Commit graph

311,532 commits

Author SHA1 Message Date
Mark Brown
c0cc6fe1d0 Merge branches 'regmap-core', 'regmap-mmio' and 'regmap-naming' into regmap-stride 2012-04-10 11:01:07 +01:00
Eyal Shapira
16f3eb530f wl12xx: increase scan timeout to 30s
In certain scenarios involving sched scan + normal scan + COEX
scan could take longer than 10s and this triggers a recovery
where it shouldn't. Increase the timeout to avoid that.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:48:10 +03:00
Eyal Shapira
6f407e5bc7 wl12xx: adaptive sched scan dwell times
Set the dwell times for sched scan according to the number
of probe requests which are going to be transmitted.
This should fix the too short dwell time problem which
prevented some of the probe requests from being transmitted
in cases of high number of SSIDs (10+) to be actively sched scanned.
However, in the common case of having up to 1-2 SSIDs that
require active scan, the dwell time would be kept to a minimum
which should conserve power. This is important as sched scan
also runs periodically while the host is suspended and there's
great importance to keep power consumption as low as possible.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
[fixed a couple of new strict checkpatch warnings]
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:47:24 +03:00
Eyal Shapira
3eba4a0e6d wl12xx: fix a memory leak of probereq template upon recovery
wlvif->probereq is zeroed when adding an interface but
the skb pointed to isn't freed when the interface is removed.
This would lead to a mem leak on every recovery.
Fix it by freeing the skb when removing the interface.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:42:49 +03:00
Suresh Siddha
fa4da365bc clockevents: tTack broadcast device mode change in tick_broadcast_switch_to_oneshot()
In the commit 77b0d60c5a,
"clockevents: Leave the broadcast device in shutdown mode when not needed",
we were bailing out too quickly in tick_broadcast_switch_to_oneshot(),
with out tracking the broadcast device mode change to 'TICKDEV_MODE_ONESHOT'.

This breaks the platforms which need broadcast device oneshot services during
deep idle states. tick_broadcast_oneshot_control() thinks that it is
in periodic mode and fails to take proper decisions based on the
CLOCK_EVT_NOTIFY_BROADCAST_[ENTER, EXIT] notifications during deep
idle entry/exit.

Fix this by tracking the broadcast device mode as 'TICKDEV_MODE_ONESHOT',
before leaving the broadcast HW device in shutdown mode if there are no active
requests for the moment.

Reported-and-tested-by: Santosh Shilimkar <santosh.shilimkar@ti.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: johnstul@us.ibm.com
Link: http://lkml.kernel.org/r/1334011304.12400.81.camel@sbsiddha-desk.sc.intel.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2012-04-10 11:42:07 +02:00
Stephen Warren
abec95adef regmap: fix compilation when !CONFIG_DEBUG_FS
Commit 79c64d5 "regmap: allow regmap instances to be named" changed the
prototype of regmap_debugfs_init, but didn't update the dummy inline used
when !CONFIG_DEBUGFS. Fix this.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 10:31:41 +01:00
Stephen Warren
d3c242e1f2 regmap: allow regmap instances to be named
Some devices have multiple separate register regions. Logically, one
regmap would be created per region. One issue that prevents this is that
each instance will attempt to create the same debugfs files. Avoid this
by allowing regmaps to be named, and use the name to construct the
debugfs directory name.

Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 10:31:41 +01:00
Alex Deucher
afceb9319f drm/radeon/kms: fix DVO setup on some r4xx chips
Some r4xx chips have the wrong frev in the
DVOEncoderControl table.  It should always be 1
on r4xx.  Fixes modesetting on DVO on r4xx chips
with the bad frev.

Reported by twied on #radeon.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-10 10:27:46 +01:00
Eliad Peller
830be7e021 wl12xx: free ap keys only in ap mode
The ap keys should be freed only when removing
ap role (otherwise, some arbitrary data might
get freed).

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:23:15 +03:00
Xi Wang
6587eb8261 drm/savage: fix integer overflows in savage_bci_cmdbuf()
Since cmdbuf->size and cmdbuf->nbox are from userspace, a large value
would overflow the allocation size, leading to out-of-bounds access.

Signed-off-by: Xi Wang <xi.wang@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-10 10:22:51 +01:00
Felipe Balbi
cd423dd363 usb: dwc3: ep0: increment "actual" on bounced ep0 case
due to a HW limitation we have a bounce buffer for ep0
out transfers which are not aligned with MaxPacketSize.

On such case we were not increment r->actual as we should.

This patch fixes that mistake.

Cc: stable@vger.kernel.org
Signed-off-by: Felipe Balbi <balbi@ti.com>
2012-04-10 12:21:44 +03:00
Moiz Sonasath
566ccdda07 usb: dwc3: ep0: Handle requests greater than wMaxPacketSize
To allow ep0 out transfers of upto bounce buffer size
instead of maxpacketsize, use the transfer size as multiple
of ep0 maxpacket size.

Cc: stable@vger.kernel.org
Signed-off-by: Moiz Sonasath <m-sonasath@ti.com>
Signed-off-by: Partha Basak <p-basak2@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
2012-04-10 12:21:43 +03:00
Felipe Balbi
6d258a4c42 usb: gadget: udc-core: stop UDC on device-initiated disconnect
When we want to do device-initiated disconnect,
let's make sure we stop the UDC in order to
e.g. allow lower power states to be achieved by
turning off unnecessary clocks and/or stoping
PHYs.

When reconnecting, call ->udc_start() again to
make sure UDC is reinitialized.

Cc: stable@vger.kernel.org
Signed-off-by: Felipe Balbi <balbi@ti.com>
2012-04-10 12:21:32 +03:00
Arnd Bergmann
4de833c337 drm/radeon: replace udelay with mdelay for long timeouts
Some architectures require that delays longer than a few
miliseconds are called through mdelay. This was triggered
on ARM randconfig builds.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Signed-off-by: Dave Airlie <airlied@redhat.com>
2012-04-10 10:21:00 +01:00
Mircea Gherzan
690142e988 wl12xx: fix DMA-API-related warnings
On the PandaBoard (omap_hsmmc + wl12xx_sdio) with DMA_API_DEBUG:

 WARNING: at lib/dma-debug.c:930 check_for_stack.part.8+0x7c/0xe0()
 omap_hsmmc omap_hsmmc.4: DMA-API: device driver maps memory fromstack

Signed-off-by: Mircea Gherzan <mgherzan@gmail.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:14:58 +03:00
Axel Lin
2e42a7dc40 regulator: max8998: Use simple equation to get selector
It's more efficient to get the best selector by simple equation.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 10:14:01 +01:00
Axel Lin
2358b7763e regulator: max8997: Use simple equation to get selector
It's more efficient to get the best selector by simple equation.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 10:14:01 +01:00
Axel Lin
0fa310ccab regulator: max8997: Remove n_bits from struct voltage_map_desc
The n_bits is only used in max8997_get_voltage_proper_val to check the valid
range for variable i.

Current code already ensures min_vol never greater than desc->max,
which means the variable i always in the valid range:
0 .. (desc->max - desc->min)/desc->step.

Thus we can remove the checking (i >= (1 << desc->n_bits) and then remove
n_bits from struct voltage_map_desc.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 10:14:01 +01:00
Eyal Shapira
c56dbd57f3 wl12xx: fix race between suspend/resume and recovery
The iteration on the wlvif list in wl1271_op_resume/suspend was
perfomed before locking wl->mutex which would lead to a kernel
panic in case a recovery was queued at the same time
and would delete the wlvifs from the list.

Signed-off-by: Eyal Shapira <eyal@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:13:58 +03:00
Victor Goldenshtein
ec414c7c78 wl12xx: fix station channel switch
Channel switch complete event wasn't handled
properly in station mode, as we checked wrong
CS flag.

Signed-off-by: Victor Goldenshtein <victorg@ti.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:13:35 +03:00
Eliad Peller
446f5ca19a wl12xx: set do_join on BSS_CHANGED_ASSOC
wl12xx sets the do_join flag (which later starts the
sta role) when the bssid was changed and the sta is
associated. However, this no longer happens after
the "mac80211: remove spurious BSSID change flag"
patch.

Fix it by setting the do_join flag on BSS_CHANGED_ASSOC
(for IBSS, do_join is already set on BSS_CHANGED_IBSS)

Signed-off-by: Eliad Peller <eliad@wizery.com>
Signed-off-by: Luciano Coelho <coelho@ti.com>
2012-04-10 12:13:28 +03:00
Thomas Gleixner
9886f44412 itimer: Use printk_once instead of WARN_ONCE
David pointed out, that WARN_ONCE() to report usage of an deprecated
misfeature make folks unhappy. Use printk_once() instead.

Andrew told me to stop grumbling and to remove the silly typecast
while touching the file.

Reported-by: David Rientjes <rientjes@google.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2012-04-10 11:00:30 +02:00
Axel Lin
65896e7362 regulator: Replace regulator_desc_[ldo|buck] by s5m8767_regulator_desc macro
Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:53:49 +01:00
Axel Lin
d35aad0cad regulator: Use one s5m8767_ops for both LDOs and BUCKs
All the callback function implementation are the same for both LDOs and BUCKs.
Merge them to one s5m8767_ops.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:53:49 +01:00
Axel Lin
321d2abaca regulator: Rework s5m8767_set_voltage to support both LDOs and BUCKs
s5m8767_set_voltage not only implement set_voltage callbacks for LDOs,
but also for BUCKs when s5m8767->buck_gpioindex is not set.
s5m8767_set_voltage_buck actually only for buck[2|3|4] when
s5m8767->buck_gpioindex is set.
Conditionally calling s5m8767_set_voltage in s5m8767_set_voltage_buck makes
the code hard to read.

I think merging s5m8767_set_voltage_buck and s5m8767_set_voltage actually
simplifies things. It's easy to use buck234_vol pointer to differentiate if
we need to control voltage for buck[2|3|4] by DVS GPIOs.

This patch reworks s5m8767_set_voltage to support both LDOx and BUCKx in all
cases.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Acked-by: Sangbeom Kim <sbkim73@samsung.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:53:48 +01:00
Mark Brown
b920eb41a8 Linux 3.4-rc2
-----BEGIN PGP SIGNATURE-----
 Version: GnuPG v2.0.18 (GNU/Linux)
 
 iQEcBAABAgAGBQJPgOpXAAoJEHm+PkMAQRiG8eUIAJ3Hjo0o5hqzxol8LWHDGF8s
 3h59qwRtts82u2psStZULPLbBzw6yjn8U5Mcrh7pffwCG2BzBL5rXcmXfvoPaET3
 TjylIj986Jv+jwZyuyZcLzYyoUtgjtq7hnt4tgVVEPlexYs7LxUl8vO8QjxT+bVO
 7eKgzLsXlBC9kw4P5M7kOBQ9NE5vmEcKf+fB3Hu3DqfVJtw9Lb7EeDQj/cbQsGqI
 vavSxe7lMTPsgMP2+hRz4txGTDJNRazkO2BdwfYrcNwTSa8tfBFlamww3y4+eJtq
 MFQfdNxQpumh2++T4IC2i/nLaawH7kWj2EdaRCRLMy9hVlb/v8/wcgd+1nfLkU8=
 =0on9
 -----END PGP SIGNATURE-----

Merge tag 'v3.4-rc2' into regulator-drivers

Linux 3.4-rc2 contains some fixes that further patches depend upon.
2012-04-10 09:52:59 +01:00
Axel Lin
a9d5801041 regulator: Fix build error for mc13783 and mc13892
Convert mc13783 and mc13892 to use a struct to pass in regulator runtime
configuration.

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:52:53 +01:00
Axel Lin
fe53d18def regulator: Fix a typo in da903x.c
Fix below build errors which is introduced by commit c172708
"regulator: core: Use a struct to pass in regulator runtime configuration".

  CC      drivers/regulator/da903x.o
drivers/regulator/da903x.c: In function 'da903x_regulator_probe':
drivers/regulator/da903x.c:541: error: 'conifg' undeclared (first use in this function)
drivers/regulator/da903x.c:541: error: (Each undeclared identifier is reported only once
drivers/regulator/da903x.c:541: error: for each function it appears in.)
make[2]: *** [drivers/regulator/da903x.o] Error 1
make[1]: *** [drivers/regulator] Error 2
make: *** [drivers] Error 2

Signed-off-by: Axel Lin <axel.lin@gmail.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:52:53 +01:00
Stephen Boyd
b3aa1584e9 workqueue: Fix workqueue_execute_end() comment
workqueue_execute_end() is called after the callback function,
not before.

Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
2012-04-10 10:49:47 +02:00
Mark Brown
ac97c62071 regulator: Fix build of ab8500
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:36:34 +01:00
Mark Brown
bbdd391556 ASoC: wm1250-ev1: Convert to module_i2c_driver
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
2012-04-10 09:31:56 +01:00
Will Deacon
34af657916 ARM: 7377/1: vic: re-read status register before dispatching each IRQ handler
handle_IRQ may briefly cause interrupts to be re-enabled during soft IRQ
processing on the exit path, leading to nested handling of VIC interrupts.

Since the current code does not re-read the VIC_IRQ_STATUS register, this
can lead to multiple invocations of the same interrupt handler and
spurious interrupts to be reported.

This patch changes the VIC interrupt dispatching code to re-read the
status register each time, avoiding duplicate invocations of the same
handler.

Acked-and-Tested-by: H Hartley Sweeten <hsweeten@visionengravers.com>

Reviewed-by: Jamie Iles <jamie@jamieiles.com>
Signed-off-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2012-04-10 09:27:42 +01:00
Kautuk Consul
dff2aa7af8 ARM: 7368/1: fault.c: correct how the tsk->[maj|min]_flt gets incremented
commit 8878a539ff was done by me
to make the page fault handler retryable as well as interruptible.

Due to this commit, there is a mistake in the way in which
tsk->[maj|min]_flt counter gets incremented for VM_FAULT_ERROR:
If VM_FAULT_ERROR is returned in the fault flags by handle_mm_fault,
then either maj_flt or min_flt will get incremented. This is wrong
as in the case of a VM_FAULT_ERROR we need to be skip ahead to the
error handling code in do_page_fault.

Added a check after the call to __do_page_fault() to check for
(fault & VM_FAULT_ERROR).

Signed-off-by: Kautuk Consul <consul.kautuk@gmail.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
2012-04-10 09:27:42 +01:00
Dmitry Torokhov
d568778298 Input: wacom_i2c - do not use irq_to_gpio
Because irq_to_gpio() is not available on many platforms let's switch
to level-triggered one-shot IRQs that will stay active as long as
there is data to be read.

Tested-by: Tobita Tatsunosuke <tobita.tatsunosuke@wacom.co.jp>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
2012-04-10 00:42:15 -07:00
JJ Ding
fb16395ee6 Input: trackpoint - use psmouse_fmt() for messages
Use psmouse_*() macros introduced in commit b5d2170436.

Signed-off-by: JJ Ding <dgdunix@gmail.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
2012-04-10 00:40:39 -07:00
JJ Ding
e3dde4fba9 Input: elantech - v4 is a clickpad, with only one button
Add pointer and buttonpad properties for v4 hardware.

Also, Jachiet reported that on Asus UX31, right button has no effect.
It turns out v4 has only one button, the right-button effect is
implemented with software when Windows driver is installed, or in
firmware when touchpad is in relative mode. So remove BTN_RIGHT
while at it.

Reported-by: Jachiet Louis <louis@jachiet.com>
Signed-off-by: JJ Ding <jj_ding@emc.com.tw>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
2012-04-10 00:40:35 -07:00
JJ Ding
a67ada7a72 Input: elantech - reset touchpad before configuring it
Acer VH40 has a Fn key toggling the touchpad on and off, but it's
implemented in system firmware, and the EC chip has to receive
reset command to activate this function. Also when this machine
wakes up after resume, psmouse_reset is necessary to bring the
touchpad back on.

Signed-off-by: JJ Ding <jj_ding@emc.com.tw>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>
Signed-off-by: Dmitry Torokhov <dtor@mail.ru>
2012-04-10 00:40:31 -07:00
Benjamin Herrenschmidt
08f1ec8a59 powerpc: Fix page fault with lockdep regression
commit a546498f3b
introduced a regression on 32-bit when irq tracing
is enabled by exposing an old bug in our irq tracing
code for exception entry.

The code would save and restore some GPRs around the
calls to the C lockdep code, however, it tries to be
too smart for its own good and restores some of the
GPRs from the exception frame (as saved there on
exception entry).

However, for page faults, we do replace those GPRs with
arguments to do_page_fault before we call transfer_to_handler
and so restoring from the exception frame is plain wrong in
this case.

This was fine as long as we didn't touch the interrupt state
when taking page fault, but when I started doing it, it would
trigger the lockdep calls and the bug.

This fixes it by cleaning up that code a bit. It did create
a small stack frame for the sake of backtraces, so let's
make it a bit bigger and use it to save and restore the
stuff we care about.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2012-04-10 17:21:35 +10:00
Jesper Juhl
507230c999 ALSA: riptide: remove redundant NULL test before release_firmware()
release_firmware() deals gracefully with NULL pointers, no need to check first.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-04-10 08:42:33 +02:00
Julia Lawall
38be95dd3d ALSA: sound/isa/sscape.c: add missing resource-release code
At the point of this error-handling code, both regions and the dma have
been allocated, so free it as done in previous and subsequent
error-handling code.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-04-10 08:42:07 +02:00
Julia Lawall
156d14da4c sound: sound/oss/msnd_pinnacle.c: add vfrees
At the point of this error-handling code, HAVE_DSPCODEH may be undefined,
so free INITCODE and PERMCODE as done elsewhere.  A jump and label are
introduced to avoid code duplication.

Signed-off-by: Julia Lawall <Julia.Lawall@lip6.fr>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
2012-04-10 08:41:54 +02:00
Paul Mundt
fd34ef9bc4 sh: kgdb: Fix up NULL pointer deref by kgdb_nmicallback.
kgdb_nmicallback expects valid register state, so just fetch the register
state with get_irq_regs() as on other platforms.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
2012-04-10 14:22:39 +09:00
Paul Mundt
10c5e4e137 sh: kgdb: Fill out sleeping_thread_to_gdb_regs() state.
Presently we're using a pretty dumbed-down implementation that copies
over register state visible from the thread info, leaving the bulk of the
switch_to state uncopied.

Given that we're also depending on register bank toggling for switch_to
optimization we ought to also explicitly zero out the GP regs that reside
in an alternate bank in order to prevent handing back garbage.

There are a few extra registers that we have state for in switch_to, so
copy those over while we're at it.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
2012-04-10 14:00:30 +09:00
Paul Mundt
fd03e81812 sh: kgdb: Individual register get/set support.
This updates sh following the generic kgdb changes adding support
for individual register get/set for kgdb/kdb use.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
2012-04-10 13:42:56 +09:00
Frank Rowand
258f742635 modpost: Fix modpost license checking of vmlinux.o
Commit f02e8a6596 ("module: Sort exported symbols") sorts symbols
placing each of them in its own elf section.  This sorting and merging
into the canonical sections are done by the linker.

Unfortunately modpost to generate Module.symvers file parses vmlinux.o
(which is not linked yet) and all modules object files (which aren't
linked yet).  These aren't sanitized by the linker yet.  That breaks
modpost that can't detect license properly for modules.

This patch makes modpost aware of the new exported symbols structure.

[ This above is a slightly corrected version of the explanation of the
  problem, copied from commit 62a2635610 ("modpost: Fix modpost's
  license checking V3").  That commit fixed the problem for module
  object files, but not for vmlinux.o.  This patch fixes modpost for
  vmlinux.o. ]

Signed-off-by: Frank Rowand <frank.rowand@am.sony.com>
Signed-off-by: Alessio Igor Bogani <abogani@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-04-09 20:52:56 -07:00
David Rientjes
83dbbdbb38 android, lowmemorykiller: remove task handoff notifier
The task handoff notifier leaks task_struct since it never gets freed
after the callback returns NOTIFY_OK, which means it is responsible for
doing so.

It turns out the lowmemorykiller actually doesn't need this notifier at
all.  It's used to prevent unnecessary killing by waiting for a thread
to exit as a result of lowmem_shrink(), however, it's possible to do
this in the same way the kernel oom killer works by setting TIF_MEMDIE
and avoid killing if we're still waiting for it to exit.

The kernel oom killer will already automatically set TIF_MEMDIE for
threads that are attempting to allocate memory that have a fatal signal.
The thread selected by lowmem_shrink() will have such a signal after the
lowmemorykiller sends it a SIGKILL, so this won't result in an
unnecessary use of memory reserves for the thread to exit.

This has the added benefit that we don't have to rely on
CONFIG_PROFILING to prevent needlessly killing tasks.

Reported-by: Werner Landgraf <w.landgraf@ru.ru>
Cc: stable@vger.kernel.org
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Colin Cross <ccross@android.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2012-04-09 20:48:44 -07:00
Paul Mundt
14f087d839 sh: kgdb: Fix up basic SMP support.
kgdb needs a kgdb_roundup_cpus() definition in the architecture backend,
so just copy over the MIPS version, which already does what we want.

Signed-off-by: Paul Mundt <lethal@linux-sh.org>
2012-04-10 12:39:55 +09:00
Alan Stern
b446b96fd1 UHCI: hub_status_data should indicate if ports are resuming
This patch (as1538) causes uhci_hub_status_data() to return a nonzero
value when any port is undergoing a resume transition while the root
hub is suspended.  This will allow usbcore to handle races between
root-hub suspend and port wakeup.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-04-09 15:43:21 -07:00
Alan Stern
a448e4dc25 EHCI: keep track of ports being resumed and indicate in hub_status_data
This patch (as1537) adds a bit-array to ehci-hcd for keeping track of
which ports are undergoing a resume transition.  If any of the bits
are set when ehci_hub_status_data() is called, the routine will return
a nonzero value even if no ports have any status changes pending.
This will allow usbcore to handle races between root-hub suspend and
port wakeup.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Sarah Sharp <sarah.a.sharp@linux.intel.com>
CC: Chen Peter-B29397 <B29397@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-04-09 15:43:21 -07:00
Alan Stern
879d38e6bc USB: fix race between root-hub suspend and remote wakeup
This patch (as1533) fixes a race between root-hub suspend and remote
wakeup.  If a wakeup event occurs while a root hub is suspending, it
might not cause the suspend to fail.  Although the host controller
drivers check for pending wakeup events at the start of their
bus_suspend routines, they generally do not check for wakeup events
while the routines are running.

In addition, if a wakeup event occurs any time after khubd is frozen
and before the root hub is fully suspended, it might not cause a
system sleep transition to fail.  For example, the host controller
drivers do not fail root-hub suspends when a connect-change event is
pending.

To fix both these issues, this patch causes hcd_bus_suspend() to query
the controller driver's hub_status_data method after a root hub is
suspended, if the root hub is enabled for wakeup.  Any pending status
changes will count as wakeup events, causing the root hub to be
resumed and the overall suspend to fail with -EBUSY.

A significant point is that not all events are reflected immediately
in the status bits.  Both EHCI and UHCI controllers notify the CPU
when remote wakeup begins on a port, but the port's suspend-change
status bit doesn't get set until after the port has completed the
transition out of the suspend state, some 25 milliseconds later.
Consequently, the patch will interpret any nonzero return value from
hub_status_data as indicating a pending event, even if none of the
status bits are set in the data buffer.  Follow-up patches make the
necessary changes to ehci-hcd and uhci-hcd.

Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
CC: Sarah Sharp <sarah.a.sharp@linux.intel.com>
CC: Chen Peter-B29397 <B29397@freescale.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2012-04-09 15:43:21 -07:00