get:
"dnsdomainname: Unknown host"

Signed-off-by: Felipe Contreras <felipe.contreras@gmail.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
15 years agokbuild: Fix size_append issue for bzip2/lzma kernel
Alek Du [Wed, 19 Aug 2009 09:18:39 +0000 (17:18 +0800)]
kbuild: Fix size_append issue for bzip2/lzma kernel

The Makefile.lib will call "echo -ne" to append uncompressed kernel size to
bzip2/lzma kernel image.
The "echo" here depends on the shell that /bin/sh pointing to.
On Ubuntu system, the /bin/sh is pointing to dash, which does not support
"echo -e" at all. Use /bin/echo instead of shell echo should always be safe.

Signed-off-by: Alek Du <alek.du@intel.com>
Acked-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
15 years agokbuild,scripts: use non-builtin echo for '-e'
Amerigo Wang [Wed, 19 Aug 2009 10:19:55 +0000 (18:19 +0800)]
kbuild,scripts: use non-builtin echo for '-e'

Alek reported that on Ubuntu, where dash is used, 'echo -e'
can't work, so let's use non-builtin echo in this case.

Reported-by: Alek Du <alek.du@intel.com>
Signed-off-by: WANG Cong <xiyou.wangcong@gmail.com>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
15 years agokbuild: fix the binrpm-pkg target to work with KBUILD_OUTPUT set
Frans Pop [Tue, 25 Aug 2009 20:15:39 +0000 (22:15 +0200)]
kbuild: fix the binrpm-pkg target to work with KBUILD_OUTPUT set

The binrpm-pkg target (binary RPM only) fails when called with
KBUILD_OUTPUT set. This patch makes it work.

For the rpm-pkg target (source + binary RPM), building with
KBUILD_OUTPUT set is not possible and also not needed as the
actual build is done in a temporary directory anyway, so check
that KBUILD_OUTPUT is not set in that case to avoid later errors.

Signed-off-by: Frans Pop <elendil@planet.nl>
Signed-off-by: Sam Ravnborg <sam@ravnborg.org>
15 years agoMerge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Sun, 11 Oct 2009 18:34:50 +0000 (11:34 -0700)]
Merge branch 'for-linus' of git://git390.marist.edu/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6: (21 commits)
  [S390] dasd: fix race condition in resume code
  [S390] Add EX_TABLE for addressing exception in usercopy functions.
  [S390] 64-bit register support for 31-bit processes
  [S390] hibernate: Use correct place for CPU address in lowcore
  [S390] pm: ignore time spend in suspended state
  [S390] zcrypt: Improve some comments
  [S390] zcrypt: Fix sparse warning.
  [S390] perf_counter: fix vdso detection
  [S390] ftrace: drop nmi protection
  [S390] compat: fix truncate system call wrapper
  [S390] Provide arch specific mdelay implementation.
  [S390] Fix enabled udelay for short delays.
  [S390] cio: allow setting boxed devices offline
  [S390] cio: make not operational handling consistent
  [S390] cio: make disconnected handling consistent
  [S390] Fix memory leak in /proc/cio_ignore
  [S390] cio: channel path memory leak
  [S390] module: fix memory leak in s390 module loader
  [S390] Enable kmemleak on s390.
  [S390] 3270 console build fix
  ...

15 years agoROMFS: fix length used with romfs_dev_strnlen() function
Bernd Schmidt [Tue, 6 Oct 2009 08:55:26 +0000 (09:55 +0100)]
ROMFS: fix length used with romfs_dev_strnlen() function

An interestingly corrupted romfs file system exposed a problem with the
romfs_dev_strnlen function: it's passing the wrong value to its helpers.
Rather than limit the string to the length passed in by the callers, it
uses the size of the device as the limit.

Signed-off-by: Bernd Schmidt <bernds_cb1@t-online.de>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6
Linus Torvalds [Sun, 11 Oct 2009 18:24:25 +0000 (11:24 -0700)]
Merge git://git./linux/kernel/git/gregkh/usb-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb-2.6: (32 commits)
  USB: serial: no unnecessary GFP_ATOMIC in oti6858
  USB: serial: fix race between unthrottle and completion handler in visor
  USB: serial: fix assumption that throttle/unthrottle cannot sleep
  USB: serial: fix race between unthrottle and completion handler in symbolserial
  USB: serial: fix race between unthrottle and completion handler in opticon
  USB: ehci: Fix isoc scheduling boundary checking.
  USB: storage: When a device returns no sense data, call it a Hardware Error
  USB: small fix in error case of suspend in generic usbserial code
  USB: visor: fix trivial accounting bug in visor driver
  USB: Fix throttling in generic usbserial driver
  USB: cp210x: Add support for the DW700 UART
  USB: ipaq: fix oops when device is plugged in
  USB: isp1362: fix build warnings on 64-bit systems
  USB: gadget: imx_udc: Use resource size
  USB: storage: iRiver P7 UNUSUAL_DEV patch
  USB: musb: make HAVE_CLK support optional
  USB: xhci: Fix dropping endpoints from the xHC schedule.
  USB: xhci: Don't wait for a disable slot cmd when HC dies.
  USB: xhci: Handle canceled URBs when HC dies.
  USB: xhci: Stop debugging polling loop when HC dies.
  ...

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6
Linus Torvalds [Sun, 11 Oct 2009 18:24:05 +0000 (11:24 -0700)]
Merge git://git./linux/kernel/git/gregkh/staging-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging-2.6:
  Staging: comedi: fix build on arches that don't want comedi drivers
  Staging: comedi: pcmcia irq fixes
  Staging: comedi: ni_pcimio: Added device id for pxi-6225.
  Staging: comedi: ni_65xx.c: fix output inversion problem.
  Staging: comedi: ni_65xx.c: fix insn_bits shift calculation.
  Staging: comedi: s526: fixes for pulse generator
  Staging: comedi: s526: Take account of arch's byte order.
  Staging: comedi: s526: Get rid of global variable 'cmReg'.
  Staging: comedi: s526: Fix number of channels on DIO subdevice
  Staging: comedi: cb_pcidio: fix "section mismatch" error
  Staging: comedi: jr3_pci: Initialize transf variable fully in jr3_pci_poll_subdevice().
  Staging: comedi: Corrected type of a printk argument in resize_async_buffer().
  Staging: p9auth: a few fixes
  Staging: rtl8192e: Add #include <linux/vmalloc.h>
  Staging: iio: Don't build on s390
  Staging: winbond: implement prepare_multicast and fix API usage
  Staging: w35und: Fix ->beacon_int breakage
  Staging: remove cowloop driver
  Staging: remove agnx driver
  Staging: comedi: serial2002: fix include build issue

15 years agoMerge branch 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelv...
Linus Torvalds [Sun, 11 Oct 2009 18:23:33 +0000 (11:23 -0700)]
Merge branch 'hwmon-for-linus' of git://git./linux/kernel/git/jdelvare/staging

* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jdelvare/staging:
  hwmon: (s3c-hwmon) Disable build for S3C64xx
  MAINTAINERS: Fix Riku Voipio's address
  hwmon: (asus_atk0110) Enable the EC
  hwmon: (asus_atk0110) Refactor the code
  hwmon: (sht15) Fix spurious section mismatch warning

15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable
Linus Torvalds [Sun, 11 Oct 2009 18:23:13 +0000 (11:23 -0700)]
Merge git://git./linux/kernel/git/mason/btrfs-unstable

* git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  Btrfs: fix file clone ioctl for bookend extents
  Btrfs: fix uninit compiler warning in cow_file_range_nocow
  Btrfs: constify dentry_operations
  Btrfs: optimize back reference update during btrfs_drop_snapshot
  Btrfs: remove negative dentry when deleting subvolumne
  Btrfs: optimize fsync for the single writer case
  Btrfs: async delalloc flushing under space pressure
  Btrfs: release delalloc reservations on extent item insertion
  Btrfs: delay clearing EXTENT_DELALLOC for compressed extents
  Btrfs: cleanup extent_clear_unlock_delalloc flags
  Btrfs: fix possible softlockup in the allocator
  Btrfs: fix deadlock on async thread startup

15 years agoheaders: remove sched.h from interrupt.h
Alexey Dobriyan [Wed, 7 Oct 2009 13:09:06 +0000 (17:09 +0400)]
headers: remove sched.h from interrupt.h

After m68k's task_thread_info() doesn't refer to current,
it's possible to remove sched.h from interrupt.h and not break m68k!
Many thanks to Heiko Carstens for allowing this.

Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
15 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6
Linus Torvalds [Sun, 11 Oct 2009 18:12:33 +0000 (11:12 -0700)]
Merge git://git./linux/kernel/git/jejb/scsi-rc-fixes-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-rc-fixes-2.6: (34 commits)
  [SCSI] qla2xxx: Fix NULL ptr deref bug in fail path during queue create
  [SCSI] st: fix possible memory use after free after MTSETBLK ioctl
  [SCSI] be2iscsi: Moving to pci_pools v3
  [SCSI] libiscsi: iscsi_session_setup to allow for private space
  [SCSI] be2iscsi: add 10Gbps iSCSI - BladeEngine 2 driver
  [SCSI] zfcp: Fix hang when offlining device with offline chpid
  [SCSI] zfcp: Fix lockdep warning when offlining device with offline chpid
  [SCSI] zfcp: Fix oops during shutdown of offline device
  [SCSI] zfcp: Fix initial device and cfdc for delayed adapter allocation
  [SCSI] zfcp: correctly initialize unchained requests
  [SCSI] mpt2sas: Bump version 02.100.03.00
  [SCSI] mpt2sas: Support dev remove when phy status is MPI2_EVENT_SAS_TOPO_PHYSTATUS_VACANT
  [SCSI] mpt2sas: Timeout occurred within the HANDSHAKE logic while waiting on firmware to ACK.
  [SCSI] mpt2sas: Call init_completion on a per request basis.
  [SCSI] mpt2sas: Target Reset will be issued from Interrupt context.
  [SCSI] mpt2sas: Added SCSIIO, Internal and high priority memory pools to support multiple TM
  [SCSI] mpt2sas: Copyright change to 2009.
  [SCSI] mpt2sas: Added mpi2_history.txt for MPI2 headers.
  [SCSI] mpt2sas: Update driver to MPI2 REV K headers.
  [SCSI] bfa: Brocade BFA FC SCSI driver
  ...

15 years agoALSA: ice1724 - Make call to set hw params succeed on ESI Juli@
David Henningsson [Sun, 11 Oct 2009 09:37:22 +0000 (11:37 +0200)]
ALSA: ice1724 - Make call to set hw params succeed on ESI Juli@

If two streams are started immediately after one another (such as a
playback and a recording stream), the call to set hw params fails with
EBUSY. This patch makes the call succeed, so playback and recording will
work properly.

Signed-off-by: David Henningsson <launchpad.web@epost.diwic.se>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoARM: Add kmap_atomic type debugging
Russell King [Sun, 11 Oct 2009 15:29:48 +0000 (16:29 +0100)]
ARM: Add kmap_atomic type debugging

Seemingly this support was missed when highmem was added, so
DEBUG_HIGHMEM wouldn't have checked the kmap_atomic type.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: boolean bit testing
Johannes Weiner [Tue, 21 Jul 2009 15:08:28 +0000 (17:08 +0200)]
ARM: boolean bit testing

Bit testing (test, testset, testclear, testchange) for bit numbers
known at compile time returns a word with the tested-for bit set.

Change it to return a true boolean value so to make it consistent with
the out-of-line path and all the other bitops implementations.

Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: update die() output
Russell King [Sun, 11 Oct 2009 14:25:05 +0000 (15:25 +0100)]
ARM: update die() output

Make die() better match x86:
- add printing of the last accessed sysfs file
- ensure console_verbose() is called under the lock
- ensure we panic outside of oops_exit()

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: Dump code/mem oops lines with the appropriate log level
Russell King [Sun, 11 Oct 2009 14:17:53 +0000 (15:17 +0100)]
ARM: Dump code/mem oops lines with the appropriate log level

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: Dump memory and backtrace as one printk per line
Russell King [Sun, 11 Oct 2009 14:03:11 +0000 (15:03 +0100)]
ARM: Dump memory and backtrace as one printk per line

dump_mem and dump_backtrace were both using multiple printk statements
to print each line.  With DEBUG_LL enabled, this causes OOPS to become
very difficult to read.  Solve this by only using one printk per line.

Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: 5756/1: ep93xx: introduce clk parent
Hartley Sweeten [Thu, 8 Oct 2009 22:44:41 +0000 (23:44 +0100)]
ARM: 5756/1: ep93xx: introduce clk parent

The clock generation system in the ep93xx uses two external oscillator's
and two internal PLLs to derive all the internal clocks.  Many of these
internal clocks can be stopped to save power.

This introduces a "parent" hierarchy for the clocks so that the users
count can be correctly tracked for power management.

The "parent" for the video clock can either be one of the PLL outputs
or the external oscillator.  In order to correctly track the "parent"
for the video clock calc_clk_div() needed to be modified.  It now
returns an error code if the desired rate cannot be generated.

Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Acked-by: Ryan Mallon <ryan@bluewatersys.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: 5754/1: ep93xx: update i2c support
Hartley Sweeten [Wed, 7 Oct 2009 23:45:00 +0000 (00:45 +0100)]
ARM: 5754/1: ep93xx: update i2c support

Update the ep93xx i2c support:

  1) The platform init code passes the configuration data for the
     i2c-gpio driver.  This allows any gpio pin do be used for the
     sda and scl pins.  It also allows the platform to specify the
     udelay and timeout.
  2) Program the gpio configuration register to enable/disable the
     open drain drivers.  Note that this really only works if the
     sda and scl pins are set to EP93XX_GPIO_LINE_EEDAT and
     EP93XX_GPIO_LINE_EECLK.
  3) Update the edb93xx.c platform init to use the new support.

Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Acked-by: Ryan Mallon <ryan@bluewatersys.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoARM: 5753/1: ep93xx: remove old EP93XX_GPIO_* defines
Hartley Sweeten [Wed, 7 Oct 2009 22:43:26 +0000 (23:43 +0100)]
ARM: 5753/1: ep93xx: remove old EP93XX_GPIO_* defines

Most of the EP93XX_GPIO_*_INT_* register defines in ep93xx-regs.h
not required due to how the ep93xx core and gpiolib support handle
gpio interrupts.  Remove the defines to prevent future confusion.

Signed-off-by: H Hartley Sweeten <hsweeten@visionengravers.com>
Acked-by: Ryan Mallon <ryan@bluewatersys.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
15 years agoALSA: ice1724: Fix surround on Chaintech AV-710
Robert Hancock [Sat, 10 Oct 2009 04:08:58 +0000 (22:08 -0600)]
ALSA: ice1724: Fix surround on Chaintech AV-710

Fix the num_total_dacs setting for Chaintech AV710. The existing comment
that only PSDOUT0 is connected is correct, but since the card is using
packed AC97 mode to send 6 channels to the codec, num_total_dacs should be
set to 6 and not 2. This allows 6-channel surround to work. Also clarify
a comment regarding the additional WM8728 codec on this card (it's connected
to the SPDIF output and always receives the same data).

Signed-off-by: Robert Hancock <hancockrwd@gmail.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
15 years agoUSB: serial: no unnecessary GFP_ATOMIC in oti6858
Oliver Neukum [Wed, 7 Oct 2009 16:07:10 +0000 (18:07 +0200)]
USB: serial: no unnecessary GFP_ATOMIC in oti6858

GFP_ATOMIC without good cause is evil.

Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoUSB: serial: fix race between unthrottle and completion handler in visor
Oliver Neukum [Wed, 7 Oct 2009 09:01:38 +0000 (11:01 +0200)]
USB: serial: fix race between unthrottle and completion handler in visor

usb:usbserial:visor: fix race between unthrottle and completion handler

visor_unthrottle() mustn't resubmit the URB unconditionally
as the URB may still be running.

the same bug as opticon.

Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoUSB: serial: fix assumption that throttle/unthrottle cannot sleep
Oliver Neukum [Wed, 7 Oct 2009 08:50:23 +0000 (10:50 +0200)]
USB: serial: fix assumption that throttle/unthrottle cannot sleep

many serial subdrivers are clearly written as if throttle/unthrottle
cannot sleep. This leads to unneeded atomic submissions. This
patch converts affected drivers in a way to makes very clear that
throttle/unthrottle can sleep. Thus future misdesigns can be avoided
and efficiency and reliability improved.

This removes any such assumption using GFP_KERNEL and spin_lock_irq()

Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoUSB: serial: fix race between unthrottle and completion handler in symbolserial
Oliver Neukum [Wed, 7 Oct 2009 07:30:49 +0000 (09:30 +0200)]
USB: serial: fix race between unthrottle and completion handler in symbolserial

usb:usbserial:symbolserial: fix race between unthrottle and completion handler

symbol_unthrottle() mustn't resubmit the URB unconditionally
as the URB may still be running.

the same bug as opticon.

Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoUSB: serial: fix race between unthrottle and completion handler in opticon
Oliver Neukum [Wed, 7 Oct 2009 07:25:10 +0000 (09:25 +0200)]
USB: serial: fix race between unthrottle and completion handler in opticon

usb:usbserial:opticon: fix race between unthrottle and completion handler

opticon_unthrottle() mustn't resubmit the URB unconditionally
as the URB may still be running.

Signed-off-by: Oliver Neukum <oliver@neukum.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
15 years agoUSB: ehci: Fix isoc scheduling boundary checking.
Sarah Sharp [Tue, 6 Oct 2009 20:45:59 +0000 (13:45 -0700)]
USB: ehci: Fix isoc scheduling boundary checking.

The EHCI driver does some bounds checking when it's scheduling an iTD for
an active endpoint.  It sets the local variable start to
stream->next_uframe and moves that variable further in the schedule if
necessary.  However, the driver fails to do anything with start before
jumping to the ready label and setting the URB's starting frame to
stream->next_uframe.  Alan Stern confirms the EHCI driver should set
stream->next_uframe to start before jumping.

Signed-off-by: Sarah Sharp <sarah.a.sharp@linux.intel.com>
Acked-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>