To prevent un-initialized address returning from MHI
state transition lookup table, move mhi_states_transition_str
array out of stack
CRs-Fixed: 2111049
Change-Id: Ife5f907a711cc0c615161f2b2a677275d6773686
Signed-off-by: Sujeev Dias <sdias@codeaurora.org>
When using debugfs dev_boot node to trigger firmware download,
debug state should be set as well so that subsequence shutdown
and powerup can be handled correctly.
Change-Id: Id85d28dca104f8938d78655d682a0a7d1491c04c
CRs-fixed: 2059087
Signed-off-by: Yue Ma <yuem@codeaurora.org>
During firmware data read, it may leads to cross the max
range. Add a check to validate data size with in the max
range.
Change-Id: I12b3c1761214a7d2af2c45ad73bf61954c8398b6
Signed-off-by: Archana Obannagari <aobann@codeaurora.org>
This reverts commit c5616f2f874faa20b59b116177b99bf3948586df.
If we re-init the per-cpu boostgroup spinlock every time that
we add a new boosted cgroup, we can easily wipe out (reinit)
a spinlock struct while in a critical section. We should only
be setting up the per-cpu boostgroup data, and the spin_lock
initialization need only happen once - which we're already
doing in a postcore_initcall.
For example:
-------- CPU 0 -------- | -------- CPU1 --------
cgroupX boost group added |
schedtune_enqueue_task |
acquires(bg->lock) | cgroupY boost group added
| for_each_cpu()
| raw_spin_lock_init(bg->lock)
releases(bg->lock) |
BUG (already unlocked) |
|
This results in the following BUG from the debug spinlock code:
BUG: spinlock already unlocked on CPU#5, rcuop/6/68
CRs-fixed: 2113062
Change-Id: I1cd780d9ba5801cf99bfe46504b18a88e45f17a8
Signed-off-by: Vikram Mulukutla <markivx@codeaurora.org>
HDR playback needs metadata to be sent to the sink
while the playback is ongoing and needs a proper
teardown sequence when the playback has ended with respect
to the infoframe being sent to the sink.
This needs a state machine to synchronize start/stop of
the playback with sending the right metadata along with
resetting the infoframe HDMI registers.
Add support for this HDR playback control state machine.
Change-Id: I58ff03bc8db99febd17e0a0853acc74668892ea8
Signed-off-by: Ch Ganesh Kumar <chganesh@codeaurora.org>
Msm8996 adp lite board has two displays, native HDMI display and one
DSI-HDMI display through ADV7533 bridge chip. So only one ADV7533
instance is needed here. And at the meanwhile, total sde_kms components
need to match the configuration, otherwise, display driver probe will be
skipped due to this mismatch.
Change-Id: If596a64a9e47416e0810039914a958cae61c7e3e
CRs-Fixed: 2113147
Signed-off-by: Guchun Chen <guchunc@codeaurora.org>
Adding capability to read OIS data, provided in the driver
register settings.
CRs-Fixed: 2114654
Change-Id: Ie10b55f597a4daa29267c6d4333ce9d9c7702d63
Signed-off-by: Xu Han <hanxu@codeaurora.org>
Added feature to configure video h/w to send empty output buffer
for skipped frame due to rate control.
Change-Id: Id47344c3bae216fec6384ff205bcb658ec27cb06
Signed-off-by: Shiju Mathew <shijum@codeaurora.org>
Support primary TDM interface dai link and export kcontrols to update
primary TDM interface slot configuration.
CRs-fixed: 2095930
Change-Id: I693b99f295f6476c96d681e10459842c0011467f
Signed-off-by: Cong Tang <congt@codeaurora.org>
Platform driver only forces FW assert for supported devices.
Return proper error code for nonsupported devices so that WLAN
host driver can handle properly.
Change-Id: Iaf88f64564b14425404404db6a7789a7913db74f
CRs-fixed: 2059087
Signed-off-by: Yue Ma <yuem@codeaurora.org>
Add CLD80211_ATTR_META_DATA vendor attribute, which will be used by
userspace RTT application to send meta data to wlan driver allowing
it to peek into RTT request message without opening up complete
definition of RTT message.
Change-Id: If2f8253b59c6860fbaecab05f1b76f11da1ddb6f
CRs-Fixed: 2114181
Signed-off-by: Naveen Rawat <naveenrawat@codeaurora.org>
Use CNSS2 platform driver instead of CNSS in AUTO platforms.
Change-Id: I0b5df067d05ceafd759b3cefab1b506a659af50e
CRs-fixed: 2071634
Signed-off-by: Yue Ma <yuem@codeaurora.org>
During recovery, there are cases where in WLAN driver running on
APPS can access CE register causing exceptions because of
hardware in reset state. Fix the issue by not resetting the
hardware in WLAN firmware during recovery but halt the CE and MAC
and reset the hardware only when reinitialization starts. For
this to work, wcn3990 ath10k driver needs to keep the clock and
regulator votes so that WLAN hardware will not go into weird
state because of not resetting it but cutting the resources.
Change-Id: I765fd4aaa2fea6f41f4760564b8d1cb8c3d9651b
Signed-off-by: Govind Singh <govinds@codeaurora.org>
Add clock and regulator nodes for WCN3990.
During recovery, there are cases where in WLAN driver running on
APPS can access CE register causing exceptions because of
hardware in reset state. Fix the issue by not resetting the
hardware in WLAN firmware during recovery but halt the CE and MAC
and reset the hardware only when reinitialization starts. For
this to work, wcn3990 ath10k driver needs to keep the clock and
regulator votes so that WLAN hardware will not go into weird
state because of not resetting it but cutting the resources.
Change-Id: Iaaa793956bed687e4d76acda93f7fa467773f013
Signed-off-by: Govind Singh <govinds@codeaurora.org>
During soundcard deregistration, free_chmap() is called as part of
pcm_free. This only frees downmix kcontrol as pcm has reference
only to downmix kcontrol followed by freeing of pcm pointer. The
other two kcontrol are freed as part of snd_ctl_remove(). However,
while freeing other two kcontrols, pcm_usr_ctl_private_free()
accesses pcm->streams which are already freed causing crash.
Fix is to add pan/scale & downmix controls as platform controls
instead of user control.
Change-Id: Ib2890122d079201e02c4b7839694d2e7618037b7
Signed-off-by: Varun Balaraj <varunb@codeaurora.org>
When there's a dynamic channel change from mono to stereo,
volume burst is observed because center channel gain for
stereo is not initialized. Initialize center channel gain
for both mono and stereo playback.
CRs-Fixed: 2095081
Change-Id: Ib92236f55b1f171603be956d00c8fcab7c4169d3
Signed-off-by: Yidong Huang <yidongh@codeaurora.org>
ION page pool and cma shrinkers return the total pooled
memory instead of the freed. This can result in an early
exit of ION page pool shrinker thus giving back less memory
to the system. For cma shrinker this change doesn't have an
effect since the callers of shrink_slab currently ignores
the return value.
Change-Id: I0b3727807d9a5f020623766d27da360d077395f3
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
Sde client customization is disabled by previous commit which
is propagated from other branch. After disabling this, 4K display
on HDMI can not be up. This is because when creating drm plane in kernel,
sde client customization will directly set possible_crtcs per plane
reporting to user space. When disabling it, user space's sdm strategy
could not get and assign correct pipes for 4K display.
Change-Id: Ie6a870c352533060f7c810bdb5c99b7c1534117d
CRs-fixed: 2107328
Signed-off-by: Guchun Chen <guchunc@codeaurora.org>
This patch disables Rx and Tx engine properly while going to suspend or
shutdown. There are chances when BT SOC or client is going bad and
disrupts the UART SW state machine.
1.Do not allow double resource_vote and hence pipe connect at port open.
2.While going to suspend, close the HW resources forcibly to prevent
UART from going bad for the subsequent resume cycle.
3.Stop Rx engine gracefully even if clocks are off to obey the framework.
4.During resource unvote, if runtime PM is disabled then go for forced
suspend.
5.In case of fatal error, stop further IPC logging.
6.Add relevant log if the Rx pipe is not empty at the time disconnecting.
7.Prevent Rx byte entering in case of back to back loopback test, as
this causes data mismatch.
8.Add the logs correctly just after reading the status instead doing it
later which doesn't get logged in case of any failure.
Change-Id: I9426e9ae8c5b07d74e95ec37651ece0d46928104
Signed-off-by: Mukesh Kumar Savaliya <msavaliy@codeaurora.org>
Pagetable pgd is allocated in arm_smmu_init_domain_context and is freed in
arm_smmu_domain_free. Client can attach/detach domain without freeing the
same. This causes memory leak. This was introduced with
commit fe676929a9 ("iommu/arm-smmu: fix a DOMAIN_ATTR_DYNAMIC memory
leak")
Free io pgtables in arm_smmu_destroy_domain_context, which is invoked at
the time of domain detach.
Change-Id: Ib53e4efdd8dca33437596f3f0ccaa8ccaae79ca0
Signed-off-by: Prakash Gupta <guptap@codeaurora.org>
Disable USB IRQs after high duty cycle IRQ triggered and re-enable
them in the delay worker. This helps to avoid the high duty cycle
IRQ storming.
CRs-Fixed: 2094725
Change-Id: Iba22a7402692b5ee5f738cc17ed9c214f6014ac0
Signed-off-by: Fenglin Wu <fenglinw@codeaurora.org>
Currently the driver skips invoking adc api's if usb is not
present i.e. vbus has fallen below 3.3V.
This presence check is not required for the vbus. Skip that
check.
Change-Id: I4a2dd91109a308b553b7407618ee8f3221127f61
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
The export of save_stack_trace_tsk() was previously backported to
msm-4.9, without a backported flag to detect support. Define a new
backport flag, so kernel modules can use this function when it is
exported.
Change-Id: Iec65e6f9fed0a9bf686f5e5c8ba4f59691eb1ee0
CRs-Fixed: 2105458
Signed-off-by: Dustin Brown <dustinb@codeaurora.org>
Dfps info is calculated for a specific msm part and needs to be
used for the same. Add a chip serial number for the dfps info so
that the PLL trim codes are used for the same msm part on which
they are generated.
Change-Id: I6f1e519c1e9278b35dfbdf3d9ba4a2ed0e1a6a78
Signed-off-by: Ashish Garg <ashigarg@codeaurora.org>
Max ASM session upperbound condition is corrected to support
8 interactive plus a primary streams playback.
Change-Id: I5ec29c70941f52af020a64cc58bf10be52354f1b
Signed-off-by: Varun Balaraj <varunb@codeaurora.org>
Add laser LED device for msm8998 MTP to support
depth sensor.
Change-Id: I8603540d56adbc0d0d3ee41959f7e235dc459a8d
Signed-off-by: Vijay kumar Tumati <vtumati@codeaurora.org>
Prevent deadlock between tasklet and delete_stream by stopping
irq during delete_stream.
Change-Id: I4aba47b5cab2722c352139f793b75d7c86cd49d8
CRs-fixed: 2076578
Signed-off-by: Rahul Sharma <sharah@codeaurora.org>
Currently the driver moves FCC to PMI as it steps down
from SMB. Instead we should step down just the SMB.
Change-Id: I7ad9ca4364a44bacb539f8b045cf97f7945155f5
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
Currently the code only waits 100mS between steps. The hardware
could take longer to deglitch its signals and status. Extend the
wait to 500mS.
Change-Id: I4a037f85f43bc670c5a357e4ecb46962b0d33598
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
If userspace sends non-null terminated channel name then
out of bounds read is possible while printing channel name.
Fix this issue by validating channel name before use.
Change-Id: I418b7f446c6a849450dd4314592c1d3810163bdc
Signed-off-by: Vidyakumar Athota <vathota@codeaurora.org>
Add range check for wdsp_cmpnt_type to be within max
and greater than 0 before accessing the array to get
the component.
CRs-fixed: 2068865
Change-Id: Ibc3128297a024f7d6d548c946a035e604f573b64
Signed-off-by: Karthikeyan Mani <kmani@codeaurora.org>
Update max tx power of country BZ/IN/KN/PA/TT/UZ/VE
Update max bandwidth of country CA
Add country GI/IQ
Remove channel 14 of country JP
Change-Id: Iacbb07b222d1c15e9109d8bbfac797fd889444a0
CRs-Fixed: 2094357
Signed-off-by: Will Huang <wilhuang@codeaurora.org>
Update upper bound of 5GHz channels for country MY from 5815 to 5835
Change-Id: Ib11cfa97316b90bc4ffb50420c84df187ff80a7f
CRs-Fixed: 2103994
Signed-off-by: Will Huang <wilhuang@codeaurora.org>
Signed-off-by: Ashish Kumar Dhanotiya <adhanoti@codeaurora.org>
Adding new submodule for lased LED and APIs to
support register WRITE/READ operations.
Change-Id: I7e6dc27a01985c223be791b16e6b7da4bbc1be74
Signed-off-by: Vijay kumar Tumati <vtumati@codeaurora.org>
In most code paths involving cgroup migration cgroup_threadgroup_rwsem
is taken. There are two exceptions:
- remove_tasks_in_empty_cpuset calls cgroup_transfer_tasks
- vhost_attach_cgroups_work calls cgroup_attach_task_all
With cgroup_threadgroup_rwsem held it is guaranteed that cgroup_post_fork
and copy_cgroup_ns will reference the same css_set from the process calling
fork.
Without such an interlock there process after fork could reference one
css_set from it's new cgroup namespace and another css_set from
task->cgroups, which semantically is nonsensical.
Change-Id: Iadec82bd07a43bf7cefcf8fc17603700fd4895c3
Cc: stable@vger.kernel.org
Fixes: a79a908fd2b0 ("cgroup: introduce cgroup namespaces")
Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Git-commit: eedd0f4cbf5f3b81e82649832091e1d9d53f0709
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
Signed-off-by: Neeraj Upadhyay <neeraju@codeaurora.org>
