Age | Commit message (Collapse) | Author |
|
of_device_id tables need to be NULL terminated. This patch fixes the
missing NULL termination within sysedp_batmon_calc module.
Bug 1539761
Change-Id: I7b1f48b85fce84337d8d2093fe7886f965d0b28f
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/437001
(cherry picked from commit 305fceeceed23782a1170ed60b87059010a83c61)
Reviewed-on: http://git-master/r/671006
Reviewed-by: Steve Rogers <srogers@nvidia.com>
|
|
Provide sysedp_dynamic_capping with the instantaneous GPU frequency
when notifying it of the GPU load. Modify the gpu/cpu priority
decision logic to choose CPU priority until GPU frequency gets "near"
the CPU-priority-limited-GPU-fmax. Introduce the priority_bias debugfs
parameter to facilitate tuning of "near". priority_bias takes a value
from 0 to 100.
Change-Id: I57df17d50cd8077a512b5932f4a304ca5e6992aa
Signed-off-by: Matt Longnecker <mlongnecker@nvidia.com>
Reviewed-on: http://git-master/r/481720
(cherry picked from commit b2ac745a45e273e849d7b190913ee97092fdebc2)
Reviewed-on: http://git-master/r/498901
GVS: Gerrit_Virtual_Submit
Reviewed-by: Winnie Hsu <whsu@nvidia.com>
|
|
Battery temperatures are reported in 0.1C units, fix the debugfs
prints to not have inconsistent unit.
Bug 200006249
Change-Id: I98ee47d57e8a14750f7746fe15c8db7268b480f0
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/415002
(cherry picked from commit 2c1c13b8941761f630cf34f9d8d5ca5d54b90fc0)
Reviewed-on: http://git-master/r/417624
Reviewed-by: Mandar Padmawar <mpadmawar@nvidia.com>
Tested-by: Mandar Padmawar <mpadmawar@nvidia.com>
|
|
The GPU should be buildable as a module, however, some tegra specific
symbols are not exported properly.
Bug 1476801
Change-Id: Ifa44a53f4b7ecea9c3405e3fb16dc262ba695c59
Signed-off-by: Arto Merilainen <amerilainen@nvidia.com>
Reviewed-on: http://git-master/r/387605
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
|
|
This patch adds a sysfs entry /sys/power/sysedp/batmon/esr. This
allows an user space process to update the current ESR value seen by
the sysedp_batmon_calc component.
sysedp_batmon_calc will compute the ratio between internal table based
ESR estimate and value set from user space. This ratio will be then
used to scale all future table based ESR estimations.
Bug 1400509
Change-Id: Ia409826f793afef4e4c8d1e42010ff31e529f827
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/337378
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-by: Steve Rogers <srogers@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Terje Bergstrom <tbergstrom@nvidia.com>
|
|
Bug 1492962
Change-Id: I753037da2a2f004133ac28986c1722b05e754150
Signed-off-by: Steve Rogers <srogers@nvidia.com>
(cherry-picked from commit c01a6dfd551eb43cc65dbeacb9a86cc3ff913026)
Reviewed-on: http://git-master/r/391475
Reviewed-on: http://git-master/r/392059
Reviewed-by: Timo Alho <talho@nvidia.com>
|
|
Bug 1486148
Change-Id: I8bf6af83530273ad5b04f1903f2d76048ff2b514
Signed-off-by: Steve Rogers <srogers@nvidia.com>
Reviewed-on: http://git-master/r/386462
(cherry-picked from commmit 2583d5bbd2910bf4420329786bb3d7560bab1615)
Reviewed-on: http://git-master/r/387831
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Timo Alho <talho@nvidia.com>
|
|
- Remove EDP_FRAMEWORK configuration
- Include drivers/edp/Kconfig from drivers/Kconfig
Change-Id: I925d933830fd85ae58320bf979d82a7df9a87f53
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/380410
Reviewed-by: Steve Rogers <srogers@nvidia.com>
Reviewed-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Bug 1475120
Change-Id: I21be3f45e624412626b6f01f2ad04307f2911a8c
Signed-off-by: Steve Rogers <srogers@nvidia.com>
Reviewed-on: http://git-master/r/378467
Reviewed-by: Timo Alho <talho@nvidia.com>
Reviewed-by: Automatic_Commit_Validation_User
|
|
Bug 1465408
Also fix the debugfs output for cap_method.
Change-Id: I1cce87ae08e81eee5b0b42dac8601764c5fad55b
Signed-off-by: Steve Rogers <srogers@nvidia.com>
Reviewed-on: http://git-master/r/370035
(cherry-picked from commmit 51b58d692653dd6075c3e728dd7f133968c99a5a)
Reviewed-on: http://git-master/r/373630
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Timo Alho <talho@nvidia.com>
|
|
Bug 1453350
Change-Id: Ia4064cf59a93f90770a61144fcdb65918cdaa1b7
Signed-off-by: Steve Rogers <srogers@nvidia.com>
Reviewed-on: http://git-master/r/363494
(cherry picked from commmit 88b6025796e4b57e750b7c615b9416c433ef7761)
Reviewed-on: http://git-master/r/364963
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Negative temperatures where incorrectly handled when calculating
maximum battery current. This patch fixes the issue.
Change-Id: I446c2b9b6cf5586fbba11a4fe0906a38a7c654d1
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/359854
(cherry picked from commit f63a5fa0d51a04befc2e113df91c076e0cab8e83)
Reviewed-on: http://git-master/r/361476
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Change the compatible field of sysedp_batmon_calc to match with nvidia
naming convention.
Bug 1391872
Change-Id: I599199813092879daaa0d533831c35b73d693f5c
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/353726
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
This patch changes the calculation of available AP+DRAM budget. It
will now take into account the maximum power reduction that HW
throttling can accomplish (using Over Current (OC) signalling). The
calculation of AP+DRAM is now as follows:
P_AP_DRAM = P_BAT - SUM(P_MAX) + min(SUM(P_MAX - P_THRESH), P_THROT)
where
P_AP_DRAM is power available for AP+DRAM
P_BAT is total power available from battery
SUM(P_MAX) is sum of maximum peak power of consumers (OC=1)
SUM(P_THRESH) is sum of maximum peak power of consumers (OC=0)
P_THROT is the maximum amount of power that HW thorttling can reduce.
Change-Id: Ife02bba22694c93ca13e42b4acdf84ac2e6a96ba
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/348943
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Add debugfs entries to sysedp_batmon_calc component.
* rbat - to show battery impedance table
* ibat - to show max. battery current table
* ocv - to show battery ocv table
* vsys_min - to show min. VSYS vlotage
* r_const - to show board impedance
Change-Id: I8a59450bdb080304630c88bbf140ad91f50f46d5
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/350657
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
This patch adds a sysedp_reactive_capping component. This component
monitors the rate of over current (OC) interrupts from soc_therm
hardware. During frequent events, sysedp_reactive_capping increases
it's power state to reduce the budget available for AP+DRAM.
sysedp_reactive_capping hooks into interrupts provided by soc_therm.
Towards system EDP framework, sysedp_reactive_capping acts as sysedp
consumer.
Change-Id: I53918c7cf63cbfd689b78abd17a77d97f5ed985a
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/348416
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
Tested-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Bug 1391872
Change-Id: I31bbe2b2fe18c8fd46b8d70980639511e21c9dd2
Signed-off-by: Kerwin Wan <kerwinw@nvidia.com>
Reviewed-on: http://git-master/r/338795
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Timo Alho <talho@nvidia.com>
Tested-by: Timo Alho <talho@nvidia.com>
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Revised system edp software framework has been put in place. This
patch removes old framework code under drivers/edp.
Bug 1431977
Change-Id: Icde6ad7066e15172863a39388904309ce2c75c39
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/350414
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
This patch creates a 'consumers' kset under sysedp directory where all
sysedp consumers are to be placed. The purpose is to allow uevents to
be sent when consumers are created so that their file permissions can
be managed from userspace.
Bug 1416849
Change-Id: I93a28e50335378eafcdecd103ff39e2bfa5d9ff8
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/346522
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
Tested-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Last character of incoming string was mistakenly stripped. This patch
fixes the issue.
Bug 1416849
Change-Id: I8741504bb5887ad54b93a5066a99bfecc55cff9f
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/345196
Reviewed-by: Automatic_Commit_Validation_User
Tested-by: Greg Heinrich <gheinrich@nvidia.com>
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
ocpeaks table defines the maximum peak power for each state when
hardware overcurrent signal is asserted. The most prominent user for
overcurrent signaling is the modem consumer.
To register sysedp consumer for user space with overcurrent peak
information, following can be written to consumer_register sysfs
entry:
consumer_name a0, a1, a2, ..., an; b0, b1, b2, ..., bn
where ai and bi are nominal and overcurrent peak values for power
state i, respectively
Bug 1416849
Change-Id: I196aa976f8b2b0f300599db6807b94bf1a486f44
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/338237
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
Tested-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
This patchs adds a check to sysedp_consumer_register to ensure that no
consumer with same name has been registered before.
Change-Id: I028b2edcef926b280d4a57037d3406d1ae83547a
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/338236
Reviewed-by: Neil Patel <neilp@nvidia.com>
GVS: Gerrit_Virtual_Submit
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Previously, only state of charge was considered when determining
battery impedance. In practice, impedance is dependant on temperature
and state of charge.
This patch adds support for two dimensional battery impedance
(rbat_lut) tables. The rbat value is bi-linearly interpolated from
table as a function of temperature and state of charge.
bug 1412781
Change-Id: I8656aa13f0ebebb848f536194b37420a0438d2ac
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/335070
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Add sysedp_get_state API function call to query current power state of
sysedp consumer.
Change-Id: I667a848e269836b7dba652a7929a5fbce6ef8d82
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/304819
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Change-Id: I7f28e373b2f96d84af98ac64d4d36c83ad18d4a6
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/309701
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
GK20A driver does initializes clocks until very late init –
clk_set_rate() is returning error until then. This patch adds graceful
handling of missing gpu clock.
Change-Id: Ia0825985425494675637c20f3d25bc77b24afb91
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/304863
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
This patch introduces a revised system-EDP software
framework. Framework consist of following components:
* sysedp - central component handling of the book keeping of consumer
power consumptions
* sysedp_batmon - periodically monitors the state of battery and
updates the available budget (in mW) to sysedp framework
* sysedp_dynamic_capping - limits the CPU, GPU, and EMC frequencies
to ensure that system will operate in the given budget.
* sysedp_consumer - A device in the platform that has noticeable peak
power consumption is called sysedp consumer. Consumers register
themselves to the sysedp framework and inform sysedp when there is
a change in their power state
Change-Id: I343d8f09082216744da41abe5e749b15cb20417a
Signed-off-by: Timo Alho <talho@nvidia.com>
Reviewed-on: http://git-master/r/304006
GVS: Gerrit_Virtual_Submit
Reviewed-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
flush_delayed_work should be used rather then
flush_delay_work_sync.
flush_delayed_work_sync just calls flush_delayed_work.
see workqueue.h
/nvidia/DEV_KERNEL/kernel/drivers/edp/psy_depletion.c
In function 'capacity_set':
nvidia/DEV_KERNEL/kernel/drivers/edp/psy_depletion.c:280:2:
warning: 'flush_delayed_work_sync' is deprecated
(declared at
/nvidia/DEV_KERNEL/kernel/include/linux/workqueue.h:454)
[-Wdeprecated-declarations]
/nvidia/DEV_KERNEL/kernel/drivers/edp/psy_depletion.c:
In function 'vsysmin_set'
/nvidia/DEV_KERNEL/kernel/drivers/edp/psy_depletion.c:297:2:
warning: 'flush_delayed_work_sync' is deprecated
(declared at
/nvidia/DEV_KERNEL/kernel/include/linux/workqueue.h:454)
[-Wdeprecated-declarations]
Change-Id: Ib6ee73feb40560aa5eb9b8941047e5313ac3aaa9
Signed-off-by: Philip Rakity <prakity@nvidia.com>
Reviewed-on: http://git-master/r/278887
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Nitin Kumbhar <nkumbhar@nvidia.com>
(cherry picked from commit 1ad8994d893cf805e84988a4f1077250f90957bb)
Signed-off-by: Ajay Nandakumar <anandakumarm@nvidia.com>
|
|
Adding a new debugfs attr for vsys_min psy depletion platform data for
tuning.
Bug 1306298
Change-Id: Id9b7a6affae2216e7bafde55eea2f8eacbe24572
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/272489
(cherry picked from commit 9361d0eaef4f7616e443fdbbefebeda92c496b87)
Reviewed-on: http://git-master/r/275628
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Add an emulator mode to the psy depletion client so that we can emulate
different depletion states. This mode requires the board to specify a
valid { capacity, OCV } lut.
Bug 1293353
Change-Id: Ie4f83ee70cec3575ee42a8edea4524d0a33da0fb
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/255648
(cherry picked from commit db1d75dc8dfaec9d0c906413d9b1eaf1df9c1e64)
Reviewed-on: http://git-master/r/256830
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Bug 1300607
Change-Id: Iafaab450fd88a41b6384256ad2c6aaf2ebb9f360
Reviewed-on: http://git-master/r/239500
(cherry picked from commit fb715d4a3cb28547c274c648ecd43c87b33a1ec2)
Signed-off-by: Diwakar Tundlam <dtundlam@nvidia.com>
Reviewed-on: http://git-master/r/253745
Reviewed-by: Automatic_Commit_Validation_User
|
|
cancel_work_sync is called with mutex held. This can cause deadlock if
there is a pending work while switching governors or while unregistering
managers (because the worker function tries to take the same lock).
Fix this by:
- getting rid of cancel_work_sync during governor switching (if there is
no governor, the scheduled work will do nothing)
- call cancel_work_sync after the manager unregistration is complete and
outside the locked code.
Bug 1313928
Change-Id: I64b270898bffc9b961cb70ca9c181c1c6e6c0118
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/247648
(cherry picked from commit 981bf6768f2ed99127550ba5554e433716586eaa)
Reviewed-on: http://git-master/r/250634
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
Fixing warnings issued by EDP governors due to incorrect power budget
math.
Bug 1313928
Change-Id: If1c9cf5c6df66ed4f2cc8a1b9399d4c57b2c7a64
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/245106
(cherry picked from commit 6870474b775573f8a060afff99fca3bb171ec12c)
Reviewed-on: http://git-master/r/250633
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
The dummy client created to reduce the EDP cap is named '.'. This
results in a runtime warning because the sysfs entries can not be
created with this name. Fixing this by changing the name to '.debug'
Change-Id: I90ac60503053ce5de8d3ea1b2d1a7627d1ea33ea
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/245108
(cherry picked from commit f4e922af79136d23991d86c17c34101a261912db)
Reviewed-on: http://git-master/r/250629
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
Add a new debugfs node for printing out the current AP+DRAM EDP
capping decisions.
Bug 1257143
Change-Id: Ie42c78233e3462ea786456ffb9e321895a437c9f
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/238035
(cherry picked from commit 7741dd6b9dd6aaeb89f5a0268fb343bcca500228)
Reviewed-on: http://git-master/r/250625
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
Moving more stuff to common code.
Bug 1257143
Change-Id: Ied0a6ecdaa0dcca53547ba8df6a20de38298e41d
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/238033
(cherry picked from commit 2c8df5eb7508b601f61ea0daef5df6eb9a63a86e)
Reviewed-on: http://git-master/r/250623
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
Let the board file specify the modem sys EDP client rather than assuming
a fixed name.
Bug 1257143
Change-Id: I2257730428ff7ddc676406f848dee737d84b4fe1
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/238032
(cherry picked from commit e5ed8f7249f357b76cc2cc628efecc031122805d)
Reviewed-on: http://git-master/r/250622
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
The loan update notifier is not called in few situations - fix those.
Bug 1257143
Change-Id: I81a6ad18c2ffab9d1c7b2ea1d38f23e5a29156c8
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/238030
(cherry picked from commit 4fbde3eaaa42b758c9b274bf8970176e0a695e1a)
Reviewed-on: http://git-master/r/250620
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
Relocating this function for increased visibility within the file.
Bug 1257143
Change-Id: I5ab244f57b75636a2273346d1806d24b66a3ab9c
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/238029
(cherry picked from commit 609693e5575d5e7ae52d0c7faecf48fd6ed1c36e)
Reviewed-on: http://git-master/r/250619
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Bharat Nihalani <bnihalani@nvidia.com>
|
|
Adding debugfs attributes for getting AP+DRAM EDP table and CPU power
limits.
Change-Id: I36f7d447a964f34a3ab6d803d753537baca1e235
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/236268
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
Reviewed-by: Diwakar Tundlam <dtundlam@nvidia.com>
|
|
The core client E-states are scaled down proportional to the gain
factor. This will reduce the difference between an available budget and
AP+DRAM EDP table entries.
Bug 1266730
Change-Id: Ie68d3a71e320433b2096e3a5d40c8bb1185bae4f
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/234947
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
For boards that do not specify an rbat lut, use the r_const as the
RBAT.
Change-Id: Ica8045c65fed34b48f2b89a5be4bc1435320924f
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/233805
(cherry picked from commit 43e96b352cdfb1438edc326da0ac4ae05e14c03e)
Reviewed-on: http://git-master/r/234945
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Isolated gpu load peaks degrade CPU performance without benefitting the
GPU use case.
This patch implements a filter to detect such events by requiring the
GPU load to be above gpu_high_threshold for at least gpu_high_count
(default is 2) number of consecutive notifications.
Bug 1293353
Change-Id: Ia14aa1d121c665251b6299cbe8b78a81f1f4c8d3
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/233597
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
While calculating the CPU freq cap for a given AP+DRAM power budget (the
allowed E-state level), performance is lost in two places.
(1) after applying the gain factor: since the E-state level
multiplied by the gain factor does not exactly match a row
in the AP+DRAM table (resulting in the selection of a lower
value).
(2) while looking up in the cpu power limit table: some delta mW
is lost as in the above case (since there is no exact match, we
are choosing a lower level)
(1) is fixed by adding the delta amount to cpu power
(2) is fixed by choosing a ceiling value (this is OK since in almost all
cases, we have some battery cap remaining).
Bug 1293353
Change-Id: I6584bfa385d8af15457d8c3faaa0d4a3169c4624
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/233596
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
core_gain is made board specific. This patch makes the needed changes in
the tegra core sys EDP client code.
Bug 1266730
Change-Id: Icc3f77d245579e8927260cb3ec0475d8b765ba62
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/233591
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
|
|
Not all fuel gauges provide OCV reading. This patch adds a lookup-table
based OCV calculation (choosen only if the board data provides a SOC vs
OCV lut).
Bug 1274401
Change-Id: I423b02a52464546f325a4828057aeb18aa50ae53
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/231561
Reviewed-by: Automatic_Commit_Validation_User
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
T148 AP+DRAM EDP tables require GPU and EMC freqs to be specified in
kHz.
Pluto and Macallan already specifies the freqs in mHz - so this patch
contains board file changes too (since these have to go as a single
change).
Bug 1266730
Change-Id: I14f96460865f775722afd217eea3401d4f0c6f18
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/231071
Reviewed-by: Mrutyunjay Sawant <msawant@nvidia.com>
Tested-by: Mrutyunjay Sawant <msawant@nvidia.com>
|
|
Power-supply depletion driver.
Bug 1274376
Change-Id: I1e18db9bb06a18bb1cc6446fe36ce8422bae67cf
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/226227
(cherry picked from commit d58d93daa615663ef9fe72c58357ec103a84803a)
Reviewed-on: http://git-master/r/226667
Reviewed-by: Automatic_Commit_Validation_User
GVS: Gerrit_Virtual_Submit
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|
|
Renaming files to align with next set of changes
Bug 1274376
Change-Id: I3f8984f29165fcdd245b9ee8fb7499cc876b3fd6
Signed-off-by: Sivaram Nair <sivaramn@nvidia.com>
Reviewed-on: http://git-master/r/226223
(cherry picked from commit a4d3a78af74e2a92e29d75e0b300ddbd64f66cbf)
Reviewed-on: http://git-master/r/226663
GVS: Gerrit_Virtual_Submit
Reviewed-by: Juha Tukkinen <jtukkinen@nvidia.com>
|