On Linux, when an interface goes down all its IPv6
addresses are deleted, so relying on knowing the previous
IPv6 addresses on the interface is brittle. Instead,
support nuking all sockets that are bound to IP addresses
that are not configured and up on the system. This
behaviour is triggered by specifying the unspecified
address (:: or 0.0.0.0). If an IP address is specified, the
behaviour is unchanged, except the ioctl now supports IPv6
as well as IPv4.
Signed-off-by: Lorenzo Colitti <lorenzo@google.com>
In mtp_read(), if wait_event_interruptible() returns an error, we need to
remove the request from the EP queue. Else, on the next call to mtp_read(), we
will attempt to enqueue the request again, potentially corrupting the
queue. This is what happens with musb_gadget_queue(), which does not check for
duplicate requests.
Based on a similar change to f_adb.c
Signed-off-by: Mike Lockwood <lockwood@android.com>
Enable dynamic high level clock gating for Cortex-A9 CPUs, as
described in 2.3.3 "Dynamic high level clock gating" of the
Cortex-A9 TRM. This may cut the clock of the integer core,
system control block, and Data Engine in certain conditions.
Add ARM errata 720791 to avoid corrupting the Jazelle
instruction stream on earlier Cortex-A9 revisions.
Change-Id: I48e51d907e593f26982ea91b0a811553f68e3c86
Signed-off-by: Todd Poynor <toddpoynor@google.com>
- Add support to PNO with adaptive scan time
- Remove logic to generate Link Down based on Roaming Event
- Improve sco/esco detection method for BT COEX
- Add new function TXPOWER to reduce power if needed
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>
This reverts commit f0b0e4bec1.
The reverted commit incorrectly calculates the size of eMMC
devices in some (all?) cases.
This revert may cause problems in cases where the bootloader was
bug-compatible and puts a GPT partition at the incorrect end of
the eMMC device.
Change-Id: Ib006acf9e517b3b8f7570220c28e19c91e7b5f25
Signed-off-by: Colin Cross <ccross@android.com>
usbhid devices have a 10s timeout waiting for the out queue to clear.
Increased the watchdog to 12s.
Change-Id: I96368fca6dff98e4eba8aedb09c23be964c8f4b4
Signed-off-by: Benoit Goby <benoit@android.com>
We used to wait for the previous interrupt packet to complete before sending
the next packet. But unfortunately the previous packet will not complete
until USB is disconnected if the host is not listening on the interrupt
endpoint (which is the case with libmtp on Linux and Mac).
To avoid hanging indefinitely in this case, we now simply return -EBUSY
if the previous interrupt packet has not completed yet.
Signed-off-by: Mike Lockwood <lockwood@android.com>
Based on the list of enabled USB functions, we can now switch the vendor ID
as well as the product ID.
Signed-off-by: Mike Lockwood <lockwood@android.com>
Exit from send_file_work immediately when a cancel request is received.
Only busy status if there is a cancel pending that has not been repoorted to userspace.
This avoids a race condition that can occur when mtp_read resets the state
to STATE_BUSY before we report status OK back to the host.
Signed-off-by: Mike Lockwood <lockwood@android.com>
If the PC on the stack is updated in entry-armv.S,
do_undefinstr can get called after the fixup. do_undefinstr
does its own fixup, and doing both causes the PC to point to
half way through an instruction.
Instead, do the fixup in do_vfp, where only the vfp code
can get called.
Change-Id: I6d966887adc8ed58d88bfe0cb3c0ba29213be488
Signed-off-by: Colin Cross <ccross@android.com>
Ensures that cpufreq_stats_free_table is called before
__cpufreq_remove_dev on cpu hotplug (which also occurs during
suspend on SMP systems) to make sure that sysfs_remove_group
can get called before the cpufreq kobj is freed. Otherwise,
the sysfs file structures are leaked.
Change-Id: I87e55277272f5cfad47e9e7c92630e990bb90069
Signed-off-by: Colin Cross <ccross@android.com>
Exclude from the watchdog the time spent waiting for children that
are resumed asynchronously and time every devices, whether or not they
resumed synchronously.
Change-Id: I84209dfd5df72842e045096c906fd61e20e6d183
Signed-off-by: Benoit Goby <benoit@android.com>
When a driver takes more than 3 seconds to suspend, dump the suspend
thread stack since BUG() might only dump the idle thread stack.
Change-Id: If854db355fdcf3b773ea20b1b5e031def6d4b114
Signed-off-by: Benoit Goby <benoit@android.com>
The exception handler in entry-armv.S checks for thumb mode and
correctly determines the exception location and instruction,
but VFP_bounce uses the uncorrected location off the stack.
If the VFP exception occured in Thumb mode, fix up the
exception location on the stack to match the value that would
be returned in ARM mode.
Fixes segfaults in userspace applications running in Thumb mode
caused by a handled VFP exception returning to the middle of the
instruction that triggered the exception.
Change-Id: I02be4c5b546561d9a967a02c1e9eb1de876cff07
Original-author: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Colin Cross <ccross@android.com>
If the host cancels a file transfer while we have a read request pending,
call usb_ep_dequeue to cancel the read.
Also return -ECANCELED from mtp_ioctl if we are canceled in MTP_RECEIVE_FILE
Signed-off-by: Mike Lockwood <lockwood@android.com>
A thread/process in cgroup_attach_task() could have called
list_del(&tsk->cg_list) after cgroup_exit() had already called
list_del() on the same list. Since it only checked for
!list_empty(&tsk->cg_list) before doing this, the list_del()
call would thus be made twice.
The solution is to leave tsk->cg_list in a valid state in
cgroup_exit() with list_del_init(&tsk->cg_list), which leaves
an empty list.
Change-Id: I4e7c1d0665fced629f5ca033c18dd98afe080e0c
Signed-off-by: Simon Wilson <simonwilson@google.com>
If an alarm was restarted with a value that moved it away from the head
of a queue, the hrtimer would not be updated. This would cause unnecessary
wakeups.
Change-Id: If379f8dd92b0bdb3173bd8d057adfe0dc1d15259
Signed-off-by: Arve Hjønnevåg <arve@android.com>
init_mm used at kernel/sched.c:idle_task_exit() has spin_lock
(init_mm.context.id_lock) that is not initialized when spin_lock/unlock
is called at an ARM machine. Note that mm_struct.context.id_lock is
usually initialized except for the instance of init_mm at
linux/arch/arm/mm/context.c
Not initializing this spinlock incurs "BUG: pinlock bad magic"
warning when spinlock debug is enabled. We have observed such
instances when testing PM in S5PC210 machines.
Change-Id: I52c6c6d035c220822a5675faaf8d163a3535985d
Signed-off-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Signed-off-by: Kyungmin Park <kyungmin.park@samsung.com>
Signed-off-by: Russell King <rmk+kernel@arm.linux.org.uk>
Signed-off-by: Colin Cross <ccross@android.com>
Don't reference the saved copy of the CPU's cpufreq policy
pointer after the governor has been stopped for the CPU.
When the governor is stopped for a CPU:
* Use del_timer_sync() to wait for a currently-running timer
function to stop.
* Delete the timer when the governor is stopped for the
associated CPU, not when the last CPU is stopped.
* Flush any speed down work ongoing.
* Reset the timestamp that is used to tell if the timer function
has had a chance to run since last idle exit.
Check the governor enabled flag for the CPU before re-arming the
timer from within the timer function and at idle exit (in case
stopping the governor at runtime).
Check the governor enabled flag for the CPU in the worker
function and thread before using the policy pointer. (There is
still a tiny window in the thread that needs more work to close.)
Change-Id: Ifaddf7a495a8dae15a579a57bdc654f7c47f6ada
Signed-off-by: Todd Poynor <toddpoynor@google.com>
If the irqsoff tracer is in use, stop tracing the interrupt disable
interval when returning to userspace. Tracing userspace execution
time as interrupts disabled time is not helpful for kernel
performance analysis purposes. Only do so if the irqsoff tracer
is enabled, to avoid overhead purely for lockdep, which doesn't
care.
Change-Id: I74cde7dd0c4c3a783af34e76bd4efc1f85315abe
Signed-off-by: Todd Poynor <toddpoynor@google.com>
If a process forked and the child process was killed by the
lowmemorykiller, the lowmemory killer would be disabled until
the parent process reaped the child or it died itself.
Change-Id: I709b1a4e1b1a1970e51d26a39fcbee57977bbc7f
Signed-off-by: Arve Hjønnevåg <arve@android.com>
The lowmemorykiller registers an atomic notifier for notfication of when
the task is freed. From this atomic notifier callback, it removes the
atomic notifier via task_free_unregister(). This is incorrect because
atomic_notifier_chain_unregister() calls syncronize_rcu(), which can
sleep, which shouldn't be done from an atomic notifier.
Fix this by registering the notifier during init, and only unregister it
if the lowmemorykiller is unloaded.
Change-Id: I1577b04e617bc2b2e39dcb490fcfc9ce660eb7ec
Signed-off-by: Rabin Vincent <rabin.vincent@stericsson.com>
Signed-off-by: Christian Bejram <christian.bejram@stericsson.com>
- Improve BT-coex to differentiate HID and SCO device during DHCP session
- Enforce DTIM=1 for any SoftAP settings to optimize internal memory logic
Signed-off-by: Dmitry Shmidt <dimitrysh@google.com>
synchronize_rcu can be very expensive, averaging 100 ms in
some cases. In cgroup_attach_task, it is used to prevent
a task->cgroups pointer dereferenced in an RCU read side
critical section from being invalidated, by delaying the
call to put_css_set until after an RCU grace period.
To avoid the call to synchronize_rcu, make the put_css_set
call rcu-safe by moving the deletion of the css_set links
into free_css_set_work, scheduled by the rcu callback
free_css_set_rcu.
The decrement of the cgroup refcount is no longer
synchronous with the call to put_css_set, which can result
in the cgroup refcount staying positive after the last call
to cgroup_attach_task returns. To allow the cgroup to be
deleted with cgroup_rmdir synchronously after
cgroup_attach_task, have rmdir check the refcount of all
associated css_sets. If cgroup_rmdir is called on a cgroup
for which the css_sets all have refcount zero but the
cgroup refcount is nonzero, reuse the rmdir waitqueue to
block the rmdir until free_css_set_work is called.
Signed-off-by: Colin Cross <ccross@android.com>
Changes the meaning of CGRP_RELEASABLE to be set on any cgroup
that has ever had a task or cgroup in it, or had css_get called
on it. The bit is set in cgroup_attach_task, cgroup_create,
and __css_get. It is not necessary to set the bit in
cgroup_fork, as the task is either in the root cgroup, in
which can never be released, or the task it was forked from
already set the bit in croup_attach_task.
Signed-off-by: Colin Cross <ccross@android.com>
If the parent device is in the DPM_RESUMING state, resume might get called
before the parent's resume has completed, because dpm_wait won't get
called. This issue was introduced by:
3d46a3c PM: Prevent waiting forever on asynchronous resume after abort
Wait for completion of the parent resume if the parent state is >= DPM_OFF
or == DPM_RESUMING
Change-Id: I14a4e16426f61efceaefedc7e772f53b7522b5dd
Signed-off-by: Benoit Goby <benoit@android.com>
Microsoft recommends that the class and subclass fields of
an IAD match the same fields from the first interface in the
collection that the IAD is grouping. In practice, we are also
finding that the protocol fields should also match. Without
this change, the default Microsoft composite driver may not
group interfaces properly, which is what allows child
function drivers with IAD's to load correctly.
Signed-off-by: Mike Lockwood <lockwood@android.com>
bFirstInterface should be set equal to bInterfaceNumber of
the first interface in a grouping. It's currently being set
only when a composite device is bound, and it does not get
updated when functions are enabled or disabled.
Signed-off-by: Mike Lockwood <lockwood@android.com>
USB-IF core team mandates that composite devices containing
IAD's must use bDeviceClass=0xEF, bDeviceSubClass=0x02, &
bDeviceProtocol=0x01 instead of the usual 0x00 in all fields.
This is not a problem currently on Linux hosts, but it is a
problem when connecting to Windows hosts. Without this change
the default Microsoft composite driver will not group
interfaces properly, which is what allows child function
drivers with IAD's to load correctly.
Signed-off-by: Mike Lockwood <lockwood@android.com>
- Added multiple ACM instance support in Android gadget
- Fixed multiple instance naming issue in ACM function
- Increased max instances from 4 to 8
Change-Id: I65f1b0be94da859bab7ec0ad7cd804b896c7c4c5
Signed-off-by: John Michelau <john.michelau@motorola.com>
