Currently, the frequency is calculated by max freq * 1.25 * util / max cap.
Add a vendor hook to adjust the frequency when the calculation
overestimate.
android_vh_map_util_freq
adjust util to freq calculation
Bug: 177854431
Signed-off-by: Yun Hsiang <yun.hsiang@mediatek.com>
Change-Id: I9aa9079f00af7d3380b19f2fe21b75cddd107d15
(cherry picked from commit 3122e3ec9672036384304fdeaa1b1815f60ba817)
Vendor scheduler may not want to rebalance between domains in some case.
New vendor hook is added for vendor to decide whether rebalance domains.
Bug: 176952463
Change-Id: Ie2edbd4b11679786096dd0170496904ae180583c
Signed-off-by: Park Bumgyu <bumgyu.park@samsung.com>
Vendor hooks required explicitly defining macros or inline functions
to handle the non-GKI build case (!CONFIG_ANDROID_VENDOR_HOOKS). Added
support for generating them automatically so the macros are no longer
required.
Both models are now supported so we can transition.
Bug: 177416721
Signed-off-by: Todd Kjos <tkjos@google.com>
Change-Id: I01acc389d315a5d509b0c48116854342a42e1058
Fixes use of %p format for u64 (should be %llx)
Fixes: e091aa59b9 ("ANDROID: tracing: Add register read and write
tracing support")
Signed-off-by: Todd Kjos <tkjos@google.com>
Change-Id: I4cd3a179e08a3fb682db6d9bb8530d504eb71720
CFI has additional overhead on indirect branches to modules as the
target is not known at kernel compile-time. This has been demonstrated
to cause problematic performance regressions on benchmarks using GKI
together with modularized scheduler callbacks attached to restricted
vendor hooks.
To restore some of the performance back, let's disable CFI around the
restricted hook call sites and issue a raw indirect call in fast paths.
We should be able to drop this patch when/if the arm64 static_call
port lands upstream [1] as this would make tracepoints circumvent some
of the CFI checks using text patching, but that still remain to be
proven.
[1] https://lore.kernel.org/linux-arm-kernel/20201028184114.6834-1-ardb@kernel.org/
Bug: 168521642
Change-Id: I7cd59f582b12fed15be64059f08122f96786e650
Signed-off-by: Quentin Perret <qperret@google.com>
In order to debug critical domain and device power issues, it may be
necessary to disallow certain idle states at runtime. Enable a vendor
hook to check if a domain idle state is allowed for powering down the
domain.
Bug: 175718935
Signed-off-by: Lina Iyer <ilina@codeaurora.org>
Change-Id: I403c0c7d272439fb338bbf54a346861559385047
Fix an issue for non-GKI builds found by the 0-day test bot:
kernel/sched/fair.c: In function 'select_task_rq_fair':
>> kernel/sched/fair.c:6766:6: error: implicit declaration of function
'trace_android_rvh_select_task_rq_fair_enabled';
did you mean 'trace_android_rvh_select_task_rq_fair'?
6766 | if (trace_android_rvh_select_task_rq_fair_enabled())
| ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
| trace_android_rvh_select_task_rq_fair
cc1: some warnings being treated as errors
Reported-by: kernel test robot <lkp@intel.com>
Fixes: c7ba09ce6a ("ANDROID: sched/fair: Sync task util for strf hook")
Signed-off-by: Todd Kjos <tkjos@google.com>
Change-Id: I1b15b9ada83dc5b2678cef3910a7aedc0d19f174
Since the sugov traces were added in kernel/sched/cpufreq_schedutil.c,
move it from power.h to sched.h.
And move the tracepoint to where the change happens.
Bug: 174488007
Test: build pass and boot to home
Signed-off-by: Jimmy Shiu <jimmyshiu@google.com>
Change-Id: Ibe18f04773c28be84db3619b1a4253bf08f0e289
Signed-off-by: Will McVicker <willmcvicker@google.com>
"flags" is added to the vendor hook parameter so that the module can know
the event type of task enqueue/dequeue.
Bug: 176917922
Signed-off-by: Park Bumgyu <bumgyu.park@samsung.com>
Change-Id: I7cc60908e301d75393bdf84861878a94de80d683
Add vendors hooks for to facilitate various scheduler value adds.
Bug: 176077958
Change-Id: I5d488ae78ce05f81e6c73b69c56128b065647fec
Signed-off-by: Shaleen Agrawal <shalagra@codeaurora.org>
At present, EAS gets disabled when on ASYM Capacity systems
if all BIG or Little CPUs gets hot-plugged. Instead of disabling
EAS by default, add trace hook and let vendor decide if EAS should
be disabled or not.
Bug: 176964092
Change-Id: I583272cc89d44f3e3a4b1c43e3f75d731092ebf6
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
A task can migrate either while it is waking or while it
is running via load balancer. Print the task status
i.e running or not in sched_migrate_task. This helps in
counting the different types of migrations without relying
on other trace events.
Bug: 176709810
Change-Id: Ib473f9ccdc78003bb1f5d2dc24354f2db7a684f5
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
Add vendor hook tracepoints to track when cpu util gets
updated and when freq is choosen.
Bug: 174488007
Signed-off-by: Saravana Kannan <skannan@codeaurora.org>
Signed-off-by: Rohit Gupta <rohgup@codeaurora.org>
Signed-off-by: Jonathan Avila <avilaj@codeaurora.org>
Signed-off-by: Jimmy Shiu <jimmyshiu@google.com>
Change-Id: Ibb22fd0337a2539820a05b1e6b54b09aeaebd040
Signed-off-by: Will McVicker <willmcvicker@google.com>
When entering cluster-wide or system-wide power mode, Exynos cpu
power management driver checks the next hrtimer events of cpu
composing the power domain to prevent unnecessary attempts to enter
the power mode. Since struct cpuidle_device has next_hrtimer, it
can be solved by passing cpuidle device as a parameter of vh.
In order to improve responsiveness, it is necessary to prevent
entering the deep idle state in boosting scenario. So, vendor
driver should be able to control the idle state.
Due to above requirements, the parameters required for idle enter
and exit different, so the vendor hook is separated into
cpu_idle_enter and cpu_idle_exit.
Bug: 176198732
Change-Id: I2262ba1bae5e6622a8e76bc1d5d16fb27af0bb8a
Signed-off-by: Park Bumgyu <bumgyu.park@samsung.com>
* aosp/upstream-f2fs-stable-linux-5.10.y:
f2fs: compress: fix compression chksum
f2fs: fix shift-out-of-bounds in sanity_check_raw_super()
f2fs: fix race of pending_pages in decompression
f2fs: fix to account inline xattr correctly during recovery
f2fs: inline: fix wrong inline inode stat
f2fs: inline: correct comment in f2fs_recover_inline_data
f2fs: don't check PAGE_SIZE again in sanity_check_raw_super()
f2fs: convert to F2FS_*_INO macro
f2fs: introduce max_io_bytes, a sysfs entry, to limit bio size
f2fs: don't allow any writes on readonly mount
f2fs: avoid race condition for shrinker count
f2fs: add F2FS_IOC_DECOMPRESS_FILE and F2FS_IOC_COMPRESS_FILE
f2fs: add compress_mode mount option
f2fs: Remove unnecessary unlikely()
f2fs: init dirty_secmap incorrectly
f2fs: remove buffer_head which has 32bits limit
f2fs: fix wrong block count instead of bytes
f2fs: use new conversion functions between blks and bytes
f2fs: rename logical_to_blk and blk_to_logical
f2fs: fix kbytes written stat for multi-device case
f2fs: compress: support chksum
f2fs: fix to avoid REQ_TIME and CP_TIME collision
f2fs: change to use rwsem for cp_mutex
f2fs: Handle casefolding with Encryption
fscrypt: Have filesystems handle their d_ops
libfs: Add generic function for setting dentry_ops
f2fs: Remove the redundancy initialization
f2fs: remove writeback_inodes_sb in f2fs_remount
f2fs: fix double free of unicode map
f2fs: fix compat F2FS_IOC_{MOVE,GARBAGE_COLLECT}_RANGE
f2fs: avoid unneeded data copy in f2fs_ioc_move_range()
f2fs: add F2FS_IOC_SET_COMPRESS_OPTION ioctl
f2fs: add F2FS_IOC_GET_COMPRESS_OPTION ioctl
f2fs: move ioctl interface definitions to separated file
f2fs: fix to seek incorrect data offset in inline data file
f2fs: call f2fs_get_meta_page_retry for nat page
Bug: 174873661
Signed-off-by: Jaegeuk Kim <jaegeuk@google.com>
Change-Id: I1edcbc31855a7b9d5fd5623ad0fd420e4885eaf2
Add a tracepoint for pause and resume which measures the
duration of time to perform the entire operation, the
cpus acted upon with this event, and the current state
of the active cpu mask. This should be sufficient
for testing pause performance.
Bug: 175959069
Change-Id: I9fc269c7d09ac78ec31612d3c552044b72b0e6e3
Signed-off-by: Stephen Dickey <dickey@codeaurora.org>
Like other filesystem does, we introduce a new file f2fs.h in path of
include/uapi/linux/, and move f2fs-specified ioctl interface definitions
to that file, after then, in order to use those definitions, userspace
developer only need to include the new header file rather than
copy & paste definitions from fs/f2fs/f2fs.h.
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Add a restricted vendor hook to check whether a set of tasks can
move to other cgorup.
Bug: 175808144
Signed-off-by: Choonghoon Park <choong.park@samsung.com>
Change-Id: If7bac83e0d2d1069b1436331989c3926645eab19
There are few vendor hooks available for binder wakeup related
functionality. As they both essentially do the same thing, we can
consolidate them into one.
Bug: 174219217
Change-Id: I44b472e7564eecbe8236ad2eb88b0433195f14d8
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
place_entity() vendor hook is meant to tweak vruntime by vendor
modules as needed, but with current form of the hook that is not
possible as vruntime is passed by it's value. Fix it by switching
to pass by reference.
Bug: 175448877
Change-Id: Ibb51592f94da31019fa98a6767d080ec61daafe6
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
Add a restricted vendor hook when a set of tasks change the cgroups in
the cpu controller. This facilitates various scheduler value adds.
Bug: 175045928
Change-Id: I544046d631f4d6c9bc2b999e054b5a296ec31a81
Signed-off-by: Abhijeet Dharmapurikar <adharmap@codeaurora.org>
Incorporate a vendor hook in the resume cpus path
so that vendor specific activities may take place.
Bug: 161210528
Change-Id: I74d03247491b004e891dbcfe06a478d00a95ba9f
Signed-off-by: Stephen Dickey <dickey@codeaurora.org>
We have debug infrastructure built on top of preempt/irq disable/enable
events. This requires modifications to the kernel tracing code. Since
this is not feasible with GKI, we started with registering to the
existing preemptirq trace events. However the performance of wide
variety of use cases are regressed as the rate of preemptirq events
is super high and generic trace events are slow.
Since GKI allows optimized trace events via restricted trace hooks,
add the same for preemptirq event.
Bug: 174541725
Change-Id: Ic8d3cdd1c1aa6a9267d0b755694fedffa2ea8e36
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Create a vendor hook inside of gfp_zone() to modify which allocations
get to enter ZONE_MOVABLE, by zeroing out __GFP_HIGHMEM inside of the
trace hook based on certain conditions.
Separately, create separate trace hooks in the swap-in and readahead
paths to affect the behavior of the tracehook in gfp_zone().
Bug: 158645321
Change-Id: I4a4f0b724267ee120a1e5661f6da5d43d7ef6fc6
Signed-off-by: Chris Goldsworthy <cgoldswo@codeaurora.org>
Vendors might want to change tasks affinity settings when they are
moving from one cpuset into the other. Add vendor hook to give control
to vendor to implement what they need.
Bug: 174125747
Change-Id: Icee0405be0bca432002dae4a26ebe945082ce052
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
Vendors might want to change tasks affinity settings when they are
moving from one cpuset into the other. Add vendor hook to give control
to vendor to implement what they need in sched_setaffinity().
Bug: 174125747
Change-Id: Ie703448147377cd62e76a58b620a7ab849a04924
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
This is to add capability for vendor to decide whether a cpufreq update
is needed, e.g. up/down rate limit.
Using restricted hook since it can be called from scheduler wakeup path.
Bug: 170511085
Signed-off-by: Wei Wang <wvw@google.com>
Change-Id: If9adea3a3e31efbf3858fbd009665a07dc70c638
(cherry picked from commit f9f3464532a045257f8138338b1beda86ef0a3be)
Signed-off-by: Will McVicker <willmcvicker@google.com>
Upstream moved the sugov to DEADLINE class which has higher prio than RT
so it can potentially block many RT use case in Android.
Also currently iowait doesn't distinguish background/foreground tasks
and we have seen cases where device run to high frequency unnecessarily
when running some background I/O.
Bug: 171598214
Signed-off-by: Wei Wang <wvw@google.com>
Change-Id: I21e9bfe9ef75a4178279574389e417c3f38e65ac
(cherry picked from commit 03177ef82bd942a3f163e826063491bae6ff0ac9)
Signed-off-by: Will McVicker <willmcvicker@google.com>
Add hooks for vendor specific find_energy_efficient_cpu logic.
Bug: 170507310
Signed-off-by: Rick Yiu <rickyiu@google.com>
Change-Id: I064b501017e32d4f22f8128bed8bf3a1508ab15b
(cherry picked from commit 2f108e2ec6e89609cbae32c5d13d6ad9f2e858cb)
Signed-off-by: Will McVicker <willmcvicker@google.com>
android_rvh_sched_nohz_balancer_kick hook allows vendor modules to
select the busiest CPU in a group during load balance. When the
load balancer could not pull tasks from this busiest CPU due to
affinity restriction, the CPU is cleared from env->cpu. This must
be passed to the vendor module, otherwise we keep selecting the
exempted CPU as the busiest CPU.
Bug: 174338902
Change-Id: Iedaa389a51849da4c3e094d731fe5e39cd909d81
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
The bdi_dev_name() returns a char [64], and
the __entry->name is a char [32].
It maybe dangerous to TP_printk("%s", __entry->name)
after the strncpy().
CC: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20201124165205.GA23937@rlk
Acked-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Hui Su <sh_def@163.com>
Signed-off-by: Jan Kara <jack@suse.cz>
The following restrict vendor hooks are added. The vendor hook
can selectively opt in for the default scheduler behavior by not
modifying the done argument.
- android_rvh_sched_newidle_balance: For newly idle load balance.
- android_rvh_sched_nohz_balancer_kick: For deciding if an idle
CPU is woken up to do nohz balance or not.
- android_rvh_find_busiest_queue: For selecting the busiest runqueue
among the CPUs in the busiest group selected in find_busiest_group.
- android_rvh_migrate_queued_task: Vendor implementations may require
both source and destination CPUs runqueue locks to be held while
calling set_task_cpu() during a task migration. Add a hook when
a queued task is migration so that vendor implementation can detach
the task and call set_task_cpu() with both runqueue locks held.
Bug: 173661641
Change-Id: I6a09226081061b6433e4231359be252a0f28f04b
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
QC value-add requires the ability to override the value displayed
by show_cpuinfo_max_freq.
Bug: 173984660
Signed-off-by: Todd Kjos <tkjos@google.com>
Change-Id: Ida11256f399f0730bc1228524b9687df3e7b8bf4
Right now, invocation of find_lowest_rq() vendor hook is made before
error checks and also, cpupri_find() isn't exported either. It would be
appropriate to move invocation of find_lowest_rq() vendor hook after
error checks are done & calling cpupri_find().
Bug: 173559623
Change-Id: I298dffd39be0451b0b154930ace4e16763c6e78d
Signed-off-by: Satya Durga Srinivasu Prabhala <satyap@codeaurora.org>
-----BEGIN PGP SIGNATURE-----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=R7xS
-----END PGP SIGNATURE-----
Merge tag 'nfsd-5.10-2' of git://linux-nfs.org/~bfields/linux
Pull nfsd fix from Bruce Fields:
"Just one quick fix for a tracing oops"
* tag 'nfsd-5.10-2' of git://linux-nfs.org/~bfields/linux:
SUNRPC: Fix oops in the rpc_xdr_buf event class
Add vendor hook to print epoch values when system enter and exit
out of suspend and resume. These epoch values are useful to know
how long the device is in suspend state. These values can be used
to synchronize various subsystem timestamps and have an unique
timestamp to correlate between various subsystems.
Bug: 172945021
Change-Id: I82a01e348d05a46c9c3921869cc9d2fc0fd28867
Signed-off-by: Murali Nalajala <mnalajal@codeaurora.org>
Backchannel rpc tasks don't have task->tk_client set, so it's necessary
to check it for NULL before dereferencing.
Fixes: c509f15a58 ("SUNRPC: Split the xdr_buf event class")
Signed-off-by: Scott Mayhew <smayhew@redhat.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Create a vendor hook for jiffies updates by the
tick_do_timer_cpu.
Bug: 148928265
Change-Id: Ia442e20d446b8ce4f2b3f2be76655e72919c76eb
Signed-off-by: Amir Vajid <avajid@codeaurora.org>
Tasklets are supposed to finish their work quickly and
should not block the current running process, but it is not
guaranteed that. Currently softirq_entry/exit can be used to
know total tasklets execution time, but not helpful to track
individual tasklet's execution time. With that we can't find
any culprit tasklet function, which is taking more time.
Add {hi}-tasklet_entry/exit trace point support to track
individual tasklet execution.
Bug: 168521633
Change-Id: I3496d15f64d020916774e673ccb4a8116ea2f2c9
Signed-off-by: Lingutla Chandrasekhar <clingutla@codeaurora.org>
[elavila: Port to mainline]
Signed-off-by: J. Avila <elavila@google.com>
refactoring.
-----BEGIN PGP SIGNATURE-----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=bpPt
-----END PGP SIGNATURE-----
Merge tag 'nfsd-5.10-1' of git://linux-nfs.org/~bfields/linux
Pull nfsd fixes from Bruce Fields:
"This is mainly server-to-server copy and fallout from Chuck's 5.10 rpc
refactoring"
* tag 'nfsd-5.10-1' of git://linux-nfs.org/~bfields/linux:
net/sunrpc: fix useless comparison in proc_do_xprt()
net/sunrpc: return 0 on attempt to write to "transports"
NFSD: fix missing refcount in nfsd4_copy by nfsd4_do_async_copy
NFSD: Fix use-after-free warning when doing inter-server copy
NFSD: MKNOD should return NFSERR_BADTYPE instead of NFSERR_INVAL
SUNRPC: Fix general protection fault in trace_rpc_xdr_overflow()
NFSD: NFSv3 PATHCONF Reply is improperly formed
Fast commits don't work with data journalling. This patch disables the
fast commit support when data journalling is turned on.
Suggested-by: Jan Kara <jack@suse.cz>
Signed-off-by: Harshad Shirwadkar <harshadshirwadkar@gmail.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20201106035911.1942128-19-harshadshirwadkar@gmail.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
