int_count variable datatype is uint64 in tz version 2.X/3.0, and
is changed to uint32 in tz version >=4.X, so updating int_count
according to the TZ version.
Change-Id: I72e128e6ad41711fd49416faf6aebfc869187ee2
Signed-off-by: Zhen Kong <zkong@codeaurora.org>
Endpoint command to request resources on ep0 is take upto ~800us and
this delay is observed for every new/re enumeration. As the api polls
(reads the register and it happens over cnoc) for every 1us to check
for completion it could result in congestion on cnoc. Hence increase
polling interval from 1us to 20us for this endpoint command.
Change-Id: I6d738b46a47b00105289e8f07b8fe0883a3fda44
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
There is a possibility of port change event triggering
xhci irq as soon as halt bit is cleared in xhci_start().
As a result before xhci state is changed from XHCI_STATE_HALTED
to 0 port change event keeps on generated until port status is
acknowledged. This does not allow xhci_start() to finish and handle
the port change event if irq is keep on getting fired on same core
where xhci_start() is running. Fix this issue by disabling irq before
starting controller and enable it back after clearing halt bit.
Change-Id: I798620f99a7ba522258455642e6e8091ebf2cd34
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
mhi clients can manage both hardware and software event rings.
Process incoming data based on event ring type instead of
channel type.
CRs-Fixed: 2018326
Change-Id: Ifa2de76034224c19a846f7be153400888f8c66de
Signed-off-by: Sujeev Dias <sdias@codeaurora.org>
When a client ask MHI host to processed X number of events, only count
transfer events as part of processed events. All other events are internal
events.
CRs-Fixed: 2018324
Change-Id: I4cfcae0ba187348442c3fc45b76d7c4b01f687bd
Signed-off-by: Sujeev Dias <sdias@codeaurora.org>
Since the PTE shareability values can be multibit fix so that they
are properly checked.
Change-Id: I36791921b0189b50cbbcb79efdb2f1a36ad9d726
Signed-off-by: Liam Mark <lmark@codeaurora.org>
Fix the symbol linking error for loadable kernel modules
which is using secure buffer hypervisor api for smmu configuration.
CRs-Fixed: 2019262
Change-Id: I3d83fd4b1854298c1283e03f220de8c673195492
Signed-off-by: Sarada Prasanna Garnayak <sgarna@codeaurora.org>
gsi_reg.h file includes the definitions of GSI registers.
The GSI Instruction RAM registers are defined twice.
This change removes the duplication.
Change-Id: Iccd256711c28a5f4698555079a7f73ce0357a92d
Signed-off-by: Ghanim Fodi <gfodi@codeaurora.org>
Increase the common CMA region size to 44MB to accommodate modem diag
buffer region.
Change-Id: Iad83f749d8ca5635ead9c637dbc2f927191414fe
Signed-off-by: Charan Teja Reddy <charante@codeaurora.org>
For sdm660 the ULPS clamp registers are moved from
DSI PHY level to interface level. This change handles
the new settings for both video mode and command mode
interface.
Change-Id: Ib99c605362aa25aacfa394c1a7ffe2886acf7035
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
If both irq chip and external interrupt controller implement
irq_disable callback then disable_irq() API disables an interrupt
at external interrupt controller level but does not disable at GIC
level due to lazy approach. During the cpu hotplug gic_set_affinity()
checks whether IRQ enabled at GIC level or not and enabling IRQ at
external interrupt controller level too after IRQ migration to new cpu
even though IRQ was disabled. So add a check in gic_set_affinity()
to check whether IRQ was disabled in descriptor before enabling.
Change-Id: Iba81881c08102ac0a5960c3a97c1c9f8c477bd56
Signed-off-by: Prasad Sodagudi <psodagud@codeaurora.org>
As per the hardware documentation, update the regulator
configuration for apc0, apc1, gfx CPR regulators to support new
speed bins in msm8996pro.
CRs-Fixed: 1083423
Change-Id: I97b9b98305e3478012664b636df26299eca66f26
Signed-off-by: Tirupathi Reddy <tirupath@codeaurora.org>
Add a VDD_APCC CPR boost corner of 1996.8 MHz for the power
cluster on MSM8996-Pro speed bin 1 parts.
Change-Id: I9bf29cdba47918770dbef5cd1f2631afccf26221
CRs-Fixed: 1010722
Signed-off-by: David Collins <collinsd@codeaurora.org>
Change the MinSVS fuse corner to virtual corner mapping for the
power cluster and performance cluster on MSM8996-Pro chips with
CPR fuse revision 2 or greater.
Change the mapping according to the following:
Power cluster: 1 (307.2 MHz) --> 3 (460.8 MHz)
Perf cluster: 1 (307.2 MHz) --> 4 (537.6 MHz)
Change-Id: I6b6ee6225d43605ad0f5ae1e16061acf12b49927
CRs-Fixed: 1004373
Signed-off-by: David Collins <collinsd@codeaurora.org>
Update the VDD_GFX CPR open-loop and closed-loop voltage
adjustments for MSM8996-Pro chips based upon characterization
results. This ensures stability and minimum power consumption.
Change-Id: I89b79f92bcb4cc54a050f8b0ba17f76ff471838f
CRs-Fixed: 989555
Signed-off-by: David Collins <collinsd@codeaurora.org>
Set the CPR IRQ affinity of the VDD_APCC CPR3 controller to be
both cores of the APPS power cluster (i.e. CPU0 and CPU1). This
ensures that neither of the CPU cores of the performance cluster
will be woken up to service a VDD_APCC CPR IRQ which was
generated when the last performance cluster core power collapsed.
Change-Id: I055e50ffcb85622ddd67d55b44d77c342e9ec074
CRs-Fixed: 949650
Signed-off-by: David Collins <collinsd@codeaurora.org>
core_ctl module no longer hotplug the CPUs. Hence remove the
/sys/devices/system/cpu/cpuX/core_ctl/cpus file interface which
pints the online/offline state of all CPUs in the cluster.
The global_state file interface can still be used to know all
the state information about all CPUs. Replace "Active" with
"Isolated" string in this file output. Otherwise we see that
an offline CPU as active.
Change-Id: Id6ffb1c3ddfe85cf34ab670ceea8b53e5b2f60e2
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
We use pr_* to print info/warn/error messages. Use pr_fmt so
that the messages are prefixed by "core_ctl:" string.
Change-Id: Icc84db30ebf0855599431366d090a2b1c94da6d0
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
The online state is cached in CPU hotplug callbacks. This is
used in determining the potential candidates for isolation or
unisolation. The current code has checks in hotplug notifier
callback to correct the online state if it is out of sync
with the actualy CPU state. All this maintenance can be
avoided by directly using cpu_online().
This patch fixes an incorrect warning message emitted from
CPU_UP_CANCELED event callback.
Change-Id: I9681ce6c5bc701507427d60db397ca4966fbdb58
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Hardware characterization has shown that the CPU LDOs can
operate safely with less than 150 mV of headroom voltage.
Reduce the CPU LDO minimum allowed headroom voltage for both the
power cluster and the performance cluster in order to save power
by allowing LDO usage in more situations.
Change-Id: I72f28569d4719540fc84cf65d8783bbfec9435ad
CRs-Fixed: 989555
Signed-off-by: David Collins <collinsd@codeaurora.org>
Update the VDD_APCC CPR open-loop and closed-loop voltage
adjustments for MSM8996-Pro based upon characterization results.
This ensures stability and minimum power consumption. Reuse the
CPR revision 1 adjustment values for future CPR revisions (2 - 7)
for forward compatibility.
Change-Id: I42f7b696987102a38db324a1c515f1c6f01e6d45
CRs-Fixed: 989555
Signed-off-by: David Collins <collinsd@codeaurora.org>
As per msm8953 design, APM state machine could take more
than 200us for mode switching. Increase SW timeout constraint
value to 500us.
Also, update the mask used to check APM switch status.
CRs-Fixed: 992695
Change-Id: Ie0f8733fb5b39172fb14547292cbd05689a50230
Signed-off-by: Tirupathi Reddy <tirupath@codeaurora.org>
The commit 639277b6c73c ("regulator: spm-regulator: Add additional
settling delay for FTS2.5 SMPS") incorrectly changed the usage
of 'uV' variable. The local 'uV' variable holds the correct voltage
level to be scaled up/down. Use it instead of vreg->uV.
CRs-Fixed: 1036738
Change-Id: I52540237a4db79c149409c6017ffc750b5abddd2
Signed-off-by: Anirudh Ghayal <aghayal@codeaurora.org>
Based on characterization add 70us settling delay on the voltage UP
to account for warm-up time and ramp-up delays for 0-10% and 90-100%
of the voltage value.
On the voltage ramp-down side add the stepper slew-rate delay and
and an additional 70us margin to avoid voltage updates while the
stepper is in progress. This could lead to voltage over/undershoot
due to buck-internal synchronization failure.
CRs-Fixed: 1036738
Change-Id: Id4230be9c4c981758bbf6860bab1f487a3b57f85
Signed-off-by: Anirudh Ghayal <aghayal@codeaurora.org>
To accommodate speed bin 2 and fuse revisions associated with it
increase the fuse combo count from 16 to 24. At present, this is
needed for msm8996proAU SOC.
Change-Id: I8555162eab2c6bfead2e0762f28525e0aa56cf10
Signed-off-by: Subbaraman Narayanamurthy <subbaram@codeaurora.org>
To accommodate speed bin 2 and fuse revisions associated with it
increase the fuse combo count from 16 to 24. At present, this is
needed for msm8996proAU SOC.
Change-Id: Ie781ffda433d491a38695caa5e55ec1ec7ff9b4a
Signed-off-by: Subbaraman Narayanamurthy <subbaram@codeaurora.org>
Add panel physical width and height properties to nt35597 truly
video and command mode panel, which will be used to calculate
the DPI of display.
Change-Id: I733eb935f55d7979af4331d7ec4bb31d21e55816
Signed-off-by: Sandeep Panda <spanda@codeaurora.org>
Temp_adj_en instantaneously reduces the voltage and quotient
when temperature enters into corresponding band which may cause
an aggressive reduction in voltage. Avoid this by not configuring
temp_adj_en bit, this only reduces the quotient and allows CPR to
react to the reduced quotient. Use temp_adj_en bit only when both
temperature based and core count based adjustments are desired.
CRs-Fixed: 1051076
Change-Id: Ia42dbdd095e51bf9b9b7e865c104dcbe8f4219da
Signed-off-by: Tirupathi Reddy <tirupath@codeaurora.org>
Use scaled gate count (GCNT) for aging measurements to reduce
the noise effect on aging data. The amount of scaling depends
on the noise effect observed in characterization results. Use
a selected scaling factor value to derive the gate count (GCNT)
used for aging measurements.
CRs-Fixed: 1025832
Change-Id: I0eb3fb08d51ddca5cd4c08e26bc83eb9f66fd3ed
Signed-off-by: Tirupathi Reddy <tirupath@codeaurora.org>
Fix the kw issues in jpeg set driver data and in jpegdma modules.
CRs-Fixed: 2017767
Change-Id: I601245553b8148c9f257d55c42367b0b8612db92
Signed-off-by: Krupal Divvela <kdivvela@codeaurora.org>
Check the validity of pointer before dereference and initialize
uninitialized variables.
Change-Id: Ifad4dd9668db91a18b493e0c7276bf6c193fb680
Signed-off-by: Venu Yeshala <vyeshala@codeaurora.org>
video frame event notification is changed so that events will be raised
for all frames both predicted and non predicted frames and invalidate
pts and dts values for predicted frames. This change is specific to
IoT targets in which only non predicted frames will be decoded
and rendered for performance improvement.
CR's-Fixed: 2005724
Change-Id: I08f648a89eb043938fcc7d3d81e83c3599a2f160
Signed-off-by: Udaya Bhaskara Reddy Mallavarapu <udaym@codeaurora.org>
pwm_free() is called in the driver incorrectly when trying to disable PWM
channel while changing the configuration. pwm_free() checks PWMF_REQUESTED
flag before calling disable() hook. The driver doesn't call pwm_request()
but only calls of_pwm_get() at the probe to get the flag once. The Flag
will be cleared when pwm_free() is called for the first time and calling it
afterwards will simply bailout not calling disable().
Replace pwm_free() with pwm_disable(), this won't cause any issue because
the PWM channel is considered statically allocated to the led devices from
hardware perspective and no need to free it in the driver.
While at it, free the PWM device if any errors when getting PWM
configuration.
CRs-Fixed: 2014600
Change-Id: I373d2a1ac83232ce94933c287b4ebe3f23519c83
Signed-off-by: Fenglin Wu <fenglinw@codeaurora.org>
Fix the order of parameters passed into read/write APIs to make
it more readable. While at it, add a pr_fmt() to the driver to
get better log information. Also, change all dev_* APIs to pr_*
APIs for better logging.
Change-Id: Ib19f48ddbaeeb3af483ac472dabb44558a6b9e43
Signed-off-by: Subbaraman Narayanamurthy <subbaram@codeaurora.org>
Currently MISC_TRIM_ERROR_RC19P2_CLK register is hardcoded to
read the trim error in 19.2 MHz RC clock. However, this can be
located differently in some PMICs. Add support to read it from
misc device as this register cannot be directly accessed from
haptics driver.
Specify the MISC peripheral and the clock trim error register in
haptics for pmi8998 and pm660 so that haptics initial drive code
can be adjusted based on error code properly.
Change-Id: I04d8dbac0f943c65272531647c54d3c1495331e7
Signed-off-by: Subbaraman Narayanamurthy <subbaram@codeaurora.org>
clock_debug_mux is required for the 'measure' functionality of the clock,
register using the clk_register_debug for the same.
Change-Id: I95b64ddd3ef3d869cf3b8ad2b4210ea78ba8f5fa
Signed-off-by: Taniya Das <tdas@codeaurora.org>
Common Clock Framework(CCF) already has a clock ops hook `debug_init` which
could be used for the debugfs clock measure functionality. Remove the APIs
from the CCF and update the 'debug_init' clock ops to the clock types which
require the measure functionality.
Change-Id: I0c01f72a9d1d1caa1b1ab73a800401c2cbc3216c
Signed-off-by: Taniya Das <tdas@codeaurora.org>
The 'measure' functionality is a debug feature which allows the user of
clocks to allow measuring the frequency of a given clock using the ring
oscillator.
Move 'measure' code to a new clk-debug file to support the functionality.
Change-Id: I229721f17d232a4ff69b5cf416b43d22fee5b72e
Signed-off-by: Taniya Das <tdas@codeaurora.org>
Update the diag header with latest log, message
and event information to match the list maintained
by the peripherals.
CRs-Fixed: 2020864
Change-Id: Icbed01bb4f90fb7d72fe7517ee6964d799f1d48e
Signed-off-by: Chris Lew <clew@codeaurora.org>
