CHROMIUM: config: Remove duplicate CONFIG_RT2xxx

Fix-up entires so global chromeos/config/base.config entries
are all we need.

Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chromium-os:36732,chromium-os:35699
TEST=emerge-{parrot,daisy,x86-alex} chromeos-kernel:
Ensure prepareconfig does not fail due to CONFIG_RT* (it fails in
all of these platforms but for other reasons than my changes)

Change-Id: I32be7178c5cf255392d9b015797b0a17cba5918f
Reviewed-on: https://gerrit.chromium.org/gerrit/38839
Tested-by: Grant Grundler <grundler@chromium.org>
Reviewed-by: Grant Grundler <grundler@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>

CHROMIUM: conifig: base added bootcache

Turned on the kernel configuration for bootcache. This does
not enable it for booting. Just compiles in the device
mapper for bootcache.

BUG=chromium-os:25441
TEST=autotest kernel_Bootcache.py

Change-Id: I299abac07c5d4fa5992c240f3197c7566d57bd83
Signed-off-by: Paul Taysom <taysom@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38330
Reviewed-by: Olof Johansson <olofj@chromium.org>

CHROMIUM: arch/arm: Remove duplicated code in syscall_trace

This is breaking the ptrace syscall on ARM. Looks like a mismerge in
5ea62181 ("CHROMIUM: arch/arm: move secure_computing into trace").

TEST=ran "strace true"
BUG=chromium-os:36102

Change-Id: Ic377577216bed3672201ca71604f650c1b8b3f0c
Signed-off-by: Michael Spang <spang@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38602
Reviewed-by: Doug Anderson <dianders@chromium.org>

samsung: snow: bitfix: Bail if any pages in a bad chunk were skipped

If we didn't store recovery info for any of the pages in a bad chunk
then we'll consider it an unrecoverable error and reboot.  The old
code only checked the first page in the chunk.  With 8K chunks this
was only a 50% hit rate.

We could just choose to recover the page whose recovery info we have.
We don't do that because there is often more than one bit error in a
chunk and it seems better to error on the side of rebooting.

BUG=chrome-os-partner:15655
TEST=suspend_stress_test

Change-Id: Ic578bcfaf19efa33f0cc118c9a2f5ca64e5063bb
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38752
Reviewed-by: Jon Kliegman <kliegs@chromium.org>

regulator: tps65090: Retry FET enable in case of failure

Sometimes the FET fails to enable but instead gives a timeout error. This
may be a fault with the TPS65090. Turning off the FET and trying again seems
to clear the fault, so do this.

BUG=chrome-os-partner:16222
TEST=manual
Build and boot kernel using a firmware which does not enable the backlight.
See that LCD backlight turns on.

Also need to test on a unit which enhibits the 'no backlight on resume'
problem.

Change-Id: Idb5d867d32b4a290c6afd3d326e8c8f8678588d3
Signed-off-by: Simon Glass <sjg@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38679
Reviewed-by: Michael Spang <spang@chromium.org>
Tested-by: Michael Spang <spang@chromium.org>
Reviewed-by: Doug Anderson <dianders@chromium.org>

CHROMIUM: rts_pstor: prevent deadlock and detection issues

Patch from Realtek that fixes a deadlock issue when combined with removing the
sdhci_pci module from the kernel. This disables the card interrupt over the bus
and something else that's undocumented.

The change also prevents issues with the card reader either not detecting an SD
card insertion or removal.

TEST=suspend, insert sd card, resume
BUG=chrome-os-partner:14934, chrome-os-partner:16053

Change-Id: Ibe81ab819662b8d6721dcfe5357d12eaf2a2335d
Signed-off-by: Derek Basehore <dbasehore@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38713
Reviewed-by: Duncan Laurie <dlaurie@chromium.org>

UPSTREAM: RTC: Avoid races between RTC alarm wakeup and suspend.

If an RTC alarm fires just as suspend is happening, it is possible for
suspend to complete and the alarm to be missed.

To avoid the race, we must register the event with the PM core.

As the event is made visible to userspace through a thread which is
only scheduled by the interrupt, we need a pm_stay_awake/pm_relax
pair preventing suspend from the interrupt until the thread completes
its work.

This makes the pm_wakeup_event() call in cmos_interrupt unnecessary as
it provides suspend protection for all RTCs that use rtc_update_irq.

BUG=chromium-os:36570
TEST=Fire wakealarm, check wakeup_count for increase:
  cat /sys/power/wakeup_count
  echo +1 >| /sys/class/rtc/rtc0/wakealarm
  sleep 1
  cat /sys/power/wakeup_count

Change-Id: I7dfc88bf6f95d0c5916b7e6ba751a56c62b5b82d
Signed-off-by: NeilBrown <neilb@suse.de>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Signed-off-by: Michael Spang <spang@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38480
Reviewed-by: Grant Grundler <grundler@chromium.org>
Reviewed-by: Jon Kliegman <kliegs@chromium.org>

UPSTREAM: cpuidle: Measure idle state durations with monotonic clock

Many cpuidle drivers measure their time spent in an idle state by
reading the wallclock time before and after idling and calculating the
difference. This leads to erroneous results when the wallclock time gets
updated by another processor in the meantime, adding that clock
adjustment to the idle state's time counter.

If the clock adjustment was negative, the result is even worse due to an
erroneous cast from int to unsigned long long of the last_residency
variable. The negative 32 bit integer will zero-extend and result in a
forward time jump of roughly four billion milliseconds or 1.3 hours on
the idle state residency counter.

This patch changes all affected cpuidle drivers to either use the
monotonic clock for their measurements or make use of the generic time
measurement wrapper in cpuidle.c, which was already working correctly.
Some superfluous CLIs/STIs in the ACPI code are removed (interrupts
should always already be disabled before entering the idle function, and
not get reenabled until the generic wrapper has performed its second
measurement). It also removes the erroneous cast, making sure that
negative residency values are applied correctly even though they should
not appear anymore.

BUG=chrome-os-partner:15864
TEST=Print /sys/.../cpuidle/state*/time in a loop, and update the
system clock concurrently.

Has been accepted upstream by Rafael J. Wysocki at
https://lkml.org/lkml/2012/11/27/435

Change-Id: If53c1e60358f62679d29130b14a68487125ff164
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Acked-by: Len Brown <len.brown@intel.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/37874
Reviewed-by: Sameer Nanda <snanda@chromium.org>

CHROMIUM: config: Don't pre-empt RT2800USB module in ARM

Allow RT2800USB module to build on ARM platforms.

Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chromium-os:35699
TEST=Boot system, do "ifconfig"

Change-Id: Ib24debfda9cfe2956a53656c3209fd6f9fcea3c9
Reviewed-on: https://gerrit.chromium.org/gerrit/38744
Reviewed-by: Olof Johansson <olofj@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

BACKPORT: drm/i915/dp: allow configuring eDP panel fitting scaling mode

LVDS allowed changing panel fitting scaling mode, while eDP didn't.
Copied relevant code from LVDS to eDP.
This also changed eDP default scaling mode to respect aspect ratio.

This patch was upstreamed as commits
53b41837935a4016852b30a6242a510e6927f9c7 and
8e740cd19fd3878453a3794e53c5252110697581,
which currently are on their way into linus' repo, and can be found in
git://people.freedesktop.org/~airlied/linux branch drm-next.

This version of the patch is a bit different from aforementioned commits,
but since backporting them would require cherry-picking 17 more commits,
it was decided to use this version instead.

BUG=chrome-os-partner:13682
TEST=Plug external monitor, switch to mirror mode,
check that aspect ratio is preserved (image is letterboxed/pillarboxed).

Signed-off-by: Yuly Novikov <ynovikov@chromium.org>
Change-Id: If1610945ccd1d69c3afa968916667fbb7de9ee40
Reviewed-on: https://gerrit.chromium.org/gerrit/33355
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

CHROMIUM: usb: dwc3: allow to configure the XHCI to use the internal clock

Remove the hardcoded assumption that we have an external PLL for the USB3 clock.

Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BUG=chrome-os-partner:16351
TEST=on Spring, plug on usb key on USB3 port and browse the content.

Change-Id: I0bf2a3604b3009abbcdfe3b0daf9d7fcccf1438c
Reviewed-on: https://gerrit.chromium.org/gerrit/38684
Tested-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Todd Broch <tbroch@chromium.org>
Reviewed-by: Olof Johansson <olofj@chromium.org>
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>

CHROMIUM: gobi: Move qcusbnet_put locking into free_dev

If the qcusbnet struct is freed through kobject_put instead of
qcusbnet_put, then we're not taking the lock that protects qcusbnet_list.
That's currently only possible if the struct is freed in cdev_put.

This bug was introduced in ae884f3 ("gobi: Keep struct qcusbnet alive
while cdev is alive").

BUG=chrome-os-partner:15849
TEST=suspend_stress_test

Change-Id: Ic34f7132e0e1a5724f02149b33f45ca71fa876f1
Signed-off-by: Michael Spang <spang@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38078
Reviewed-by: Ben Chan <benchan@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>

CHROMIUM: exynos: dts: do not use external PLL for USB3 on Spring

Move the enable gpio configuration for the USB3 PLL to the machine
device tree files, and do not define it on Spring, so the dwc3 driver
assumes that we only use the internal clock.

Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BUG=chrome-os-partner:16351
TEST=on Spring, plug on usb key on USB3 port and browse the content.

Change-Id: If8f91a34a0cadedada2ad627681a7fe83a6e5c12
Reviewed-on: https://gerrit.chromium.org/gerrit/38685
Tested-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Todd Broch <tbroch@chromium.org>
Reviewed-by: Olof Johansson <olofj@chromium.org>
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>

drm/exynos: do not support CONTIG gem buffer allocation

Currently, drm framework only supports allocation and mapping of
non-contig gem buffers. In the next versions of drm (v3.7),
this option to choose contig/non-contig is not supported. So for now,
we just return error in case user asks for contig gem buffers.

BUG=chrome-os-partner:15569
TEST=Built kernel and booted on snow.

Change-Id: Ibb164e812f4d175b6601145dffab1aaec28de4f5
Signed-off-by: Prathyush K <prathyush.k@samsung.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/38291
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Michael Spang <spang@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>
Tested-by: Akshu Agrawal <akshu@chromium.org>

UPSTREAM: perf tools: do not flush maps on COMM for perf report

This fixes a long-standing bug caused by the lack of separate COMM and EXEC
record types, which makes "perf report" lose track of symbols when a process
renames itself.

With this fix (suggested by Stephane Eranian), a COMM (rename) no longer
flushes the maps, which is the correct behavior.  An EXEC also no longer
flushes the maps, but this doesn't matter because as new mappings are created
(for the executable and the libraries) the old mappings are automatically
removed.  This is not by accident: the functionality is necessary because DLLs
can be explicitly loaded at any time with dlopen(), possibly on top of existing
text, so "perf report" handles correctly the clobbering of new mappings on top
of old ones.

An alternative patch (which I proposed earlier) would be to introduce a
separate PERF_RECORD_EXEC type, but it is a much larger change (about 300
lines) and is not necessary.

Signed-off-by: Luigi Semenzato <semenzato@chromium.org>
Tested-by: Stephane Eranian <eranian@google.com>
Acked-by: Stephane Eranian <eranian@google.com>
Cc: "Eric W. Biederman" <ebiederm@xmission.com>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Cc: Alexander Viro <viro@zeniv.linux.org.uk>
Cc: Andi Kleen <ak@linux.intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Lucas De Marchi <lucas.demarchi@profusion.mobi>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Olof Johansson <olofj@chromium.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Robert Richter <robert.richter@amd.com>
Cc: Sonny Rao <sonnyrao@chromium.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Stephen Wilson <wilsons@start.ca>
Cc: Tejun Heo <tj@kernel.org>
Cc: Vasiliy Kulikov <segoon@openwall.com>
Link: http://lkml.kernel.org/r/1345585940-6497-1-git-send-email-semenzato@chromium.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
(cherry picked from commit 9fdbf671ba7e8adb2cbb93d716232ebcab55f6bd)

BUG=chromium-os:26115
TEST=manual (see bug/original commit in 3.0)

Change-Id: Ie1fe97535430a989c3c925b30aa90e462e6eb37d
Signed-off-by: Olof Johansson <olofj@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38676
Reviewed-by: Luigi Semenzato <semenzato@chromium.org>

drm/exynos: dp: Reduce video stream detection time

Reduce the time we take for video stream detection by removing the
"done_count" wait of 1.1s since it's not blocking anything from
happening.

Also replace the msleep(100) with a usleep_range since we don't need to
wait 100ms to retry the detection.

BUG=chrome-os-partner:15261
TEST=Tested on snow, display on time reduced from 1.3s to 90ms

Signed-off-by: Sean Paul <seanpaul@chromium.org>
Change-Id: Ib4ef14352256b6b6627a3dff8bf77db2f368e547
Reviewed-on: https://gerrit.chromium.org/gerrit/38470
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

watchdog: s3c2410: Avoid excessive petting during cpufreq

On more modern Samsung SoCs (like exynos5) the watchdog is
parented on a clock that doesn't change.  Avoid all of the extra
petting, disabling, and re-enabling of the watchdog in the case
that the clock never changes.

NOTE: As pointed out in
  <https://gerrit.chromium.org/gerrit/#/c/31465/>

...the proper way would be to register a notifier for the clock
directly, but CONFIG_COMMON_CLK isn't enabled.

BUG=chrome-os-partner:9403
TEST=If I open /dev/watchdog now it eventually dies.

Change-Id: I4f982dc85caf091444633aff94b90d524bdfe829
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/32566
Reviewed-by: Olof Johansson <olofj@chromium.org>
Commit-Ready: Grant Grundler <grundler@chromium.org>
Reviewed-by: Grant Grundler <grundler@chromium.org>
Tested-by: Grant Grundler <grundler@chromium.org>

CHROMIUMOS: mwifiex: Disable transmit aggregation

Disable requesting transmit aggregation.  In situations where
the firmware receives a particular type of ADDBA response, it
enters a Block-Ack-Request loop and no further frames are
transmitted.  Until this larger issue is addressed, we cannot
safely request aggregation.

Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chrome-os-partner:12192
TEST=Associate with modified mac80211-based AP which returns
a Block-Ack-ADDBA response with only 1 buffer allowed.

Change-Id: Icb532a118bb90eca867784fd9b05bf850404e974
Reviewed-on: https://gerrit.chromium.org/gerrit/38483
Reviewed-by: Christopher Wiley <wiley@chromium.org>
Reviewed-by: mukesh agrawal <quiche@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

CHROMIUM: gobi: Fix qcusbnet leak in devqmi_open

We're taking a reference to qcusbnet in devqmi_open and not releasing
it in the error path. This plugs that leak.

BUG=chrome-os-partner:16158
TEST=suspended the machine a few times, started and stopped cromo

Change-Id: I729aa5160baa4a64beaa1b7b92ac39e35319f799
Signed-off-by: Michael Spang <spang@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37953
Reviewed-by: Ben Chan <benchan@chromium.org>
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>

CHROMIUM: gobi: Fix qcusbnet leak in QMI_CLOSE ioctl

The handle cleanup path in the QMI_CLOSE ioctl leaks a reference to
qcusbnet (taken in devqmi_open). This patch unifies that cleanup code
with the code from devqmi_release and fixes the leak.

BUG=chrome-os-partner:16158
TEST=suspended the machine a few times, started and stopped cromo

Change-Id: I970a20fd67ebf8d1e65526ccca41f13c2aa97b44
Signed-off-by: Michael Spang <spang@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37955
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Ben Chan <benchan@chromium.org>

CHROMIUM: gobi: Remove leaky replacement of fops in devqmi_write

The devqmi_write function sometimes overwrites devqmi_fops with
def_chr_fops. When that happens, the devqmi_fops reference is leaked,
and the file is never cleaned up because devqmi_release is not called.

It's not clear what this is for, and it looks like the other file
callbacks already check device_valid() before doing anything.
Therefore we can probably just remove this.

BUG=chrome-os-partner:16158
TEST=suspend_stress_test; stop cromo; cat /sys/module/gobi/refcnt

Change-Id: Ie8aecf98784165629864c90b60ee1c20d2106bef
Signed-off-by: Michael Spang <spang@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38058
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>

drm/exynos: Clean up fimd mode handling

FIMD has been using the panel timing data straight from the driver's
platform data instead of the data that is provided via the drm layer.
This makes things difficult if we want to use EDID data from the dp
driver, since it would be ignored. This patch changes FIMD to just
use the platform panel data as a seed to get_panel. From then on, we use
the overlay data that is set via modeset.

This patch also removes the panel array, since we only want one
resolution from platform data.

BUG=chrome-os-partner:11158
TEST=Tested on snow, no regressions noted

Change-Id: I345c3663509e6820a745b85c4b9a37f77f42c0d2
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38409
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: dp: Check hp gpio in hp detect function

Check the hotplug gpio in the hpd function so we know when it's
asserted. This fixes the return value of the is_connected callback.

BUG=chrome-os-partner:16280
TEST=None

Change-Id: I1ab0c7d830ecdc7b7f38ffd81895038978e4e57a
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38319
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>

drm/i2c: ptn3460: Add dev node check to ready wait

Check for the ptn dev node in the ready wait function to avoid blocking
boards without ptn for 30 seconds.

BUG=chrome-os-partner:16280
TEST=None

Change-Id: I0057e7b72d9841447a56167617b9f081f5bbbe18
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38317
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>

CHROMIUM: config: Add RT2800USB module

Add support for the RT2800 USB WiFi device.

Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chromium-os:35699
TEST=Boot system, do "ifconfig"
CQ-DEPEND=Iea82c1e6b1d7f9c5181cef11fcf25a5ee2da114b

Change-Id: Ic02543da81191c8e6bf39e9e90d8fc2c86128b39
Reviewed-on: https://gerrit.chromium.org/gerrit/36834
Reviewed-by: Olof Johansson <olofj@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

drm/exynos: Let drm handle edid allocations

There's no need to allocate edid twice and do a memcpy when drm helpers
exist to do just that. This patch cleans that interaction up, and
doesn't keep the edid hanging around in the connector.

BUG=chromium-os:36513
TEST=Tested on snow boot/hotplug/idle suspend/lid suspend

Change-Id: Ifb971a7a6c6ae66ce431b9a256886d9699adbd09
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38406

drm/exynos: Add drm notification of hotplug to dp

Instead of polling the connector for hotplug, use the hotplug workqueue
to fire a drm hotplug notification when we detect a cable change.

This will also update userspace when the state has changed, and notify
fbdev.

At this point in time, the patch should be a no-op, but it's important
if we want to use EDID to mode set.

BUG=chrome-os-partner:11158
TEST=Tested on snow, no regressions detected

Change-Id: Ib6b0c789f6abc45a118a04dcb8a73e8e35fd4a85
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38408
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

Revert "DRM/exynos: Fix Memory Leak: free EDID block returned by drm_get_edid()."

This reverts commit e04984d087c95d7dd09be7cb6cb84e1f339c4f17. The patch causes kernel panic since it frees that memory twice once in the panel_op, and then again in the connector function.

Change-Id: I26173e1664733acea4abad9a731c4f7d98efe18a
Reviewed-on: https://gerrit.chromium.org/gerrit/38405
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>
Commit-Ready: Sean Paul <seanpaul@chromium.org>
Tested-by: Sean Paul <seanpaul@chromium.org>

CHROMIUM: platform: rename nvram to vbc

As we are actually accessing vboot context and vboot context is not
limited to be stored on nvram, it is clearer to just call it vbc.

This change also removes obsolete comments and makes API interface
consistent (all read/write should return number of bytes instead of zero
on success).

BUG=none
TEST=manual, add debugging sysfs entry to read/write vboot context and
     make sure it works.

Change-Id: Ifd097a1a8af9edb776eed83fe47cb332f8a3445d
Signed-off-by: Che-Liang Chiou <clchiou@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37686
Reviewed-by: Simon Glass <sjg@chromium.org>

CHROMIUM: vmx: control vmx correctly on resume

On x86 chromeos systems with vmx support, the feature is controlled by
the disablevmx flag. There are several boot paths for the CPUs
and this patch ensures that the resume path for the BSP is covered.

Rename cpu_disable_vmx to cpu_control_vmx. It's a bit more
descriptive since it does not disable vmx in all cases.
Further, have the function itself test the variable; this
is more in keeping with how other such capabilities
are controlled anyway and it keeps the variable in
file scope. Finally, make it all conditional on
CONFIG_CHROMEOS.

We use the variable to control the enable/disable instead
of saved CPU context. I could add the FEATURES MSR
to the saved context and use that on restore but
that adds more changes, and I've seen no evidence that anyone
outside chromeos is interested in disabling VMX, so it
makes more sense to me to limit the footprint of this
patch.

BUG=chrome-os-partner:10449
TEST=apply and boot with this patch.
dmesg | grep Disabling.VMX
Note that all four CPUs have it disabled.
Also
rdmsr 0 0x3a
and see the value.
Suspend any way you want. (powerd_suspend, close the lid, whatever).
Repeat and notice the Disabling message appears again,
for all four cores this time,
and the 0x3a MSRs all have the correct value.

Change-Id: Ie0160db31a1a058ce4d3e4f8752cd0e5d696ea84
Signed-off-by: Ronald G. Minnich <rminnich@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38235
Reviewed-by: Olof Johansson <olofj@chromium.org>
Reviewed-by: Kees Cook <keescook@chromium.org>

DRM/exynos: Fix Memory Leak: free EDID block returned by drm_get_edid().

This was just posted on the dri-devel list, and affects us also. It's
not a direct cherry-pick since we moved the code around.

BUG=none
TEST=by hand

Change-Id: Ibd76bc48094e466fe574af23d37c8df1e0262a8d
Signed-off-by: Egbert Eich <eich@suse.de>
Reviewed-on: https://gerrit.chromium.org/gerrit/38334
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Commit-Ready: Stéphane Marchesin <marcheu@chromium.org>
Tested-by: Stéphane Marchesin <marcheu@chromium.org>

CHROMIUM: ath9k: do not schedule rx poll on invalid state

During system suspend, the rx poll timer is cancelled at
the driver stop. But occasionally it will be scheduled again by
beacon reception before stopping rx. Since the timer is never
stopped again, it tries to queue work on suspend path.
This patch ensures that rx poll timer is cancelled on driver stop.

Signed-off-by: Rajkumar Manoharan <rmanohar@qca.qualcomm.com>
BUG=chrome-os-partner:15168
TEST=Boot, run suspend-resume test with powersave ON.

Change-Id: I2eae50a4b35335249f0a530c9eb6161951abe5fc
Signed-off-by: Ashok Nagarajan <asnagarajan@chromium.org>
Tested-by: Ashok Nagarajan <asnagarajan@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38316
Reviewed-by: mukesh agrawal <quiche@chromium.org>
Reviewed-by: Paul Stewart <pstew@chromium.org>

arm: exynos5: Add Adaptive Body Bias Control for exynos5

ABB improves operation voltage margin and reduces leakage current
of logic devices. ABB control depends on ASV group.

BUG=chrome-os-partner:15327
TEST=suspend_stress_test along with poweroffs and reboots

Change-Id: I6c7c405cacdf36d6a2e01d76b046e9dc2c112f60
Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Signed-off-by: Jun Kim <junkim@samsung.com>
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37619

UPSTREAM: mwifiex: report error to MMC core if we cannot suspend

When host_sleep_config command fails we should return error to
MMC core to indicate the failure for our device.

The misspelled variable is also removed as it's redundant.

Cc: "3.0+" <stable@vger.kernel.org>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chrome-os-partner:15129
TEST=Boot, associate, speed test, suspend, resume

Change-Id: I9d53425b29050274840030e63f95969601ee622f
Reviewed-on: https://gerrit.chromium.org/gerrit/38285
Reviewed-by: Bing Zhao <bzhao@marvell.com>
Reviewed-by: mukesh agrawal <quiche@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

UPSTREAM: mwifiex: fix system hang issue in cmd timeout error case

Reported by Tim Shepard:
I was seeing sporadic failures (wedgeups), and the majority of those
failures I saw printed the printouts in mwifiex_cmd_timeout_func with
cmd = 0xe5 which is CMD_802_11_HS_CFG_ENH.  When this happens, two
minutes later I get notified that the rtcwake thread is blocked, like
this:
      INFO: task rtcwake:3495 blocked for more than 120 seconds.

To get the hung thread unblocked we wake up the cmd wait queue and
cancel the ioctl.

Cc: "3.4+" <stable@vger.kernel.org>
Reported-by: Tim Shepard <shep@laptop.org>
Signed-off-by: Bing Zhao <bzhao@marvell.com>
Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chrome-os-partner:15129
TEST=Boot, associate

Change-Id: Icdf3dd71b12b8110c5b13df5c5c9ebbb0820441c
Reviewed-on: https://gerrit.chromium.org/gerrit/38284
Reviewed-by: Bing Zhao <bzhao@marvell.com>
Reviewed-by: mukesh agrawal <quiche@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

CHROMIUM: ASoC: samsung: fix error for hdmi-audio

The result of the device tree look-up for the hdmi-audio device
was checked against the wrong variable.
As a consequence, if the hdmi-audio device is disabled, the kernel will
crash at startup instead of skipping the plugin.

Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BUG=chrome-os-partner:14489
TEST=boot on Spring with hdmi-audio disabled in the device tree
and see we are longer crashing in the max98095 probing.

Change-Id: I613b53633eb7814e63a7459704f3780ded62a2a0
Reviewed-on: https://gerrit.chromium.org/gerrit/38261
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>
Tested-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Dylan Reid <dgreid@chromium.org>

Documentation: mfd/max77686 devicetree bindings update

This change documents the enable/disable-low-jitter
bindings for the mfd/max77686.

Signed-off-by: Will Drewry <wad@chromium.org>
TEST=none
BUG=chrome-os-partner:15785

Change-Id: I0c980d5a2fa730c8d8681041a5627d84c3f1f4bb
Reviewed-on: https://gerrit.chromium.org/gerrit/38238
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>
Reviewed-by: Olof Johansson <olofj@chromium.org>
Reviewed-by: Doug Anderson <dianders@chromium.org>
Commit-Ready: Will Drewry <wad@chromium.org>
Tested-by: Will Drewry <wad@chromium.org>

CHROMIUM: staging: gdm72xx: protect access of rx / tx structs

This patch applies spinlock to protect access to rx / tx structs in
certain call sites, which fixes the following crash in gdm_suspend.
It also fixes usb_set_intfdata() in gdm_usb_probe to avoid setting an
already freed phy_dev.

<5>[ 4996.815018] [<7f0074b0>] (gdm_suspend+0x1c/0x2b4 [gdmwm]) from [<803020a4>] (usb_suspend_both+0x80/0x1a0)
<5>[ 4996.815055] [<803020a4>] (usb_suspend_both+0x80/0x1a0) from [<80302c84>] (usb_runtime_suspend+0x38/0x64)
<5>[ 4996.815089] [<80302c84>] (usb_runtime_suspend+0x38/0x64) from [<802becc0>] (__rpm_callback+0x48/0x78)
<5>[ 4996.815118] [<802becc0>] (__rpm_callback+0x48/0x78) from [<802bf8dc>] (rpm_suspend+0x394/0x5ec)
<5>[ 4996.815145] [<802bf8dc>] (rpm_suspend+0x394/0x5ec) from [<802c0550>] (pm_runtime_work+0x8c/0xa4)
<5>[ 4996.815177] [<802c0550>] (pm_runtime_work+0x8c/0xa4) from [<800456cc>] (process_one_work+0x264/0x438)
<5>[ 4996.815209] [<800456cc>] (process_one_work+0x264/0x438) from [<80045acc>] (worker_thread+0x22c/0x3b8)
<5>[ 4996.815239] [<80045acc>] (worker_thread+0x22c/0x3b8) from [<8004a43c>] (kthread+0x9c/0xa8)
<5>[ 4996.815270] [<8004a43c>] (kthread+0x9c/0xa8) from [<8000f160>] (kernel_thread_exit+0x0/0x8)
<0>[ 4996.815295] Code: e92d4000 e8bd4000 e2800020 eb4ab9a1 (e5905000)

BUG=chrome-os-partner:16223
TEST=Run network_MobileSuspendResume.wimax_stress test.

Change-Id: I5a45f7e5fc3ca35681005fea40303f4c8c017ac0
Signed-off-by: Ben Chan <benchan@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38172
Reviewed-by: Sameer Nanda <snanda@chromium.org>

CHROMIUM: add DEVTMPFS_SAFE support

If DEVTMPFS_SAFE==y, automount devtmpfs with nosuid and noexec.

BUG=chromium-os:32629
TEST=adhoc
mount | grep devtmpfs

Change-Id: I7a330fb7bd1e0da7e42fe2194a958c6878c0ce68
Signed-off-by: Elizabeth Fong-Jones <ellyjones@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38061
Reviewed-by: Olof Johansson <olofj@chromium.org>

CHROMIUM: cros5250-common: add enable-low-jitter to max77686

Signed-off-by: Will Drewry <wad@chromium.org>
BUG=chrome-os-partner:15785
TEST=boots with change 38180 and i2cget returns 0x0f as per that change.

Change-Id: I70f305c23ee8205f9af3e81addbcaa350434a27a
Reviewed-on: https://gerrit.chromium.org/gerrit/38229
Reviewed-by: Doug Anderson <dianders@chromium.org>
Commit-Ready: Will Drewry <wad@chromium.org>
Tested-by: Will Drewry <wad@chromium.org>

CHROMIUM: mfd/max77686: enable low-jitter mode

This change adds three valid device-tree configured settings
for the max77686 "low jitter" mode:
* max77686,enable-low-jitter
* max77686,disable-low-jitter
* (no entry)

With either entry, the third bit of the 32KHZ register
is updated.

Low jitter mode is desirable on some platforms with components that are
sensitive to the clock signal.  The clock sometimes appears to stutter when
being updated.  For most uses, this is inconsequential, but if the RTC is
supplying a signal to more sensitive hardware, it can cause undesirable
results.

BUG=chrome-os-partner:15785
TEST=It boots :) semenzato@ has confirmed that if the register is set, the errors do not reproduce with the scope.

To check the register state:
  $ i2cget -f -y 0 0x9 0x7f
  0x0f

Reported-by: Luigi Semenzato <semenzato@chromium.org>
Suggested-by: Luigi Semenzato <semenzato@chromium.org>
Signed-off-by: Will Drewry <wad@chromium.org>
Change-Id: Ib40d1d0728535b10c8438eaa6eacd74fb83cb697
Reviewed-on: https://gerrit.chromium.org/gerrit/38180
Reviewed-by: Doug Anderson <dianders@chromium.org>
Commit-Ready: Will Drewry <wad@chromium.org>
Tested-by: Will Drewry <wad@chromium.org>
Reviewed-by: Olof Johansson <olofj@chromium.org>

CHROMIUM: ASoC: samsung: Spring is using max98095 codec

Spring is also using the Maxim max98095 codec for audio.
Add it to the list of supported boards.

Signed-off-by: Vincent Palatin <vpalatin@chromium.org>
BUG=chrome-os-partner:14490
TEST=on Spring, plug a headset, play a video on Youtube and hear the
sound.

Change-Id: If17a3ba240943412acf03b24058218c40781944f
Reviewed-on: https://gerrit.chromium.org/gerrit/38205
Reviewed-by: Dylan Reid <dgreid@chromium.org>
Commit-Ready: Vincent Palatin <vpalatin@chromium.org>
Tested-by: Vincent Palatin <vpalatin@chromium.org>

CHROMIUM: md: bootcache save string args

dm-bootcache now makes a separate copy of any string arguments
that are passed to it.

BUG=chromium-os:35683
TEST=ran bootcachetest on stumpy, arm

Change-Id: I48f62671377109a596eea150361c492fe19a96a7
Signed-off-by: Paul Taysom <taysom@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37736
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>

drm/exynos: Implement DPMS_STANDBY for HDMI

Enable the "blue screen" when we get a DPMS_STANDBY request for HDMI.
This allows us to turn off the mixer, but maintain audio over HDMI when
we turn the screen idle.

BUG=chrome-os-partner:15709
TEST=Tested on snow by ensuring audio continues to play during idle

Change-Id: I2639c0d1cd7384881b5fd651039fef71b27c76ed
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38122
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>

UPSTREAM: ath9k_hw: Fix wrong peak detector DC offset

An issue is reported in AR9462 that NF_cal_not_done is not observed
when HW peak detector calibration is disabled. At that state, the HW
is stuck at NF calibration which prevents tx output. The root cause
is wrong peak detector offset calibrated by HW. To resolve this issue,
peak detector calibration is done manually by SW for AR9462.

Signed-off-by: Rajkumar Manoharan <rmanohar@qca.qualcomm.com>
Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chrome-os-partner:36266
TEST=Boot and run bandwith tests

Change-Id: I9b79e9b51ef86cc73800b7bb3e80875b1c0e2705
Reviewed-on: https://gerrit.chromium.org/gerrit/38025
Reviewed-by: mukesh agrawal <quiche@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

CHROMIUM: mwifiex: Add SDIO register debugging

This patch retrieves firmware/hardware status upon tx watchdog
timeout or command timeout. Hopefully with this debug patch we
can get some clues on what is causing the timeout issue.

BUG=chrome-os-partner:15129
TEST=Boot, connect to network

Change-Id: I058fe07746e6a1ccd8debde0c0d3d8debd700502
Signed-off-by: Paul Stewart <pstew@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37972

CHROMIUMOS: Revert "mwifiex: Set timeout for mwifiex_wait_queue_complete"

This was meant to isolate issues with command timeouts for debugging
purposes.  We are proceeding with a different debugging path now.

This reverts commit e0514c155f95de286656a3d4306df852ea107987.

Signed-off-by: Paul Stewart <pstew@chromium.org>
BUG=chrome-os-partner:15129
TEST=Boot, connect to network

Change-Id: I9b2fa79803605bc8b92d93a57fb0c850113c9fac
Reviewed-on: https://gerrit.chromium.org/gerrit/38119
Reviewed-by: mukesh agrawal <quiche@chromium.org>
Commit-Ready: Paul Stewart <pstew@chromium.org>
Tested-by: Paul Stewart <pstew@chromium.org>

CHROMIUM: config: enable usb net drivers (dm9601/net1080/rtl8150) for all targets

Some configs had these drivers enabled while others did not.  Update the
common arch configs to all enable these drivers.

BUG=chromium-os:35943
TEST=`emerge-daisy chromeos-kernel` included these three drivers
TEST=`emerge-stumpy chromeos-kernel` included these three drivers
TEST=`emerge-x86-alex chromeos-kernel` included these three drivers

Change-Id: I9615109a5b17af293855595ba8f06f397db360c4
Signed-off-by: Mike Frysinger <vapier@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37296
Reviewed-by: Doug Anderson <dianders@chromium.org>
Reviewed-by: Grant Grundler <grundler@chromium.org>

CHROMIUM: arm: exynos: fix math in ppmu driver

The define doesn't wrap the argument properly so the math ends up wrong.

BUG=chromium-os:36402
TEST=check /proc/cpu/alignment and "perf top -G --symbols do_alignment_ldrstr" after applying

Change-Id: Ifd342de96c682bc2a2acaf7a9bfb73f0d5d2f516
Signed-off-by: Olof Johansson <olofj@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/38123
Reviewed-by: Doug Anderson <dianders@chromium.org>
Reviewed-by: Bryan Freed <bfreed@chromium.org>

CHROMIUM: mfd: chromeos_ec: support command version numbers

ec supports versioned command and non-versioned command should be
deprecated.  To minimize changes to ec clients, driver will accept not
8-bits but 16-bits command code and interpret top 8-bits as version
number.  Because current ec clients all use commands of version-zero,
they will not be effected by this API change.

BUG=chrome-os-partner:15609
TEST=manual, add debugging sysfs entry to exercise kernel code path, and
     make sure ec driver still works on snow.

Change-Id: I640e7756baddd20d8bc14c7f8ba10c13e60ed8ff
Signed-off-by: Che-Liang Chiou <clchiou@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37496
Reviewed-by: Simon Glass <sjg@chromium.org>

CHROMIUM: Input: atmel_mxt_ts - recalibrate on system resume

In the suspend path, if we don't allow device wakeup, we put the
chip into deepsleep mode and the chip stops scanning during suspen.
On resume if the environment changes, the calibrated baseline before
suspend will no longer be valid.

In this patch we force a recalibration on resume if device wakeup is
disabled during suspend to handle the environment change.

Signed-off-by: Yufeng Shen <miletus@chromium.org>
BUG=chrome-os-partner:16171
TEST=I don't have a controlled environment to test this. So I only test
     normal suspend/resume to make sure no noise touches happen on resume
     and touch devices work as expected.
     1. With lid open, using powerd_suspend to suspend the system. Wakeup
     the system and make sure the touch device still works.
     Run "demsg | grep atmel"  to make sure no calibration message reported.
     2. Use lid close to suspend the system. Wakeup the system and make
     sure touch device still works.
     Run "demsg | grep atmel"  to make sure calibration message are reported.
     3. Also notice the case of lid open caused system resume, if something is
     on the touch surface (like opening the lid and quickly put the palm on the
     touch surface for a while), the system will get calibrated into a wrong
     baseline and touch device then won't work.

Change-Id: I62cb47fa1c97917a2c0f968e41ee4cd13f12187c
Reviewed-on: https://gerrit.chromium.org/gerrit/38051
Reviewed-by: Benson Leung <bleung@chromium.org>
Commit-Ready: Yufeng Shen <miletus@chromium.org>
Tested-by: Yufeng Shen <miletus@chromium.org>

CHROMIUM: Input: cyapa - Remove CYAPA_CMD_GROUP_CTRL

Remove CONTROL_BASE define and CYAPA_CMD_GROUP_CTRL
command class.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=builds clean.

Change-Id: I6f62508f70af2c0f91c8d3d5c97cc3cdd5439034
Reviewed-on: https://gerrit.chromium.org/gerrit/37741
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>

arm: exynos: Retry "wfi" on failure

We are seeing suspend failures occuring once in a while. In case
there still exists a race with another interrupt let us retry "wfi"
a 100 times before failing.

BUG=chrome-os-partner:14956
TEST=suspend_stress_test with no suspend failures yet.

Change-Id: I1c59ae77e12be40dec1ee027f8f1d0e0a2a2f66c
Signed-off-by: Abhilash Kesavan <a.kesavan@samsung.com>
Reviewed-on: https://gerrit.chromium.org/gerrit/37991
Reviewed-by: Doug Anderson <dianders@chromium.org>
Reviewed-by: Jon Kliegman <kliegs@chromium.org>
Commit-Ready: Jon Kliegman <kliegs@chromium.org>
Tested-by: Jon Kliegman <kliegs@chromium.org>

CHROMIUM: pm-check: make pm-check sum function stronger

Add a rotate to the sum function and switch back to 32-bit.
The rotate seems to help in catching more instances of corruption than
a pure checksum.  So far we haven't seen sum + ror miss a corruption
in over 2000 known cases whereas pure sum misses 10 of those.  There
seems to be some kind of optimizer bug when I try to use 64-bit + ror
so let's go back to 32-bit and take a speed hit.  This code seems to
run in about 760 msec vs about 540 msec for the pure checksum.

Signed-off-by: Sonny Rao <sonnyrao@chromium.org>
BUG=chrome-os-partner:16148
TEST=run suspend_stress_test, ensure errors are still detected

Change-Id: I0adaec6f73d9acd752059840f37f707d5d2a2f59
Reviewed-on: https://gerrit.chromium.org/gerrit/38053
Reviewed-by: Doug Anderson <dianders@chromium.org>
Commit-Ready: Sonny Rao <sonnyrao@chromium.org>
Tested-by: Sonny Rao <sonnyrao@chromium.org>

drm/exynos: add the plumbing for drm debugfs

Also added simple gem object counting.

BUG=chromium-os:36060
TEST=Verified that the debugfs interface is present.

localhost ~ # cat /sys/kernel/debug/dri/0/exynos_gem_objects
5 objects, 12709888 bytes
localhost ~ # stop ui
ui stop/waiting
localhost ~ # cat /sys/kernel/debug/dri/0/exynos_gem_objects
1 objects, 4227072 bytes

Change-Id: I4652eedbdda5be68f055b45e316c999114d80b3b
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37858

CHROMIUM: Input: cyapa - Remove CYAPA_MAX_TOUCHES define

It's only used once in struct cyapa_reg_data, so lets just
hard code it there.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=builds clean.

Change-Id: I6d39e70f369b0825238022a38111e2fa42862c5e
Reviewed-on: https://gerrit.chromium.org/gerrit/37745
Reviewed-by: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>

CHROMIUM: Input: cyapa - compare return value of memcmp to 0

Check the return value of memcmp for more clarity.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=builds clean

Change-Id: Ie2235a59df9347d938917e86d2c69d27af073338
Reviewed-on: https://gerrit.chromium.org/gerrit/37744
Reviewed-by: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>

CHROMIUM: Input: cyapa - change to xy_hi and x_lo y_lo

Based on upstream feedback, change these variable names.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=builds clean.

Change-Id: Ieb16c22fcc95c657f493f9f8c69854aa475e391e
Reviewed-on: https://gerrit.chromium.org/gerrit/37743
Reviewed-by: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>

CHROMIUM: Input: cyapa - Remove gen enum

Remove the gen enum. The other two values of the enum
aren't used anywhere in this driver (we only support Gen3).

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=Builds clean.

Change-Id: If1f26dbccb64fc852151d8c5de6ea3577213f5f9
Reviewed-on: https://gerrit.chromium.org/gerrit/37742
Reviewed-by: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>

CHROMIUM: cpufreq_interactive:  correct cpu idle time calculation

Each cpu has a timer that tracks load and decides whether to adjust the
clock for that processor.  The timer callback may, however, execute on
any cpu, not just the one it manages.  Fix the idle time calculation to
take this into consideration--it was assuming smp_processor_id() returned
the cpu id of the processor to manage, causing incorrect decisions.

Signed-off-by: sleffler@chromium.org
BUG=none
TEST=run cpu intensive benchmark like octane on a multi-core system; check trace results and note that sometimes v8 runs for extended times on cpu operating at low freq (contributing to poor score)

Change-Id: I2b5c21b5d8cae31679e844ac67037cc760c5dfc0
Reviewed-on: https://gerrit.chromium.org/gerrit/37830
Reviewed-by: Sonny Rao <sonnyrao@chromium.org>
Reviewed-by: Sameer Nanda <snanda@chromium.org>
Tested-by: Sam Leffler <sleffler@chromium.org>
Commit-Ready: Sam Leffler <sleffler@chromium.org>

CHROMIUM: ARM: MALI: add sysfs file which prints amount of memory allocated

BUG=chromium-os:36060
TEST=Verified that the sysfs file is present.

localhost ~ # cat /sys/devices/platform/mali.0/memory
122068992 bytes
localhost ~ # stop ui
ui stop/waiting
localhost ~ # cat /sys/devices/platform/mali.0/memory
28672 bytes

Change-Id: Ic11cf2470a6e17676da48ce4aa3a73444a475942
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37862

drm/exynos: make event parameter optional for page_flip

We no longer need to enforce that event is non-NULL. We also need
to be able to call page_flip with a NULL event so that we can
call it from the mode_set paths.

BUG=chrome-os-partner:15349,chrome-os-partner:14965
TEST=Multiple VT switch, HDMI plug/unplug, sign in/out, suspend/resume,
     idle suspend/resume.

Change-Id: Ia139cd36a3f78fbf3c96651176a42289f07722e6
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37572

Revert "drm/exynos: implement an optimized page_flip"

This reverts commit 69518a3c650de9186f345a5e7feaddbadfc394d0.

We want to be able to call page_flip from mode_set and mode_set_base
so that all functions which cause a buffer to be scanned out are
synchronized.

BUG=chrome-os-partner:15349,chrome-os-partner:14965
TEST=Multiple VT switch, HDMI plug/unplug, sign in/out, suspend/resume,
     idle suspend/resume.

Change-Id: I198962b65bb8a1ab5b885c053c19f7f8077cb9c6
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37571

CHROMIUM: drm/i915: Propermy terminate DMI list

Not sure why this wasn't triggered by our testing, but
Intel reports that they had the issue.

BUG=chromium-os:36276
TEST=by hand

Change-Id: Iad63bca07eab05e750bb165387f028e1dbf3de54
Reviewed-on: https://gerrit.chromium.org/gerrit/37921
Reviewed-by: Stuart Abercrombie <sabercrombie@chromium.org>
Reviewed-by: Russell Webb <russell.webb@intel.com>
Tested-by: Russell Webb <russell.webb@intel.com>
Reviewed-by: Olof Johansson <olofj@chromium.org>
Commit-Ready: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: move checks outside of exynos_drm_crtc_update

Refactor the code so that exynos_drm_crtc_update cannot fail.

BUG=chrome-os-partner:15349,chrome-os-partner:14965
TEST=Multiple VT switch, HDMI plug/unplug, sign in/out, suspend/resume,
     idle suspend/resume.

Change-Id: I77335d6142f2363ca19b113ddcd5909c3a067121
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37570
Reviewed-by: Sean Paul <seanpaul@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: remove NULL check from exynos_drm_overlay_update

We really shouldn't be handling this NULL pointer. If there
is a NULL pointer here, we've hit a kernel bug. No need for a
BUG_ON since the kernel will panic on any NULL dereference.

This allows us to make exynos_drm_overlay_update void.

BUG=chrome-os-partner:15349,chrome-os-partner:14965
TEST=Multiple VT switch, HDMI plug/unplug, sign in/out, suspend/resume,
     idle suspend/resume.

Change-Id: I39c4ee44260eaeb98594dfbe01f2b885039ab23c
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/35248
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

KDS: fix deadlock in early release path

We deadlock, under the following scenario:

1) A grabs lock non-exclusive
2) B grabs lock exclusive
3) C grabs lock non-exclusive
4) B releases lock before ever acquiring
5) A releases lock

After 5, C should acquire the lock but the current implementation
does not do this.

BUG=chromium-os:36071,chrome-os-partner:15349,chrome-os-partner:14965
TEST=I had some patches which were seeing the deadlock. After this fix,
     I no longer get a deadlock.
TEST=Multiple VT switch, HDMI plug/unplug, sign in/out, suspend/resume,
     idle suspend/resume.

Change-Id: I4df35f08f9f31ad95fbe686e48bf421e25f9930b
Signed-off-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37569
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/ptn-exynos: Synchronize PTN init with hotplug

Gate the DP hotplug work function on ptn3460 initailization if the
bridge chip is present. This ensures we can completely set up the
hardware before trying to read EDID and configure video.

BUG=chrome-os-partner:11158
TEST=Tested on snow, boot/suspend-resume/idle suspend-resume

Change-Id: I7f2ca812bdaaaba5350deed29939afec5d4e8a23
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/36794

drm/ptn3460: Remove free_me

This free_me variable is ugly and useless. Since we don't even define
the platform_data struct outside of this driver, it'd be hard for
arch-specific code to instantiate it.

BUG=None
TEST=Tested on snow

Change-Id: I87564b43d53c2ec02bdf6a4ef73cc7f69c4a2ce9
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37624

drm/exynos: Consolidate suspend/resume

Remove all of the pm_ops from the various platform devices and move to a
consolidated suspend/resume process from the drm driver. This allows us
to properly order operations when we suspend/resume. It also makes the
dpms tracking in exynos_drm_encoder meaningful.

BUG=None
TEST=Tested with/without hdmi & idle/lid suspend/resume. No regressions.

Change-Id: I25bab3843c87960416b1218a48c60c5124404aff
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37642
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: dp: Mask interrupts instead of disable

Instead of running disable_irq on suspend, and subsequently enable_irq
on resume, just mask off the interrupt before suspending. This avoids
the ugly disable_irq in probe, and any future issues with mismatched irq
enable/disables.

BUG=None
TEST=Tested on snow idle suspend/lid suspend, boot, etc.

Change-Id: I0e913b1c563b3b0081c074eac52c8d887d5dca58
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37621
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: Convert dp driver to use panel_ops

This patch changes the dp driver to implement panel_ops. This gets
rid of the dp_dev hack that was introduced when dp moved into drm.

BUG=chrome-os-partner:11158, chrome-os-partner:9855,
    chrome-os-partner:10717
TEST=Tested on snow, no regressions detected

Change-Id: Ia0e2939f635a2406a1ac29e172cb1ae59d13acbe
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/36701
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: Refactor manager/display/overlay

The exynos_drm_manager/display/overlay architecture wasn't really
serving its purpose. It attempted to split callbacks into logical
sections, but those sections didn't really make sense, and required us
to have a layer of abstraction such as exynos_drm_hdmi on top of the
controller and panel to call the appropriate functions.

This patch removes manager/display/overlay as well as exynos_drm_hdmi,
and adds a new abstraction which splits the callbacks to controller and
panel.

Right now, this just cleans things up, making the code a little more
readable. In the short term, it allows us to hook fimd into just
the controller side, and add dp/mipi as panel implementations. In the
long term, it removes all of the device pointers, so we should be able
to remove all of the various platform drivers more easily.

BUG=chrome-os-partner:11158,chrome-os-partner:10717
TEST=Tested on snow, hdmi hotplugged, booted with/without hdmi,
suspend/resume tested. No regressions detected.

Change-Id: Ia1bd6bbd67d49af2abecf36c43bfe765ae00f38d
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/36477
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

drm/exynos: Fix up encoder->dpms locking

encoder->dpms was both being read and written without the struct mutex
lock. Fix this so we're synchronized.

BUG=None
TEST=Tested on snow, ensured we only called dpms once

Change-Id: I7d2b34e908273660c31c0290e3c6c632556297fc
Signed-off-by: Sean Paul <seanpaul@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37643
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>
Reviewed-by: Stéphane Marchesin <marcheu@chromium.org>

CHROMIUM: Input: cyapa - Use static const for unique_str

Use static contst for unique_str "CYTRA" used to compare
with product_id.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=Builds clean.

Change-Id: I667cfe0dfef652e0a377079bfacedc41b8412891
Reviewed-on: https://gerrit.chromium.org/gerrit/37740
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>

CHROMIUM: Input: cyapa - Use u8 instead of unsigned char

Change type of cmd and len in cyapa_cmd_len to u8 for consistency
with the rest of the driver.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=Builds clean.

Change-Id: I726546c730e54db88d683e14199df83bc34b2311
Reviewed-on: https://gerrit.chromium.org/gerrit/37739
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>

CHROMIUM: Input: cyapa - Clean up whitespace and comments

Minor cleanup for preparation for upstreaming this driver.

* Add attribution for djkurtz and bleung for cleanup and restructuring
* Remove kernel_doc comments for i2c_reg_read and i2c_reg_write
* Fix comment for cyapa_poll_state for correct ms values and spellcheck.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chromium-os:21047
TEST=Builds clean

Change-Id: I75028fb0a674e3bb30aae4c5af4a8c581a734065
Reviewed-on: https://gerrit.chromium.org/gerrit/37738
Reviewed-by: Daniel Kurtz <djkurtz@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>

UPSTREAM: xhci: fix null-pointer dereference when destroying half-built segment rings

xhci_alloc_segments_for_ring() builds a list of xhci_segments and links
the tail to head at the end (forming a ring). When it bails out for OOM
reasons half-way through, it tries to destroy its half-built list with
xhci_free_segments_for_ring(), even though it is not a ring yet. This
causes a null-pointer dereference upon hitting the last element.

Furthermore, one of its callers (xhci_ring_alloc()) mistakenly believes
the output parameters to be valid upon this kind of OOM failure, and
calls xhci_ring_free() on them. Since the (incomplete) list/ring should
already be destroyed in that case, this would lead to a use after free.

This patch fixes those issues by having xhci_alloc_segments_for_ring()
destroy its half-built, non-circular list manually and destroying the
invalid struct xhci_ring in xhci_ring_alloc() with a plain kfree().

BUG=chrome-os-partner:15124
TEST=Hard to reproduce since the underlying bug was fixed already.

Has been accepted upstream by Sarah Sharp at
https://lkml.org/lkml/2012/11/12/313

Change-Id: I8035a390ab410873b97bce23dded5f0cfc14dcf6
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/36745
Reviewed-by: Vincent Palatin <vpalatin@chromium.org>

CHROMIUM: tcp: Avoid merging segments on the OOO queue into a cloned SKB

TCP tries to merge socket buffers in the out-of-order queue when they
are sequential. This is generally a good thing to save memory and speed
up processing, but can lead to problems when it expands a buffer that
has previously been cloned.

We have encountered a bug with a confused/broken receive queue, that we
think might have been caused by this problem, as the smsc95xx driver
uses cloned buffers in its receive path. Thanks to edumazet for
pointing that out and suggesting this fix.

This patch makes sure that this code path only expands socket buffers
that have not been cloned and just enqueues them normally in the other
case. In upstream kernels starting with 3.5, the affected code has been
refactored into tcp_try_coalesce with commit 1402d36, which has later
been patched to fix this issue in commit 923dd34. The fix has not been
backported into the upstream 3.4 kernel.

BUG=chromium-os:35827
TEST=None

Change-Id: I2cdf09946a3cbb0cbff35b1ac09313352b8f1e94
Signed-off-by: Julius Werner <jwerner@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37666
Reviewed-by: Eric Dumazet <edumazet@chromium.org>
Reviewed-by: Sameer Nanda <snanda@chromium.org>

CHROMIUM: Input: atmel_mxt_ts - Return early from mxt_exit_bl on error

In the case that a firmware update fails, the driver
will try to proceed to mxt_input_dev_create, which will try to
dereference the NULL object table. Bail out early so this does
not happen.

Signed-off-by: Benson Leung <bleung@chromium.org>
BUG=chrome-os-partner:10688
TEST=The following instructions will write the wrong firmware to the
touchpad device.
1. cd /sys/bus/i2c/devices/1-004b
2. echo maxtouch-ts.fw > fw_file
3. echo 1 > update_fw
4. dmesg | tail
This will result in a failed touchpad update. Check that dmesg shows:
[  158.495164] atmel_mxt_ts 1-004b: bootloader version: 32
[  170.563491] atmel_mxt_ts 1-004b: mxt_read_reg: i2c read failed
[  170.563513] atmel_mxt_ts 1-004b: Failed to initialize on exit bl. error = -6
Check that the system does not panic in this situation.

To recover from this state, simply shut the machine down completely using
sudo shutdown -P now

Change-Id: Ib773efd8b76aced9f5faab0b51745db7192e78f9
Reviewed-on: https://gerrit.chromium.org/gerrit/37860
Reviewed-by: Yufeng Shen <miletus@chromium.org>
Commit-Ready: Benson Leung <bleung@chromium.org>
Tested-by: Benson Leung <bleung@chromium.org>

CHROMIUM: mfd: chromeos_ec: update EC<->AP protocol

The ec_commands.h file has been added quite a few new commands. Bring
the new version.

Sync ec_commands.h file of commit
  3c575ccb02ff58eaae3e7c3dd8e2e86ad25ac106

BUG=none
TEST=build kernel for daisy and link

Change-Id: If17bba42df105a92aa31d6ddfedc0cdae682c279
Signed-off-by: Che-Liang Chiou <clchiou@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37677

CHROMIUM: remove implementation details from public header

chromeos_platform_read_nvram() and chromeos_platform_write_nvram() are
implementation details of chromeos_set_need_recovery() that are not used
outside drivers/platform.  So remove them from public header.

BUG=none
TEST=build kernel

Change-Id: Ie3a67dbe05b9764dc680dd1e5cf15dd91160d61a
Signed-off-by: Che-Liang Chiou <clchiou@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37685

CHROMIUM: pm-check: optimize xor and sum for A-15

This speeds up the xor and checksum functions in pm-check

before:
[  202.557153] s3c_pm_check: Suspend memory scan took 1135837 usecs
[  202.557153] s3c_pm_check: Resume memory scan took 1097304 usecs

after:
[   63.292634] s3c_pm_check: Suspend memory scan took 553787 usecs
[   63.292634] s3c_pm_check: Resume memory scan took 531468 usecs

BUG=chrome-os-partner:15655
TEST=run suspend_stress_test, ensure errors are still detected

Change-Id: Icf20432c107a25efde77271f3956e82fc8a3a7a2
Signed-off-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Sonny Rao <sonnyrao@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37781

CHROMIUM: Input: atmel_mxt_ts - make suspend power acquisition interval configurable

Create sysfs for T7 IDLE/ACTV-ACQINT settings to have a configurable power
acquisition setting during system suspend.

Also change the default T7 IDLE/ACTV-ACQINT values during system suspend to be
32ms to have a sensitive wakeup-from-touch response.

Signed-off-by: Yufeng Shen <miletus@chromium.org>
BUG=chrome-os-partner:15968
TEST=1. go to touch device sysfs directory
        echo 254 > power/suspend_acq_interval_ms
        Try powerd_suspend and use very quick, gentle touch to wake up the system
        and notice that sometimes the system does not wake up from touch.
     2. go to touch device sysfs directory
        echo 32 > power/suspend_acq_interval_ms
        Try powerd_suspend and use very quick, gentle touch to wake up the system
        and make sure the system always wakes up.

Change-Id: Ib7ad7a6b81699b71bcb165df06f275e55a68e7c6
Reviewed-on: https://gerrit.chromium.org/gerrit/37836
Reviewed-by: Benson Leung <bleung@chromium.org>
Commit-Ready: Yufeng Shen <miletus@chromium.org>
Tested-by: Yufeng Shen <miletus@chromium.org>

PM / Sleep: samsung: pm-check: Add ability to skip unallocated pages

Memory corruption in unallocated pages isn't harmful and we can save
time by only checking allocated pages.

BUG=None
TEST=Ran with all memory allocated in userspace.
   Saw errors on machine with bad bios.  No errors on good bios.
   Timing tests:
  - Memory fully allocated: 4.428 second suspend, 1.154 resume.
  - Chrome on login screen: 0.995 suspend, 0.319 resume
  - Chrome UI stopped: 0.591 suspend, 0.195 resume
  - pm_check_skip_unused = 0: 4.510 suspend, 1.155 resume

Change-Id: I3b996b17b3f12885f86595684e2d0e14b11adbdc
Signed-off-by: Jonathan Kliegman <kliegs@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37716
Reviewed-by: Doug Anderson <dianders@chromium.org>

samsung: snow: bitfix: Set interleaving based on corruption unit

We'd like to process all pages with the same destination one after
another.  Tell pm-check about this.

BUG=chrome-os-partner:15655
TEST=Timed before and after and saw 500-600ms speedup.
Before:
 [   24.709808] s3c_pm_check: Suspend memory scan took 3947182 usecs
 [   24.709808] s3c_pm_check: Resume memory scan took 1098695 usecs
 [   37.209790] s3c_pm_check: Suspend memory scan took 4020150 usecs
 [   37.209790] s3c_pm_check: Resume memory scan took 1099690 usecs
 [   51.569801] s3c_pm_check: Suspend memory scan took 4017198 usecs
 [   51.569801] s3c_pm_check: Resume memory scan took 1099794 usecs
After:
 [   33.923909] s3c_pm_check: Suspend memory scan took 3493848 usecs
 [   33.923909] s3c_pm_check: Resume memory scan took 1104389 usecs
 [   44.273912] s3c_pm_check: Suspend memory scan took 3444726 usecs
 [   44.273912] s3c_pm_check: Resume memory scan took 1104217 usecs
 [   57.753915] s3c_pm_check: Suspend memory scan took 3429833 usecs
 [   57.753915] s3c_pm_check: Resume memory scan took 1104522 usecs

Change-Id: Id57e5c76788ec1cfb4d09d1a99e6b2e10a56f184
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37725
Reviewed-by: Jon Kliegman <kliegs@chromium.org>

samsung: pm-check: Add interleaving support

When pm-check is used in conjunction with bitfix we can get much
faster processing times if we change the order that we process chunks.
Add support for changing the order that we process pages in save so
that bitfix can use it.

BUG=chrome-os-partner:15655
TEST=Timed before and after and saw 500-600ms speedup.  See associated
bitfix change (dd13e37d3ca428eb89ee6d9afe3b258ec4ca0bfd) for details.

Change-Id: If2b720a75f5274df0e5dc11d77662d73a4581ffb
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37690
Reviewed-by: Jon Kliegman <kliegs@chromium.org>

samsung: snow: bitfix: Use linux xor code

Linux already has some xor code that was optimized for RAID.  Include
it and use it for bitfix.  This appears to save 200-300 ms.

BUG=chrome-os-partner:15655
TEST=Timed before and after:
Before:
 [   28.344096] s3c_pm_check: Suspend memory scan took 4341511 usecs
 [   28.344096] s3c_pm_check: Resume memory scan took 1204971 usecs
 [   40.794096] s3c_pm_check: Suspend memory scan took 4372845 usecs
 [   40.794096] s3c_pm_check: Resume memory scan took 1205185 usecs
 [   53.689130] s3c_pm_check: Suspend memory scan took 4348640 usecs
 [   53.689130] s3c_pm_check: Resume memory scan took 1205305 usecs
After:
 [   34.604046] s3c_pm_check: Suspend memory scan took 4000453 usecs
 [   34.604046] s3c_pm_check: Resume memory scan took 1202673 usecs
 [   48.804021] s3c_pm_check: Suspend memory scan took 3991251 usecs
 [   48.804021] s3c_pm_check: Resume memory scan took 1202789 usecs
 [   62.234132] s3c_pm_check: Suspend memory scan took 4161455 usecs
 [   62.234132] s3c_pm_check: Resume memory scan took 1203661 usecs
 [   83.439127] s3c_pm_check: Suspend memory scan took 4002259 usecs
 [   83.439127] s3c_pm_check: Resume memory scan took 1205180 usecs

Change-Id: I17a6d6b8c072ac0fe04ae2fa8762374aafcae7ff
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37691
Reviewed-by: Jon Kliegman <kliegs@chromium.org>
Reviewed-by: Sonny Rao <sonnyrao@chromium.org>

samsung: snow: bitfix: Panic if too many bits fixed in 1 chunk

If we see too much corruption in a single chunk then that's a sign
that we're in a state that we can't recover from (either a bitfix
internal error or a corruption of some memory that can't be checked).
Panic in that case.

BUG=chrome-os-partner:15655
TEST=suspend_stress_test
TEST=In code change MAX_BYTES_TO_FIX to 3 and then re-run test.
See a reboot and see the panic in /var/spool/crash:
  <7>[  163.818802] s3c_pm_check: Restore CRC error at 71734000 (6da6143e vs 1d894763)
  <6>[  163.818802] bitfix_recover_chunk: Attempting recovery at 71734000
  <6>[  163.818802] ...fixed 0x71735644 from 0xbadc0de1 to 0x00000000 (4 bytes)
  <5>[  163.818802] ------------[ cut here ]------------
  <2>[  163.818802] kernel BUG at .../arch/arm/mach-exynos/bitfix-snow.c:504!
  <0>[  163.818802] Internal error: Oops - BUG: 0 [#1] SMP ARM
  <5>[  163.818802] Modules linked in: i2c_dev uinput isl29018(C) industrialio(C) sbs_battery rfcomm btmrvl_sdio btmrvl bluetooth fuse nf_conntrack_ipv6 nf_defrag_ipv6 ip6table_filter xt_mark ip6_tables mwifiex_sdio mwifiex cfg80211 spidev uvcvideo videobuf2_vmalloc asix smsc95xx usbnet joydev
  <5>[  163.818802] CPU: 0    Tainted: G         C    (3.4.0 #4)
  <5>[  163.818802] PC is at bitfix_recover_chunk+0x220/0x244
  <5>[  163.818802] LR is at bitfix_recover_chunk+0x20c/0x244
  ...
  <5>[  163.818802] [<80024f5c>] (bitfix_recover_chunk+0x220/0x244) from [<8002786c>] (s3c_pm_runcheck+0x158/0x2a0)
  ...

Change-Id: Iaad53d821231f419cb1e44a9e1ef3c8ec36973d2
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37600
Reviewed-by: Jon Kliegman <kliegs@chromium.org>

samsung: pm-check: Speed up memory exclusion check

Changed the memory exclusion change to assume that we're working
on one page at a time.  This is mostly useful in bitfix which
can do a simple calculation instead of a loop.

BUG=chrome-os-partner:15655
TEST=Timed before and after:
Before:
 [   65.769646] s3c_pm_check: Suspend memory scan took 4304997 usecs
 [   65.769646] s3c_pm_check: Resume memory scan took 1176677 usecs
 [   80.684717] s3c_pm_check: Suspend memory scan took 4319001 usecs
 [   80.684717] s3c_pm_check: Resume memory scan took 1172676 usecs
 [  101.599638] s3c_pm_check: Suspend memory scan took 4331624 usecs
 [  101.599638] s3c_pm_check: Resume memory scan took 1173586 usecs

With samsung: pm-check: Change exclusions to be per page:
 [   83.759770] s3c_pm_check: Suspend memory scan took 4695243 usecs
 [   83.759770] s3c_pm_check: Resume memory scan took 1271669 usecs
 [   99.309735] s3c_pm_check: Suspend memory scan took 4695435 usecs
 [   99.309735] s3c_pm_check: Resume memory scan took 1270742 usecs
 [  115.154743] s3c_pm_check: Suspend memory scan took 4600779 usecs
 [  115.154743] s3c_pm_check: Resume memory scan took 1270537 usecs

With samsung: pm-check: Speed up memory exclusion check:
 [   28.344096] s3c_pm_check: Suspend memory scan took 4341511 usecs
 [   28.344096] s3c_pm_check: Resume memory scan took 1204971 usecs
 [   40.794096] s3c_pm_check: Suspend memory scan took 4372845 usecs
 [   40.794096] s3c_pm_check: Resume memory scan took 1205185 usecs
 [   53.689130] s3c_pm_check: Suspend memory scan took 4348640 usecs
 [   53.689130] s3c_pm_check: Resume memory scan took 1205305 usecs
TEST=suspend_stress_test

Change-Id: I71b6f84e12af6d2aa28373262650e8fa25abb1ea
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37679
Reviewed-by: Jon Kliegman <kliegs@chromium.org>

samsung: snow: bitfix: Fix missing kunmap_atomic() when bitfixing

A recent change ("samsung: pm-check: Change exclusions to be per
page") introduced a bug where pages wouldn't get unmapped after
bitfix.  Fix the bug.

BUG=chrome-os-partner:15655
TEST=suspend_stress_test

Change-Id: I25d86c5b583309e8ff1115daca436cc625fee9c7
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37754
Reviewed-by: Michael Spang <spang@chromium.org>

arm: xor: Add prefetch options

On my current system these options aren't faster than the assembly but
it doesn't hurt to try them.  This also avoids a warning about unused
statics.

BUG=None
TEST=No more warning.  'dmesg | grep xor' at bootup still shows
we're using the arm4regs function.

Change-Id: I4d566a79efb7bc7dbbbd7d766681b089ffe07326
Signed-off-by: Doug Anderson <dianders@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37687
Reviewed-by: Sonny Rao <sonnyrao@chromium.org>

CHROMIUM: MALI: added stack traces for gpu faults

Add stack traces for gpu faults.  This is temporary for helping with
problems in the field.

Signed-off-by: Gabriele Paoloni <gabriele.paoloni@arm.com>
BUG=chrome-os-partner:1556
TEST=build and boot

Change-Id: Id79c6e4d339ae362df9497beb936ebdfd45931ea
Reviewed-on: https://gerrit.chromium.org/gerrit/37349
Commit-Ready: Sam Leffler <sleffler@chromium.org>
Tested-by: Sam Leffler <sleffler@chromium.org>
Reviewed-by: Anush Elangovan <anush@chromium.org>

PM / Sleep: exynos: mct: Mark hi2 local static suspend_volatile_bss

If ktime_get is called, the local static hi2 gets updated when clock->read()
is called.  As this is called when testing timings in pm-check, make sure to mark it
is marked as suspend_volatile_bss.

BUG=None
TEST=With timings on and bitfix enabled, saw hi2's address being fixed.
  After patch saw it no longer being affected.
Signed-off-by: Jonathan Kliegman <kliegs@chromium.org>
Change-Id: I060dbd6a46085def936a14b6dfbb051570c40a85
Reviewed-on: https://gerrit.chromium.org/gerrit/37711
Tested-by: Jon Kliegman <kliegs@chromium.org>
Reviewed-by: Doug Anderson <dianders@chromium.org>
Commit-Ready: Jon Kliegman <kliegs@chromium.org>

PM / Sleep: suspend_volatile: Remove spinlock in overlap check

pm_does_overlap_suspend_volatile is only called when scheduling disabled
and can not be interrupted.  It is also called frequently.  Remove the
spinlock acquistion to speedup suspend/resume times.

BUG=None
TEST=Compared timings before and after.
  Ensured no warnings in log.
  Did cycle of 200 suspend/resumes on fixed bios successfully.
  Still caught errors on bad bios correctly.

Change-Id: Ifce238e223e25121b48c880407f712305cdde825
Signed-off-by: Jonathan Kliegman <kliegs@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37708
Reviewed-by: Doug Anderson <dianders@chromium.org>

samsung: pm-check: Change exclusions to be per page

Increase granularity to page-level exclusions.  This helps
set the way for doing smarter exclusions based on if a page
is allocated or not.

BUG=None
TEST=Run on machines with and without bad memory
  Confirm 500 runs on good machine with no false positives
  Confirm bitfix and pm_check properly detect and correct errors

Change-Id: Ibeab83dbc822637dc0b4857bb9883acc5b6ab7c5
Signed-off-by: Jonathan Kliegman <kliegs@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/37646
Reviewed-by: Doug Anderson <dianders@chromium.org>

CHROMIUM: ath9k: dump_eep_power in debugfs only with eep_ops support

The ath9k driver is adding a dump_eep_power debugfs file
even when there is no corresponding eep_ops support for
dump_eep_power.

Accessing dump_eep_power on platforms without support for
dump_eep_power leads to kernel crashes, so do not add the
debugfs entry for those platforms.

Signed-off-by: Jason Abele <jabele@chromium.org>
BUG=None
TEST=Do the following:
ls /sys/kernel/debug/ieee80211/phy0/ath9k/
make sure dump_eep_power does not exist on mario
Change-Id: I10cc18c4322481efc49cea7d608852b60cdcf904
Reviewed-on: https://gerrit.chromium.org/gerrit/37726
Reviewed-by: Sam Leffler <sleffler@chromium.org>
Commit-Ready: Jason Abele <jabele@chromium.org>
Tested-by: Jason Abele <jabele@chromium.org>

CHROMIUM: md: bootcache if init fails, mark invalid

On any failure that doesn't fully initialize the bootcache,
it should set mark the cache as invalid so the bootcache
utility can rebuild the bootcache.

BUG=chromium-os:35378
TEST=ran bootcachetest on stumpy, arm

Change-Id: I3331ba0bc505445310c52a7c73a334106fd7141b
Signed-off-by: Paul Taysom <taysom@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/35837
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>

CHROMIUM: md: bootcache don't delete sys files

Because the device mapper already deletes any sys files
created by dm-bootcache, it shouldn't delete them
when cleaning up.

BUG=chromium-os:35377
TEST=ran bootcachetest on stumpy, arm

Change-Id: I73d40a6de17b15827da2dd421a29294e286ebc1a
Signed-off-by: Paul Taysom <taysom@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/35836
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>

CHROMIUM: md: bootcache check metadata and data lengths

bootcache now verifies that the lengths for metadata and data
make sense and fit with in the partition.

BUG=chromium-os:35379
TEST=ran bootcachetest on stumpy, arm

Change-Id: I6beee3a5b1aa723aaf08c673a004bc4924afc42d
Signed-off-by: Paul Taysom <taysom@chromium.org>
Reviewed-on: https://gerrit.chromium.org/gerrit/35578
Reviewed-by: Mandeep Singh Baines <msb@chromium.org>