There are only two reasons for us to define a OPP table these days, and
one of them is to define the voltage levels that we send to the GMU.
That is silly because we have a perfectly good power level definition
already handy. Move the levels to the GPU definition and use them
accordingly.
Change-Id: Ic0dedbad2d3c84a2de7cda5311ff99e96434b9e5
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Life goes on, and so does the kernel. Update some API changes since
msm-4.19 including helper functions for vmfault functions, renaming
a memory statistic, updating cmd-db and removing __mutex_user which
is not longer available.
Change-Id: Ic0dedbad0b24042f3683a4be99888a0aa86baf78
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Add a copy of the Adreno GPU bandwidth governor from msm-4.19 as of
01dafb76 ("Merge "power: qpnp-qnovo5: Add standalone class param"")
Like the GPU frequency governor, move it to the KGSL driver directory
to share local headers.
Change-Id: Ic0dedbadbdeef532a3e847db759b9840282064da
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Add a copy of the Adreno GPU frequency governor as of
01dafb76 ("Merge "power: qpnp-qnovo5: Add standalone class param"").
Instead of keeping the governor in drivers/devfreq, move it to the GPU
driver directory so that it can share a header file with the GPU driver.
Change-Id: Ic0dedbada21ca0310ce1ce69126fa0096c70c2e6
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
There was a time, long long ago, when having a "zero" power level made
sense because a "zero" powerlevel wasn't really "zero". That is no longer
the case and it is silly to force the DT to define it for EVERY. SINGLE.
TARGET. Get rid of it and assume that zero is, well... zero.
Change-Id: Ic0dedbadc5551585de893cdb5b3ed948a62aa430
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
The system power management functions (suspend/resume) are currently
registered with the adreno device but implemented at the KGSL level
which requires function hooks for the actual adreno functionality.
Move the functions back to adreno where they belong, get rid of the
function hooks and remove the legacy suspend/resume handlers that
we haven't needed in years.
Change-Id: Ic0dedbad4a09f165ffd9707e66c45e5b516962e3
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Commit 7a6368ee0b ("msm: kgsl: Add support for the Adreno GPU") stopped
exporting the uapi header. Unfortunately not all the user-mode libraries
got the message. Start exporting it again to prevent breakage while
everybody figures out how to transition correctly.
Change-Id: Ic0dedbadc187d018b9ff34c0a93f156d525e8b16
Signed-off-by: Jordan Crouse <jcrouse@codeaurora.org>
Add support in SPI QUPv3 driver to make it a DLKM.
Change-Id: Ic251466331862eb9a6e72455cac8ac4388b01045
Signed-off-by: Vipin Deep Kaur <vkaur@codeaurora.org>
Fix QCA_WLAN_VENDOR_ATTR_FREQ attribute ID to match the official
value from the QCA vendor command definitions (qca-vendor.h).
Change-Id: I549752fe5c29de5346cf9263ccc16266cafbbecd
Signed-off-by: Lior David <liord@codeaurora.org>
Signed-off-by: Maya Erez <merez@codeaurora.org>
Signed-off-by: Alexei Avshalom Lazar <ailizaro@codeaurora.org>
Prepare driver for FTM/AOA while unassociated.
1. Added attributes allowing user space to specify the frequency
where FTM/AOA peer is listening. When specified it will override
the entry in the kernel scan results cache. If not specified,
the kernel scan results cache will be used like today.
This gives user space location framework more flexibility - it
can maintain its own cache of peers, or perform scans less
often (since entries in the kernel scan results cache expire
quickly)
2. Remove associated check when starting FTM session, so session
can start even when unassociated.
Change-Id: Ib0ba52c506e1d8d59ab8c12043acd792405d69a0
CRs-Fixed: 1094147
Signed-off-by: Lior David <liord@codeaurora.org>
Signed-off-by: Maya Erez <merez@codeaurora.org>
Signed-off-by: Alexei Avshalom Lazar <ailizaro@codeaurora.org>
Added initial support for fine timing measurement(FTM)
and angle of arrival (AOA). These features allow a station
to get an accurate indoor location by measuring
distance (FTM) and azimuth/elevation (AOA) to one or more
stations/APs with a known location.
The following APIs are supported:
1. Get FTM/AOA capabilities, currently hard-coded for a
single-shot measurement to an associated AP.
2. Start FTM session.
3. Receive FTM results.
4. Start standalone AOA measurement.
5. Receive AOA results.
Change-Id: I54ebc1bdff6ed7d91644165b16eb0a4240069321
CRs-Fixed: 1057307
Signed-off-by: Lior David <liord@codeaurora.org>
[merez@codeaurora.org: fix compilation errors]
Signed-off-by: Maya Erez <merez@codeaurora.org>
[ailizaro@codeaurora.org: trivial conflict fixes]
Signed-off-by: Alexei Avshalom Lazar <ailizaro@codeaurora.org>
[liord@codeaurora.org: fix conflicts, SPDX license]
Signed-off-by: Lior David <liord@codeaurora.org>
Signed-off-by: Alexei Avshalom Lazar <ailizaro@codeaurora.org>
The function qcom_scm_get_sec_dump_state dereferences __scm without
checking for NULL and could potentially lead to a NULL pointer exception.
This scenario can happen if the clinet calls qcom_scm_get_sec_dump_state
before qcom_scm driver is even probed. Hence, check for NULL before
dereferencing it.
Change-Id: Ic6f7c2b2d616645ea69a8c594c90dc01d689637f
Signed-off-by: Raghavendra Rao Ananta <rananta@codeaurora.org>
Tunables update accesses the monitor's registers for some devices.
When the device is suspended, the clocks for those devices are turned
off and updating the tunables of a suspended device causes the device
to crash. Fix this by checking the flag dev_suspended before allowing
the tunable updates.
Change-Id: I3a0938edf761a9ea6d39ffeec19f35c228d5c306
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: updated to utilize suspend_count from upstream and resolved minor merge conflicts]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
There is a possibility to switch the governors from sysfs even when the
device is in suspended state. This can cause a NOC error at times when
trying to access the device's monitor registers in a suspended state. This
change fixes this issue. Utilize suspend_count to know if the device is in
suspended state or not. Check if the device is suspended before switching
the governor from sysfs.
Change-Id: I15055aa51daa35272be4667e5bafb8ccd7933098
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: updated to utilize suspend_count from upstream, resolved minor merge conflicts and updated commit text]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
There is a race condition when the event governor_store is being executed
from sysfs and the device issues a suspend. The devfreq data structures
would become stale when the suspend tries to access them in the middle
of the governor_store operation. Fix this issue by taking a lock around
suspend and resume operations so that these operations are not concurrent
with the other events from sysfs.
Change-Id: Ifa0e93915a920cec3e0429966328a1128d61098b
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: updated to merge better with upstream changes and removed renaming since done in previous commit]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Currently, concurrent writes to sysfs entries leave the possibility
for race conditions within the devfreq framework. For example,
concurrently executing max_freq_store and governor_store can result
in attempting to perform an update_devfreq() before the new governor's
start handler can be executed.
A more concrete case is a race between polling_interval_store and
governor_store. Because no lock is used after calling into the event
handler of the old governor and there's nothing preventing work from
being queued after the monitor is stopped, it's possible to
accidentally cause delayed work to be queued on the governor being
switched to. This can be seen if you create two threads, one which
changes a device's governor between simple_ondemand and performance,
and one which changes its polling interval between 45 and 50.
All of these races can be addressed with the introduction of a lock
that prevents sysfs operations from interleaving in this fashion.
Change-Id: Ia6887dcb2d69dc2576837a6c09fed55a28943abc
Signed-off-by: Jonathan Avila <avilaj@codeaurora.org>
[avajid@codeaurora.org: renamed to event lock and only used when CONFIG_QCOM_DEVFREQ_ICC is enabled]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
This reverts commit 4585fbcb53. The change
to be reverted would add probe order-dependent numbers to the end of each
devfreq device, making our post boot script (among others) incredibly
brittle.
Change-Id: I8fdb679ae651d9f425ffd60e2141613ae20bb029
Signed-off-by: Jonathan Avila <avilaj@codeaurora.org>
[avajid@codeaurora.org: resolved trival merge conflicts]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Having OPP levels of 0 and 1 causes a minor error in OPP
debugfs since an OPP level of 0 causes the OPP index to be
used instead (1 in this case) which later clashes with an
actual OPP of 1. Update the qoslat OPP levels to avoid this.
Change-Id: I23c0e736c9f1bfe77a99af44ced180d488a52eeb
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Currently bw_hwmon governor does not allow the device
to be resume if the resume_freq is set to zero. But there
can be devices that vote for zero frequency when there is
no bandwidth on the device. This does not mean that the device
should not be resume from suspend. Remove the check with prevents
resume, instead add a check in the devfreq framework to check if
device is suspended or not before calling the governor resume.
Change-Id: Ib82a6a36308aee52e8bb989fddca92265de6c4a4
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: removed suspended check in devfreq framework]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
The IO percentage limits the values to 100, increase it
since some devices might need a higher value.
Change-Id: I7f19017b426832d9a1a11bd02195a4d15634c041
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: resolved trivial merge conflict]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Kona has new LLCC PMU version. This brought in changes in the
monitor enable, disable and configuration registers and hence
made code changes accordingly to read cache misses for multiple
CPUs.
Change-Id: I4d25487966a31f12143e2f99264a25d8c38f0188
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: updated version enum, fixed disable/clear masks and made minor styling changes]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Some chips have hardware that can count misses for LLCC at a per-CPU level.
This PMU serves as an intermediary that allows us to retrieve these values
for use in other drivers.
Change-Id: I1dc3090a64ec7d5b12a36b0395c42930128287fe
Signed-off-by: Jonathan Avila <avilaj@codeaurora.org>
[avajid@codeaurora.org: removed BEAC registers, unused variables and made minor styling changes]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Add a devfreq device that has the ability to vote for a
QoS level to ensure better memory latency.
Change-Id: I33094b45ec250a58f244ecb1b7b8476df8791425
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Currently, we calculate freq and stall_pct multiple times for
the same shared PMU counters. Calculate these once per cpu_grp
to prevent repeating the same calculation multiple times later.
Change-Id: I3b594e01ca50a583f45174e69a24f0b0eca32dde
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
[avajid@codeaurora.org: made minor styling changes]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Currently, each individual arm-memlat-mon instance separately reads from
a few shared PMUs, which leads to unnecessary IPI overhead. Tweak the
arm-memlat-mon design to allow multiple monitors to share certain event
counts, lowering overhead.
Change-Id: Ic3358afd1853efd84a4c3d5f3d66c3a86df6e5e7
Signed-off-by: Jonathan Avila <avilaj@codeaurora.org>
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
[avajid@codeaurora.org: squashed minor fixes and made minor styling changes]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Some targets support different DDR types. Detect the DDR type
and populate the frequency table accordingly. OPP framework
supports opp-supported-hw bit map which allows the selected
frequencies to be added to the opp table based on the hardware
version check. Implement the same in the bandwidth monitor device
so that the right frequencies would be added to the opp table.
This patch also adds checks for the latency based device to detect
the DDR type at runtime and add the corresponding frequency map.
Change-Id: Ice5a0b14da67b3f2f07e98bc7349220da7d4efdb
Signed-off-by: Santosh Mardi <gsantosh@codeaurora.org>
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: resolved minor conflicts and made minor styling changes]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
In bytes_to_mbps function, the parameter is all unsigned,
so change the decleration of the function to include unsigned long
long to avoid compilation errors in 32 bit environment.
Also changed the return value as unsigned long to avoid
any data loss possible in 64 bit environment.
Change-Id: I90aebfe32e86ebc17c414d6f3c0c3cbd6dddb0bd
Signed-off-by: Santosh Mardi <gsantosh@codeaurora.org>
Some BWMON devices requires clocks to be enabled, hence
add necessary support in the bwmon driver to enable the
required clocks if any.
Change-Id: Ie06731f764ff24d602ae2a86dea4f39ce75df800
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: made minor styling changes]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
If there is a failure in starting hw monitor required for
memlat governor, return the error code to devfreq framework
so that the framework will make sure the governor is
switched back to previous governor.
Change-Id: I76705725fc8b94d5341b1b8af7c1bd2b79e7642a
Signed-off-by: Santosh Mardi <gsantosh@codeaurora.org>
[avajid@codeaurora.org: minor change to check ret < 0]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
The BWMON governor start is returning success on GOV_START event
without checking for the return value of start_monitor.
The return value of start_monitor is not being returned to
ret variable. This would cause the governor to start successfully
even when the monitor failed to start causing a NULL pointer
derefence when accessing the device attributes. Fix it by checking
the return value of start_monitor.
Change-Id: I8c1f6933d44ae4533c6b81ccda8a5c4c0da3779c
Signed-off-by: Rama Aparna Mallavarapu <aparnam@codeaurora.org>
[avajid@codeaurora.org: minor change to check ret < 0]
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
