FCC and Float voltage is configured by battery driver using
power_supply framework (main_psy), re-run FCC/FV election once
main_psy is available to ensure FCC/Float voltage is configured
and reflected on hardware.
While at it, add check for valid "pl_psy" before using it.
CRs-fixed: 2028082
Change-Id: I2f5dc174eacf325ba27186b07c89bb7d438f061b
Signed-off-by: Ashay Jaiswal <ashayj@codeaurora.org>
A livelock can be created in the system by lowmemorykiller trying to kill
the same task again and again. Below is the livelock condition.
P0 -> binder_main_lock(W)
P1 -> binder_main_lock(T)-> mmap_sem(W)
P2 -> mmap_sem(T) -> lowmem_scan::scan_mutex(W)
P3 -> lowmem_scan::scan_mutex(T) -> send SIGKILL P0
Here multiple tasks are contending on scan_mutex, and binder_main_lock.
Change lowmem_scan mutex_lock with mutex_trylock, so in case of lowmem_scan
lock contention, task will be allowed to reclaim from other shrinkers. This
will also maintain the serialization of lowmemorykiller trigger.
If a task is pending MEMDIE'ing, remove sleep before falling back on other
shrinkers.
If the task selected to be killed is MEMDIE'ing and in un-interruptible
sleep state, do not repeat kill but fallback on other shrinkers without any
delay.
Change-Id: I12131622f7fa7b422c6d5d09f782af848300e412
Signed-off-by: Prakash Gupta <guptap@codeaurora.org>
The register BATTERY_CHARGER_STATUS continues to show charging state
(like FULLON or TAPER mode) in cases where charging was paused due
to input OV. Look at BATTERY_CHARGER_STATUS_7_REG to determine if
charging was paused.
This fix modifies previous similar change for JEITA hard condition but
continue to cover the case.
Change-Id: Ibc5f4f5e85651708b656f06814008b0c319db02d
Signed-off-by: Harry Yang <harryy@codeaurora.org>
Currently a legacy cable is always assumed which causes decreased
performance from non-legacy HVDCP adapters with 10k ohm Rp.
Fix this by disabling and re-enabling Type-C to rerun the legacy cable
detection.
Moreover, the legacy cable IRQs are not used and cause unnecessary
type-c-change IRQs to fire. Disable them.
CRs-Fixed: 2020132
Change-Id: I57fc3762251ead028298f01b06d66f52fd119c6b
Signed-off-by: Nicholas Troast <ntroast@codeaurora.org>
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
Currently the USB Type-C status is retrieved from the hardware in real
time. The Type-C change IRQ should trigger on every change of the Type-C
status, therefore it is not necessary to read the status registers
multiple times if a Type-C change IRQ has not triggered. Furthermore,
workarounds which force UFP/DFP mode, or disable Type-C altogether could
mislead the software into thinking a removal has happened.
Cache all of the USB Type-C status registers upon receiving a USB Type-C
change IRQ, and use the cached status where appropriate.
CRs-Fixed: 2020132
Change-Id: I99f2ff29633207898ae803672162db0c3cec80dc
Signed-off-by: Nicholas Troast <ntroast@codeaurora.org>
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
During USB removal a lot of cleanup happens; votables are reset, flags
are cleared, etc. After the cleanup is finished there is a chance that
USB power supply consumers may set properties before getting the USB
removal notification. This can lead to many problems where ICL limits
are set based on the previous insertion, or APSD is disabled due to a
late setting of PD_ACTIVE.
Introduce a lock which prevents USB power supply consumers from setting
properties when USB has been removed. This lock will ensure that the
next insertion starts with a clean slate.
CRs-Fixed: 2020132
Change-Id: I05a4145289b6097e41afc30aa09782722fa03fb6
Signed-off-by: Nicholas Troast <ntroast@codeaurora.org>
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
Currently it is expected upon USB removal that consumers will receive a
notification that USB has been removed and will cleanup after
themselves. Unfortunately this makes it very difficult to keep track of
all of the necessary cleanup steps upon removal.
Also, in preparation of not allowing consumers to set properties after a
USB removal has happened it is now the responsibility of the charger
driver to cleanup on their behalf.
Moreover, since we don't have a separate removal/insertion interrupt,
the removal/insertion code may run even if the typeC cable is not
physically inserted/removed. Fix it by tracking a typec_present flag.
CRs-Fixed: 2020132
Change-Id: Ia514abaa4e12f72daec17fd1e95f3c51cc38a15f
Signed-off-by: Nicholas Troast <ntroast@codeaurora.org>
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
It is possible that LSM (Listen Stream Manager) could have an usecase
that does not need any audio calibration. In such cases, currently
the driver returns an error and does not allow the usecase setup with
LSM causing the usecase to fail. Fix this by allowing zero size audio
calibration for LSM.
CRs-fixed: 2031503
Change-Id: I42d89792a4fd6a0d6e1908a36720680b63ff0c30
Signed-off-by: Bhalchandra Gajare <gajare@codeaurora.org>
Currently the CC2 removal workaround starts whenever PD issues a hard
reset. When PD issues a hard reset it is not guaranteed that VBUS will
fall since the source may not even be PD capable.
The CC2 removal workaround should only run during the time that VBUS is
low and CC is debounced.
Fix this by scheduling the CC2 removal workaround when VBUS falls and CC
is debounced, and cancel the workaround when either VBUS rises, or the
removal detection is successful.
CRs-Fixed: 2020132
Change-Id: I6475d37911d90805ed8b3bb4b3a26a9f7557ebd6
Signed-off-by: Nicholas Troast <ntroast@codeaurora.org>
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
AFE sidetone is currently enabled based on a combination
of mixer control setting and default config in ACDB.
This change will limit enablement to mixer control setting alone.
Change-Id: I038d51177adc3e1da45ea7fdf9386362390f181d
Signed-off-by: Vikram Panduranga <vpandura@codeaurora.org>
Refactor the fs readpage/write tracepoints to move the
inode->path lookup outside the tracepoint code, and pass a pointer
to the path into the tracepoint code instead. This is necessary
because the tracepoint code runs non-preemptible. Thanks to
Trilok Soni for catching this in 4.4.
Change-Id: I7486c5947918d155a30c61d6b9cd5027cf8fbe15
Git-commit: d854b68890
Git-repo: https://android.googlesource.com/kernel/common/
Signed-off-by: Mohan Srinivasan <srmohan@google.com>
Signed-off-by: Runmin Wang <runminw@codeaurora.org>
Adds tracepoints in ext4/f2fs/mpage to track readpages/buffered
write()s. This allows us to track files that are being read/written
to PIDs.
Change-Id: I26bd36f933108927d6903da04d8cb42fd9c3ef3d
Signed-off-by: Mohan Srinivasan <srmohan@google.com>
Git-commit: 32cbbe5953
Git-repo: https://android.googlesource.com/kernel/common/
[runminw@codeaurora.org: resolve trivial merge conflicts]
Signed-off-by: Runmin Wang <runminw@codeaurora.org>
After cdm has been used, need to reset cdm block,
or the next HDMI device will be affected by the
cdm config.
Change-Id: I4eb879202cc3547d9149b3352377c3395ebfe6b3
Signed-off-by: zhaoyuan <yzhao@codeaurora.org>
Queue the msg receive task in service notifier queue
so that it does not run concurrently with other notifier task.
This avoid an issue where adsp ssr is stuck due to deadlock
between msg receive and service arrive task.
Change-Id: I6ef9b765ae74eeb32021c2848ffc06d70df19c1b
Signed-off-by: Avaneesh Kumar Dwivedi <akdwived@codeaurora.org>
Make opened device count atomic variable to avoid probable race
condition. Race condition leads to memory leak and list corruption.
Change-Id: I4da98f27d36f616bc8fa7b1a848c20cc7eea04e5
Signed-off-by: Divya Ojha <dojha@codeaurora.org>
There are some cases where MDSS GSDC is not turned off after FB
PM suspend ever after clock ref count is 0 as runtime suspend
is not triggered. Ensure that MDSS GDCC is toggled in these
cases.
Change-Id: I33389ad736960b619b32a9bec4b2b157eed4d20b
Signed-off-by: Jayant Shekhar <jshekhar@codeaurora.org>
WARN_ON is unnecessary if pre-alloc table doesn't have any free
memory. Remove the WARN_ON as error log is enough to find out
missing entry.
Change-Id: I5a46e1f259e88d1a19f05195f5d7bb0745d072c3
CRs-fixed: 2030272
Signed-off-by: Prashanth Bhatta <bhattap@codeaurora.org>
GPIO21 is connected to SMB1381 STAT pin for interrupt detection,
configure it to pull up for sensing SMB1381's interrupt.
CRs-Fixed: 2033882
Change-Id: Ib88aac4acb3b1094adb13839cccf1aa27903b9c7
Signed-off-by: Yingwei Zhao <cyizhao@codeaurora.org>
In case of very slow insertion of SDP/DCP there is a possibility
that D+/D- makes contact while APSD is in progress. This will
result in an incorrect type detection.
Fix this by doing a APSD rerun after charger-type detection is
complete.
CRs-Fixed: 2032590
Change-Id: I0037b90f29dbe65a2cdb2771d5caceff77862f03
Signed-off-by: Ashay Jaiswal <ashayj@codeaurora.org>
Since on some platforms DSI_CMD_OFFSET register has become
double buffered, so dynamic fps update and DSI DMA command
transfer can not happen at the same time. This changes makes
panel status check which in turn does a DSI DMA transfer, and
dfps update mutually exclusive.
Change-Id: If8591c55d31669dbf3f565db041c04bcd6cb616a
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
When the need for CPUs is more than the active CPUs and there are some
isolated CPUs, we wakeup the core_ctl thread to unisolate some CPUs.
The core_ctl task can't unisolate any CPU if all of them are isolated
by other clients. Track the number of isolated CPUs by core_ctl and
wakeup the core_ctl task when adjustment is really possible.
Change-Id: I11ef10860532df25cbde572aabd4b925320db8fe
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
The cluster->active_cpus is not updated in eval_need(). The new need
for CPUs is compared against the previous cluster->active_cpus. If
another client isolates a CPU, cluster->active_cpus becomes stale and
we fail to detect the change in need for CPUs.
Change-Id: Ib58b8f0bd03dd2b4a174de2ac54eb0c60c59f9f7
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
We are interested in only CPU_ONLINE and CPU_DEAD notifications. Don't
do anything when other notifications arrive.
Change-Id: Iea2e0e1c93e67ef278ee7c5a9813fbab6cea5c74
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Enable register read based ESD check mechanism for nt35597
truly command mode panel used on sdm660 platform, as per HW
recommendation.
Change-Id: I11f897a24d29f215901c2d95d1c5070716ca6515
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
In the current implementation DSI driver is masking the
fake ACK and overflow errors that might occur during if
BTA mechanism is used to check ESD. But if register read
mechanism is used to check ESD, then also embedded BTA will
be triggered and fake ACK and overflow errors might be
reported. Mask the same for register based ESD check also.
Change-Id: If1ee0a7cc0171b96a3b7298aa5201372c6eb8139
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
When the downstream DP device is physically disconnected, ensure
that all pending events are removed and the event thread is parked.
Reset all the software state so that subsequent connection events
are handled correctly.
CRs-Fixed: 2034023
Change-Id: Ie94b1da903b8e78509220e373bec4ff54026885c
Signed-off-by: Aravind Venkateswaran <aravindh@codeaurora.org>
Fix race condition in the release of the mdp debugfs functions
panel_debug_base_release and mdss_debug_base_release by adding
the lock for unpreempted freeing of the buffer so that multiple
concurrent processes cannot affect the release which can possibly
lead to use-after-free operation on the buffer.
Change-Id: I9586081b65ae2eb0e7f6e30c606ee748ae9ef7e8
Signed-off-by: Harsh Sahu <hsahu@codeaurora.org>
Enable page poisoning by default on MSM8998
and SDM660.
Change-Id: If0b873888f21fd7e7057df43cf223e3c50372bed
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
Add a config option to enable page poisoning by
default. The kernel command line option "page_poison"
can be used to change the behaviour during boot.
Change-Id: Ie70763841191a722b1c6125dfad119a29ed0f605
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
On SPARSEMEM systems page poisoning is enabled after buddy is up, because
of the dependency on page extension init. This causes the pages released
by free_all_bootmem not to be poisoned. This either delays or misses the
identification of some issues because the pages have to undergo another
cycle of alloc-free-alloc for any corruption to be detected.
Enable page poisoning early by getting rid of the PAGE_EXT_DEBUG_POISON
flag. Since all the free pages will now be poisoned, the flag need not be
verified before checking the poison during an alloc.
Link: http://lkml.kernel.org/r/1490358246-11001-1-git-send-email-vinmenon@codeaurora.org
Acked-by: Laura Abbott <labbott@redhat.com>
Tested-by: Laura Abbott <labbott@redhat.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Akinobu Mita <akinobu.mita@gmail.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
[vinmenon@codeaurora.org: resolve trivial merge conflicts.
Remove the redundant free pages RO feature from the
page_poison.c file which is the reason for conflicts +
squash the addendum commit 40961ef8d65f51093bc94de110b97b590b6b9275
('mm-enable-page-poisoning-early-at-boot-v2')]
Git-commit: c5b7cd344fd6341e6db79e55c0f1f4d1d9c67a7e
Git-repo: git://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git
Change-Id: I1bb1f99d3a2e1135131911905e0916c837ba9d8a
Signed-off-by: Vinayak Menon <vinmenon@codeaurora.org>
