[ARM] fiq glue: Align fiq stacks

Change-Id: I956f05d0b0ce48572e611765e56a439dc036c052
Signed-off-by: Colin Cross <ccross@android.com>

mmc: Add "ignore mmc pm notify" functionality

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

[ARM] fiq debugger: Allow selection of target cpu

Change-Id: I676bac08ba12dfa506aea16800fc80432b4bc83d
Signed-off-by: Colin Cross <ccross@android.com>

epoll: make epoll_wait use the hrtimer range feature

This make epoll use hrtimers for the timeout value which prevents
epoll_wait() from timing out up to a millisecond early.

This mirrors the behavior of select() and poll().

Change-Id: Ic6489ab6194935df348d31b8e92239227ffd3f17
Signed-off-by: Shawn Bohrer <shawn.bohrer@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

select: rename estimate_accuracy to select_estimate_accuracy

Make it a subsystem-specific identifier because we wish to amke it
non-static in the next patch ("epoll: make epoll_wait() use the hrtimer
range feature").

Change-Id: Ie333f2051a3dc73dc8b0615b4fdccef63591c918
Cc: Shawn Bohrer <shawn.bohrer@gmail.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Davide Libenzi <davidel@xmailserver.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

mmc: sdio: Fix enable_hs and enable_wide in sdio_reset_comm()

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

ARM: fiq debugger: Add tty to fiq debugger

Change-Id: I80347cdb70cda104b96562c63f972c1f217e3822
Signed-off-by: Colin Cross <ccross@google.com>

ARM: Add generic fiq serial debugger

Change-Id: Ibb536c88f0dbaf4766d0599296907e35e42cbfd6
Signed-off-by: Iliyan Malchev <malchev@google.com>
Signed-off-by: Arve Hjønnevåg <arve@android.com>

ARM: Add fiq_glue

Change-Id: I27d2554e07d9de204e0a06696d38db51608d9f6b
Signed-off-by: Arve Hjønnevåg <arve@android.com>
Signed-off-by: Colin Cross <ccross@android.com>

sched: use the old min_vruntime when normalizing on dequeue

After pulling the thread off the run-queue during a cgroup change,
the cfs_rq.min_vruntime gets recalculated. The dequeued thread's vruntime
then gets normalized to this new value. This can then lead to the thread
getting an unfair boost in the new group if the vruntime of the next
task in the old run-queue was way further ahead.

Cc: Arve Hjønnevåg <arve@android.com>
Signed-off-by: Dima Zavin <dima@android.com>

sched: normalize sleeper's vruntime during group change

If you switch the cgroup of a sleeping thread, its vruntime does
not get adjusted correctly for the difference between the
min_vruntime values of the two groups.

This patch adds a new callback, prep_move_task, to struct sched_class
to give sched_fair the opportunity to adjust the task's vruntime
just before setting its new group. This allows us to properly normalize
a sleeping task's vruntime when moving it between different cgroups.

More details about the problem:
  http://lkml.org/lkml/2010/9/28/24

Cc: Arve Hjønnevåg <arve@android.com>
Signed-off-by: Dima Zavin <dima@android.com>

net: wireless: bcm4329: Fix setting HT clock race conditions in driver start

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

Allow CONFIG_STACKTRACE to be enabled by itself.

This allows us to get a kernel stacktrace for a thread though /proc.
Also enable it by default.

Change-Id: If8c21cd02feaf9863f4841ace524fa30c7328d49
Signed-off-by: Arve Hjønnevåg <arve@android.com>

net: wireless: bcm4329: Fix setting HT clock race conditions

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

usb: gadget: android: Remove WAKEUP flag from bmAttributes

Remove USB_CONFIG_ATT_WAKEUP from bmAttributes as we don't
support remote wakeup. This fixes an issue with the USB 2.0
compliance test tool.

Change-Id: Ic15e44710069f3f05ace6a82bbfa2b85a3649027
Signed-off-by: Benoit Goby <benoit@android.com>

Support lseek(2) in ashmem driver

Signed-off-by: Bjorn Bringert <bringert@android.com>
Change-Id: I509d18b21832e229737ea7ebaa231fb107eb61d7

l2tp: l2tp_eth: Fix section mismatch warning

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix interrupt enabling in case of error

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Update to Version 4.218.248.6

Fix watchdog time rescheduling, fix first scan no-return

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Enable packet filtering during low power mode

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Ignore error if scan results are empty

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Add debug print for cscan failure

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix roaming setting on resume

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

mmc: subtract boot sectors from disk size for eMMC 4.3+ devices

the csd sector count reported by eMMC 4.3+ cards includes the boot
partition size; subtract this from the size reported to the disk
since the boot partition is inaccessible

Change-Id: I601b83aa0159b7aa446409ea8c945b256dd0b5b1
Signed-off-by: Gary King <gking@nvidia.com>

mmc_block: Allow more than 8 partitions per card

Set the GENHD_FL_EXT_DEVT flag, which will allocate minor numbers
in major 259 for partitions past disk->minors.

Also remove the use of disk_devt to determine devidx from md->disk.
md->disk->first_minor is always initialized from devidx and can
always be used to recover it.

Signed-off-by: Colin Cross <ccross@android.com>

EFI/GPT: Register named partitions.

Change-Id: If318512de67d3d152192aa66b074427f4e5b5c2e
Signed-off-by: Todd Poynor <toddpoynor@google.com>

fs: partitions: Add support for named partitions

Adds a new file in /sys/block/<block>/<partition> called partition_name
that contains the name of the partition, if specified by the partition
handler.

Change-Id: I6648ed95eabefd1d00edbfdfd99eeb971d15f4b3
Signed-off-by: Colin Cross <ccross@android.com>

net: wireless: bcm4329: Fix roaming failure case

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

PM: Prevent waiting forever on asynchronous resume after abort

Only wait on a parent device during resume if the parent device is
suspended.

Consider three drivers, A, B, and C.  The parent of A is C, and C
has async_suspend set.  On boot, C->power.completion is initialized
to 0.

During the first suspend:
suspend_devices_and_enter(...)
 dpm_resume(...)
  device_suspend(A)
  device_suspend(B) returns error, aborts suspend
 dpm_resume_end(...)
   dpm_resume(...)
    device_resume(A)
     dpm_wait(A->parent == C)
      wait_for_completion(C->power.completion)

The wait_for_completion will never complete, because
complete_all(C->power.completion) will only be called from
device_suspend(C) or device_resume(C), neither of which is called
if suspend is aborted before C.

After a successful suspend->resume cycle, where B doesn't abort
suspend, C->power.completion is left in the completed state by the
call to device_resume(C), and the same call path will work if B
aborts suspend.

Signed-off-by: Colin Cross <ccross@android.com>

power: wakelock: call __get_wall_to_monotonic() instead of using wall_to_monotonic

Change-Id: I9e9c3b923bf9a22ffd48f80a72050289496e57d8

net: Fix CONFIG_RPS option to be turned off

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix HW_OOB interrupt processing

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

Revert "net: wireless: bcm4329: Fix HW_OOB interrupt processing"

This reverts commit 8bd035daa820dc5612ae311262c71dc133871046.

Revert "net: wireless: bcm4329: Fix interrupt enabling for level interrupt"

This reverts commit 261d21bbfffef6261696d0d13672d2e4f9d76f05.

net: wireless: bcm4329: Fix interrupt enabling for level interrupt

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix HW_OOB interrupt processing

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

USB: gadget: android: Disable MTP when RNDIS function is enabled.

Signed-off-by: Mike Lockwood <lockwood@android.com>

net: wireless: bcm4329: Update to Version 4.218.248

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

wakelock: Fix operator precedence bug

Change-Id: I21366ace371d1b8f4684ddbe4ea8d555a926ac21
Signed-off-by: Colin Cross <ccross@google.com>

ashmem: Update arguments of shrinker for 2.6.35

Change-Id: Ie527d18f3352ede06d565826c8d35ded1638203a
Signed-off-by: Colin Cross <ccross@google.com>

lowmemorykiller: Update arguments of shrinker for 2.6.35

Change-Id: I6bb09b36639527f91c48704118acad5e50c4163f
Signed-off-by: Colin Cross <ccross@google.com>

net: wireless: bcm4329: Send "HANG" message only once

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

Allow multiple pmem master mmap()s.

Signed-off-by: Jamie Gennis <jgennis@google.com>
Change-Id: Icbe619c92e0ebb391f0a93f81937705452a67d87

net: wireless: bcm4329: Disable packet filtering

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Turn OFF packet filtering during DHCP session

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Increase PMU_MAX_TRANSITION_DLY to 1 sec

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix "setsuspend" behavior

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Add memory barriers to wait functions

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix unregister_early_suspend() in dhd_detach()

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix dhd_bus_watchdog() race conditions

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix scan timeout for abg case

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Force scan when driver is loaded

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Add wakelock processing to WEXT requests

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Update to Version 4.218.246 and setband fix

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Update to 4.218.245 (combo scan)

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

USB: g_mass_storage: Always allow disabling mass storage by writing to lun file

For android builds we disable the check for curlun->prevent_medium_removal.
Instead we let the framework manage unmounting policy, as we sometimes need
to unmount after the media has been removed.
This also helps support hosts that do not inform the device when the media
has been unmounted.

Signed-off-by: Mike Lockwood <lockwood@android.com>

yaffs: Use init_timer_on_stack for timer on stack

Change-Id: I517b94a1ca22c1690fd6421cf9892a05039a3def
Signed-off-by: Arve Hjønnevåg <arve@android.com>

input: gpio_event: make driver be more robust against incorrectly configured lvl trigger irqs

Change-Id: Ie378600668500dcffeaaddeaba3628e5c2141aa4
Signed-off-by: Dima Zavin <dima@android.com>

Input: gpio_event: Don't call gpio_cansleep before gpio_request.

If the gpio did not exist it would crash in gpiolib.

Signed-off-by: Arve Hjønnevåg <arve@android.com>

net: wireless: bcm4329: Add wakelock processing in ioctl and messaging

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Fix packet filter to only filter out non-unicast frames

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

ashmem: Fix ASHMEM_SET_PROT_MASK.

Change-Id: I1412cc9560de8c4feb1162fc30922f0e3362a476
Signed-off-by: Arve Hjønnevåg <arve@android.com>

scheduler: cpuacct: Enable platform callbacks for cpuacct power tracking

Platform must register cpu power function that return power in
milliWatt seconds.

Change-Id: I1caa0335e316c352eee3b1ddf326fcd4942bcbe8
Signed-off-by: Mike Chan <mike@android.com>

scheduler: cpuacct: Enable platform hooks to track cpuusage for CPU frequencies

Introduce new platform callback hooks for cpuacct for tracking CPU frequencies

Not all platforms / architectures have a set CPU_FREQ_TABLE defined
for CPU transition speeds. In order to track time spent in at various
CPU frequencies, we enable platform callbacks from cpuacct for this accounting.

Architectures that support overclock boosting, or don't have pre-defined
frequency tables can implement their own bucketing system that makes sense
given their cpufreq scaling abilities.

New file:
cpuacct.cpufreq reports the CPU time (in nanoseconds) spent at each CPU
frequency.

Change-Id: I10a80b3162e6fff3a8a2f74dd6bb37e88b12ba96
Signed-off-by: Mike Chan <mike@android.com>

USB: gadget: f_mtp: Add ioctl for sending events via the interrupt endpoint

Signed-off-by: Mike Lockwood <lockwood@android.com>

misc: add akm8975 compass driver

Originally written by HTC. Contributions by Motorola and AKM.
  misc: Import akm8975 from Motorola

Major style and code cleanups by Praveen Bharathi <pbharathi@motorola.com>
  misc: akm8975: clean up code violations in akm8975.c
  misc: akm8975: Clean up coding style, add suspend and resume

Change-Id: I4196913f15aec2dfbed47506d3dc085aada8e92d
Signed-off-by: Dima Zavin <dima@android.com>

drivers: usb: gadget: rndis: Fix section mismatch warning

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

drivers: usb: gadget: mass_storage: Fix section mismatch warning

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

drivers: input: misc: Fix section mismatch warning

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

drivers: usb: gadget: Fix section mismatch warning

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Add driver SETSUSPEND command

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Fix BUS DOWN in IOCTL

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

USB: composite: Add usb_composite_force_reset utility to force enumeration

Use this rather than calling usb_gadget_disconnect and usb_gadget_connect
directly to avoid sending USB disconnect events to userspace when resetting
the bus.

Signed-off-by: Mike Lockwood <lockwood@android.com>

network: wireless: bcm4329: Set special OUI: 02:1A:11:FH:HH:HH for SoftAP

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

USB: gadget: android: Remove unused function android_usb_set_connected()

Signed-off-by: Mike Lockwood <lockwood@android.com>

USB: gadget: composite: Move switch_set_state calls to a work queue

Signed-off-by: Mike Lockwood <lockwood@android.com>

USB: gadget: f_mtp: USB gadget function driver for MTP and PTP device support.

Signed-off-by: Mike Lockwood <lockwood@android.com>

USB: composite: Add support for functions overriding USB_DT_STRING descriptors

Needed for MTP support.

Signed-off-by: Mike Lockwood <lockwood@android.com>

USB: gadget: composite: Add userspace notifications for USB state changes

Add switch to notify current USB configuration.  This can be used to detect
USB connect and disconnect events.

Broadcast a change via the usb_composite class when a USB function is
enabled or disabled.

Rename usb_function.hidden to usb_function.disabled.

Signed-off-by: Mike Lockwood <lockwood@android.com>

USB: gadget: f_mass_storage: Integrate android customizations

Integrate support for android composite driver and platform data
within CONFIG_USB_ANDROID_MASS_STORAGE.

Signed-off-by: Mike Lockwood <lockwood@android.com>

Implement read(2) in ashmem driver

Bug: 2595601
Change-Id: I47c0016f594f9354fb8658ccb26e3d395bcb137b
Signed-off-by: Bjorn Bringert <bringert@android.com>

network: wireless: bcm4329: Reduce driver loading time

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Add 2.6.35 compatibility and fix memory leak in set_multicast_list

Signed-off-by: Dmitry Shmidt <dimitrysh@android.com>
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: wireless: bcm4329: Add SoftAP MAC address randomization

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

net: activity_stats: Add statistics for network transmission activity

When enabled, tracks the frequency of network transmissions
(inbound and outbound) and buckets them accordingly.
Buckets are determined by time between network activity.

Each bucket represents the number of network transmisions that were
N sec or longer apart. Where N is defined as 1 << bucket index.

This network pattern tracking is particularly useful for wireless
networks (ie: 3G) where batching network activity closely together
is more power efficient than far apart.

New file: /proc/net/stat/activity

output:

Min Bucket(sec) Count
              1 7
              2 0
              4 1
              8 0
             16 0
             32 2
             64 1
            128 0

Change-Id: I4c4cd8627b872a55f326b1715c51bc3bdd6e8d92
Signed-off-by: Mike Chan <mike@android.com>

network: wireless: bcm4329: Update to Version 4.218.239
    Add reading mac address from platform data
    Add dhd_os_proto_block protection for dhd_preinit_ioctls
    Revert dhdsdio_clk() changes

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Fix wlan card removal

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Fix MAC address import

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Update to Version 4.218.238

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: Add get_mac_addr functionality to platform

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Remove unnecessary set_freezable() calls

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Add "HANG" event and console monitoring

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

network: wireless: bcm4329: Add bcm4329 driver

Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>

staging: android: lowmemorykiller: Remove bitrotted codepath

Now that we're murder-synchronous, this code path will never be
called (and if it does, it doesn't tell us anything useful other
than we killed a task that was already being killed by somebody
else but hadn't gotten its' signal yet)

Signed-off-by: San Mehat <san@google.com>

staging: android: lowmemorykiller: Fix task_struct leak

As it turns out, the CONFIG_PROFILING interfaces leak a
task struct if the notifier chain returns NOTIFY_OK.. doh.

This patch reworks lowmemkiller to use the new generic task
free notifier chain.

Signed-off-by: San Mehat <san@google.com>

sched: Add a generic notifier when a task struct is about to be freed

This patch adds a notifier which can be used by subsystems that may
be interested in when a task has completely died and is about to
have it's last resource freed.

  The Android lowmemory killer uses this to determine when a task
it has killed has finally given up its goods.

Signed-off-by: San Mehat <san@google.com>

binder: Fix memory corruption via page aliasing

binder_deferred_release was not unmapping the page from the buffer
before freeing it, causing memory corruption.  This only happened
when page(s) had not been freed by binder_update_page_range, which
properly unmaps the pages.

This only happens on architectures with VIPT aliasing.

To reproduce, create a program which opens, mmaps, munmaps, then closes
the binder very quickly.  This should leave a page allocated when the
binder is released.  When binder_deferrred_release is called on the
close, the page will remain mapped to the address in the linear
proc->buffer.  Later, we may map the same physical page to a different
virtual address that has different coloring, and this may cause
aliasing to occur.

PAGE_POISONING will greatly increase your chances of noticing any
problems.

Change-Id: I6941bf212881b8bf846bdfda43d3609c7ae4892e

Signed-off-by: Christopher Lais <chris+android@zenthought.org>

staging: android: lowmemkiller: Substantially reduce overhead during reclaim

This patch optimizes lowmemkiller to not do any work when it has an outstanding
kill-request. This greatly reduces the pressure on the task_list lock
(improving interactivity), as well as improving the vmscan performance
when under heavy memory pressure (by up to 20x in tests).

Note: For this enhancement to work, you need CONFIG_PROFILING

Signed-off-by: San Mehat <san@google.com>

[ARM] Do not call flush_cache_user_range with mmap_sem held

We can't be holding the mmap_sem while calling flush_cache_user_range
because the flush can fault. If we fault on a user address, the
page fault handler will try to take mmap_sem again. Since both places
acquire the read lock, most of the time it succeeds. However, if another
thread tries to acquire the write lock on the mmap_sem (e.g. mmap) in
between the call to flush_cache_user_range and the fault, the down_read
in do_page_fault will deadlock.

Also, since we really can't be holding the mmap_sem while calling
flush_cache_user_range AND vma is actually unused by the flush itself,
get rid of vma as an argument.

Change-Id: If55409bde41ad1060fa4fe7cbd4ac530d4d9a106
Signed-off-by: Dima Zavin <dima@android.com>

USB: android gadget: mass storage: Fix format issue for Vista Host

As part of formating the mass storage device, Host sends the INQUIRY
scsi command. As per the standard, the command length for this
command should be 6 bytes, whereas the Vista host sends 12 bytes.

When the command length of the command is not equal to the standard
length, the device sends a phase error as part of the status phase.
When the host receives a phase error, it re-enumerates, hence the
error.

If the command is INQUIRY, and the command length is 12 bytes,
treating this as a good command and not sending the phase error
to the host fixes this issue.

Signed-off-by: Velempati Chiranjeevi <c_cvelem@qualcomm.com>
Signed-off-by: Mike Lockwood <lockwood@android.com>