Even if we use status = "disabled" in mem-offline node, still the
update_memory_limit() will go and update memory_limit variable and
thus the removed memory is never added to system as the
mem-offline driver is disabled. Correct this logic.
Change-Id: I80102830f0a5a82dcd4cddfaddb0de21bfdd0110
Signed-off-by: Charan Teja Reddy <charante@codeaurora.org>
Signed-off-by: Isaac J. Manjarres <isaacm@codeaurora.org>
Consider the memory limit enforced by boot parameter and
restrict adding beyond this boundary during a memory hotplug.
Change-Id: Ibd13e47ee13004381a1ddff649150fae13429c87
Signed-off-by: Vijayanand Jitta <vjitta@codeaurora.org>
Signed-off-by: Isaac J. Manjarres <isaacm@codeaurora.org>
Check for the current state of sub system restart notifier when
the LOCATOR_UP notification is given to slimbus driver and do the
slimbus initialization if the current state is already set to
SERVREG_NOTIF_SERVICE_STATE_UP_V01.
This change is needed in the cases where LOCATOR_UP notification
to slimbus driver got delayed and the state of the sub system restart
notifier is already set to SERVREG_NOTIF_SERVICE_STATE_UP_V01 as
slimbus will be expecting the SERVREG_NOTIF_SERVICE_STATE_UP_V01
notification which is not going to come as the notifier state is
already changed.
Change-Id: I4804bbb13e065ac78be7a442993286532402db9d
Signed-off-by: Prudhvi Yarlagadda <pyarlaga@codeaurora.org>
Prevent system suspend from happening during slimbus initialization
process where the NGD Slimbus component gets enummerated with
ADSP slimbus component and then queries the logical addresses of
the clients connected.
This is to prevent the issue where the slimbus driver goes to
suspend due to system suspend after the enumeration phase is
done after an SSR/PDR. Driver again resumes for querying logical
addresses of clients connected assuming enumeration process is
already done whereas it has to be re-enumerated as driver got
suspened and resumed.
Change-Id: I42ab3f7a6061a1c464d436aa13e43d27eb312fd4
Signed-off-by: Prudhvi Yarlagadda <pyarlaga@codeaurora.org>
Enable support for ultrasonic fingerprint
driver in lahaina defconfig.
Change-Id: Ib66316b7b3e6fc186f6f5a92c50e62e1332e92b8
Signed-off-by: Abir Ghosh <abirg@codeaurora.org>
Add support to calculate maximum current available for flash module.
Clients of flash LED makes use of this API qti_flash_led_prepare()
to query maximum available current for different flash LED channels.
Change-Id: I26988b1afb1e57a190aa25061189b71ca3f7ea95
Signed-off-by: Shyam Kumar Thella <sthella@codeaurora.org>
Add changes to make touch events notifier static.
Change the return type of qbt_poll from unsigned int
to __poll_t.
Change-Id: Ib43f6bb5ba77cb48d1e47e1ecd1d5b810f7c9159
Signed-off-by: Abir Ghosh <abirg@codeaurora.org>
Change the way the HOSTLD and HOSTAR are assigned in the envsetup.sh
so that the original values don't get overwritten.
Change-Id: Ia2dd91d47b1ab3a527305973a67d8b5b5368a607
Signed-off-by: Prakruthi Deepak Heragu <pheragu@codeaurora.org>
Commit 052553af6a ("ufs/phy: qcom: Refactor to use phy_init call") puts
enabling regulators & clks, calibrating UFS PHY, starting serdes and
polling PCS ready status into phy_power_on.
Due to UFS PHY's registers are retained after power collapse, so
calibrating UFS PHY, starting serdes and polling PCS ready status need to
be done only when start hba, but not needed every time when phy_power_on
is called during resume. Keep the codes which enables PHY's regulators &
clks in phy_power_on but move the rest steps into PHY's general calibration
func, namely phy_calibrate.
Commit 3f6d1767b1 ("phy: ufs-qcom: Refactor all init steps into
phy_poweron") removes the phy_power_on/off from ufs_qcom_setup_clocks.
To have a better power saving, remove the phy_power_on/off calls from
resume/suspend path and put them back to ufs_qcom_setup_clocks, so that
PHY's regulators & clks can be turned on/off along with UFS's clocks.
At last, add a mutex lock to protect the usage of is_phy_pwr_on against
possible racing.
Change-Id: I1ed5d5d6968c3f16018e9723eab010ea7ed798b2
Signed-off-by: Can Guo <cang@codeaurora.org>
The bus bandwidth voting is required to be done before the bus clocks
are enabled, and the unvoting is required to be done only after the bus
clocks are disabled.
Change-Id: Ie231d9c373c0d8c8f3dd90b8322ff24fabdfae26
Signed-off-by: Can Guo <cang@codeaurora.org>
CONFIG_HAVE_LOW_MEMORY_KILLER is no longer present on 5.4, so remove a
check to see if it's enabled or not.
Change-Id: Ifcd88a0e29b7da6dede7f98fcb4c99ffc7bf3c02
Signed-off-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
This setting is useful to debug memory intensive test scenarios,
which are close to the limit of what the device supports.
Change-Id: Id09f43736bd3c9586b4dac4fae570e78089f5bb3
Signed-off-by: Patrick Daly <pdaly@codeaurora.org>
[cgoldswo@codeaurora.org: Add macro introduced in obsolete patch]
Signed-off-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
Enable CONFIG_PAGE_EXTENSION_PAGE_FREE and CONFIG_PAGE_FREE_TIME, which
provides information on (1) if a page from buddy is free or not, (2) the
timestamp of when a given page was freed, if it currently is free.
Change-Id: I702c93192e031bbeb023879e47c2202f0c8d65f9
Signed-off-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
When allocating higher order pages, currently only the head page is set
with page owner info, leaving the tail pages with previous or stale info.
This patch sets the info for tail pages too and sets the time of free in
page owner info of all pages during free. This patch also adds an extra
info to page extension flags to help check if page is in buddy or not.
Change-Id: I96d13485af698d586a4dd7b1092a379a645c8fcb
Signed-off-by: Sudarshan Rajagopalan <sudaraja@codeaurora.org>
[cgoldswo@codeaurora.org: Fix merge conflicts, add GKI #ifdef guards]
Signed-off-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
When Android has a userspace low memory killer, we want to configure
certain parts of the reclaim paths in the kernel.
HAVE_USERSPACE_LOW_MEMORY_KILLER can be used to perform this configuration,
and was removed when reverting a patch that defined it. Re-add it.
Change-Id: Ic069c92a9d7928f0402705df8375d5384e9f743c
Signed-off-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
If MHI force wake get returns failure, it means MHI force wake
put has already been done. Add checks to avoid double calling.
Change-Id: I14f8cef5e6736130e76c287748d7ce3ff0023505
Signed-off-by: Yue Ma <yuem@codeaurora.org>
Retry a few times if PCI link training fails to avoid device side is
not ready in time.
Change-Id: Icd88673b19a4d6f877f597d1c636438f932f85a7
Signed-off-by: Yue Ma <yuem@codeaurora.org>
With LTO, modules with a large number of compilation units maybe end
up exceeding the for loop argument list in the shell. Reduce the
probability for this happening by including only the modules that have
exported symbols.
Bug: 150234396
Change-Id: I4a289aff47e1444aca28d1bd00b125628f39bcd5
Suggested-by: Hsiu-Chang Chen <hsiuchangchen@google.com>
Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Git-commit: 1245167
Git-repo: https://android.googlesource.com/kernel/common
Signed-off-by: Prakruthi Deepak Heragu <pheragu@codeaurora.org>
The context fault handler uses a lot of magic numbers
that now have proper constants defined, so use those
constants.
Change-Id: Ic775bdbb4d617d4f96c935d7f18516e9d4740b9e
Signed-off-by: Isaac J. Manjarres <isaacm@codeaurora.org>
