This feature should not be enabled in release but can be useful for
developers who need to monitor register accesses at some specific places.
Helped me identify a bug in u-boot, by comparing the register accesses
from the linux driver with the ones from its u-boot variant.
Signed-off-by: Tudor Ambarus <tudor.ambarus@microchip.com>
Link: https://lore.kernel.org/r/20200320065058.891221-1-tudor.ambarus@microchip.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently pmc_core_lpm_display() uses an array of the struct pointers,
i.e. tgl_lpm_maps for Tiger Lake directly to iterate through and to get
the number of (live) status registers which is hard coded and can not
be re-used for the future platforms that support sub-states. To maintain
readability, make pmc_core_lpm_display() generic, so that it can be re-used
for future platforms.
Cc: Chen Zhou <chenzhou10@huawei.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: David E. Box <david.e.box@intel.com>
Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Gayatri Kammela <gayatri.kammela@intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
This patch fix a spelling typo in error message.
Signed-off-by: Masanari Iida <standby24x7@gmail.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Fair number of changes including bug fixes done to change version.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
For platforms where multiple packages/die, this makes "Package-" key
duplicate. To make unique, add die and cpu id to key name.
So "Package-0" key name will change to "Package-0-die-x:cpu-x".
For example:
$sudo ./intel-speed-select -f json perf-profile info
Intel(R) Speed Select Technology
Executing on CPU model:106[0x6a]
{
"package-0:die-0:cpu-0": {
"perf-profile-level-0": {
"cpu-count": "32",
"enable-cpu-count": "32",
...
...
"package-1:die-0:cpu-16": {
"perf-profile-level-0": {
"cpu-count": "32",
"enable-cpu-count": "32",
"enable-cpu-mask": "ffff0000,ffff0000",
...
...
For non json format, there is no change. Here when print_package_info()
is called, it will return the level to print for other information.
This level is used formatting. Also in some function duplicate code
was there to print package,die and CPU information. Replace all that
code with a call to print_package_info().
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
In addition to total CPU count also display "enabled-cpu-count" for
perf-profile info command. This will show number of CPUs in the
"enable-cpu-mask".
For example:
perf-profile-level-4
cpu-count:32
enable-cpu-count:16
enable-cpu-mask:e42d0000,e42d0000
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
When user specifies invalid option, display "Unknown Option: ignore",
instead of "no match". Also display error for garbage on the command
line.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Specifying "avx2" and "avx512" option for display filter doesn't work
with short option "-r", only works with --try-type. Also compare full
6 characters for "avx512" string.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
CLX doesn't have capability to change the feature in the hardware, but
this acts as "--auto | -a" option. So even if user didn't specify the
option, use this as --auto | -a to set cpufreq scaling frequency limits.
Also remove perror with debug_printf as they don't bring any value.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
When --cpu or -c is used to specify target CPUs and non of them are valid,
display error.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
This change adds improved error display and handling for commands related
to core-power feature. The changes include:
- Replace perror with helpful error message
- Use ordered priority for SKX based platform by default as the
proportional priority is not supported
- Don't show weight and epp in help and also give error when user
tries to set them in SKX based platforms
- Range check for epp and weights and display error
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Treat a case when mailbox/mmio command can't be handled by the kernel
drivers when the module is removed or send a command which no driver can
handle. In this case ENOTTY result is returned, so print error.
Also when the isst_if_mmio module is removed, we can't send CLOS message
messages via Mailbox on non SKX based platforms. When this module is
removed, isst_platform_info.mmio_supported is set to 0. So it can't be
used as a condition to send via mailbox. Here replace check for Skylake
based platform to send via mailbox, other platforms can't use mailbox in
lieu of MMIO.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
This change adds improved error display and handling for commands related
to turbo-freq feature. The changes include:
- Replace perror/fprintf with helpful error message
- Error for not specifying TDP level when required
- Show error for invalid bucket number
- Show message to enable core-power before enabling turbo-freq feature
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
This change adds improved error display and handling for commands related
to base-freq feature. The changes include:
- Replace perror/fprintf with helpful error message
- Error for not specifying TDP level when required
- For CLX show help which shows limitation
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Improve output of perf-profile commands:
get-config-enabled
get-lock-status
Instead of showing 0/1, show meaningful strings. Also show error when
command is failed.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Enhance help to specify CPU and clos by an example.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
When priority type for core-power enable command is anything more than 1
display error before change to 1, which is ordered priority.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Before looking for information about the base-freq or turbo-freq details,
first check if the feature is supported at that level. If not print error
and return.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
This change adds improved error display and handling for commands related
to perf-profile feature. The changes include:
- When invalid TDP level is passed. display error and exit
- Replace perror with helpful error message
- Show error when TDP level can't be set
- Print error when information can't be read for a level
- Validate user options for invalid level
- Display error for TDP lock status
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Add a common API which can be used to print all error and information
messages. In this way a common format can be used.
For json output an error index in suffixed to make unique error key.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Add additional information, which will allow user to detect available
features. This will allow users to check presence of features before
continue to test.
A sample output:
$sudo ./intel-speed-select --info
Intel(R) Speed Select Technology
Executing on CPU model:85[0x55]
Platform: API version : 1
Platform: Driver version : 1
Platform: mbox supported : 1
Platform: mmio supported : 0
Intel(R) SST-PP (feature perf-profile) is not supported
Only performance level 0 (base level) is present
TDP level change control is locked
Intel(R) SST-TF (feature turbo-freq) is supported
Intel(R) SST-BF (feature base-freq) is supported
Intel(R) SST-CP (feature core-power) is supported
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Enhance help message which adds some example. The changes include:
- Print help when options are not recognized.
- For CLX, display only options which are applicable.
- Sort options in alphatical order.
- Disply help() instead of error:
"Feature name and|or command not specified"
- Remove duplicate display of
Intel(R) Speed Select Technology
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
When the device file "/dev/isst_interface" is not present, instead of
failing on access, check at the start and print a helpful warning.
Here CLX platform is an exception, which doesn't depend on the device
file. So continue for CLX platform.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Once the CPU is offline, the topology information (core-id, package-id,
die-id) is not accessible via sysfs. So when user selects a config level
more than base config 0 and offlined CPUs to match the config level,
to return to base config he has to manually online CPUs before. Without
this CPUs information mapping from Punit CPU numbering will lot work
as it needs atlest package id for each CPU.
To avoid this additional steps store the topology information in a file
, which is created on the very first run after boot. Since system boots
in base config and all CPUs are online, we can get information about
every CPU.
Once any of the APIs like get_physical_package_id(),
get_physical_core_id() or get_physical_die_id() fails to read from
sysfs, read from the stored mapping file.
This mapping file is stored in /tmp file system. so on every boot
it is recreated to make sure that any new CPUs are added to the
system before boot are taken into account.
But don't use the stored physical device id when trying to get
information for CPU to send message in for_each_online_package_in_set().
Here use the real value from syfs and in case fails try the next CPU.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Currently /sys/devices/system/cpu/cpu0/topology/thread_siblings is used to
get the max CPU count. But when CPU0 is offline, then this file will be
absent.
So add processing so that we can get count from any first CPU in the
system. which is online.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
When "-o" option for force online/offline is used with command:
perf-profile set-config-level
If the config level calls for CPU 0 online/offline, then call fails
as there is special kernel setup required for CPU 0 online/offline
and the currently not setup for that.
But when call is for online CPU 0, then don't fail. Just warn that
this system is not setup for CPU 0 online/offline.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Instead of displaying 0 and 1 for enable status, display "disabled"
and "enabled" respectively.
Similarly for priority type, display "ordered or proportional" instead
of 0 and 1.
An example display:
$intel-speed-select -c 1 core-power info
Intel(R) Speed Select Technology
..
package-0
die-0
cpu-1
core-power
support-status:supported
enable-status:enabled
clos-enable-status:enabled
priority-type:proportional
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
In addition to CLOS enable status, also show the core-power feature
status. This will help why clos enable status didn't give desired
results as the core-power feature may be disabled or unsupported.
The new display looks as follows:
$intel-speed-select core-power info
Intel(R) Speed Select Technology
..
package-0
die-0
cpu-0
core-power
support-status:supported
enable-status:enabled
clos-enable-status:1
priority-type:0
In the above display "support-status" and "enable-status", shows the
status of the core-power feature and "clos-enable-status", shows the
status of the clos.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Currently "-c" is a mandatory option for "core-power info" command. Make
this optional as this is a per package/die property. When not specified,
it will print info for every package/die.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
When CPU is offline, we can't get package id. So print error for this
and don't use output.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Here topology_max_cpus is used for total CPU count, not the last CPU
number. So remove "-1".
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Even for the products using MMIO, this message needs to be sent via
mail box. The previous fix done for this didn't properly address this.
That fix simply removed sending command via MMIO, but still didn't
trigger sending via mailbox.
Add additional condition to check for CLOS_PM_QOS_CONFIG, when MMIO
is supported on a platform.
Fixes: cd0e637065 (tools/power/x86/intel-speed-select: Use mailbox for CLOS_PM_QOS_CONFIG)
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Here is a pull request with interconnect changes for the 5.7-rc1 merge
window. It contains just driver updates, and these are:
- Refactoring of the SDM845 driver, which is now improved to better
represent the hardware.
- New driver for SC7180 platforms.
- New driver for OSM L3 interconnect hardware found on SDM845/SC7180
platforms.
Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
-----BEGIN PGP SIGNATURE-----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=jHM0
-----END PGP SIGNATURE-----
Merge tag 'icc-5.7-rc1' of https://git.linaro.org/people/georgi.djakov/linux into char-misc-next
Georgi writes:
interconnect changes for 5.7
Here is a pull request with interconnect changes for the 5.7-rc1 merge
window. It contains just driver updates, and these are:
- Refactoring of the SDM845 driver, which is now improved to better
represent the hardware.
- New driver for SC7180 platforms.
- New driver for OSM L3 interconnect hardware found on SDM845/SC7180
platforms.
Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
* tag 'icc-5.7-rc1' of https://git.linaro.org/people/georgi.djakov/linux:
interconnect: qcom: Add OSM L3 support on SC7180
dt-bindings: interconnect: Add OSM L3 DT binding on SC7180
interconnect: qcom: Add OSM L3 interconnect provider support
dt-bindings: interconnect: Add OSM L3 DT bindings
interconnect: qcom: Allow icc node to be used across icc providers
interconnect: qcom: Add SC7180 interconnect provider driver
dt-bindings: interconnect: Add Qualcomm SC7180 DT bindings
interconnect: qcom: sdm845: Split qnodes into their respective NoCs
interconnect: qcom: Consolidate interconnect RPMh support
dt-bindings: interconnect: Update Qualcomm SDM845 DT bindings
dt-bindings: interconnect: Add YAML schemas for QCOM bcm-voter
dt-bindings: interconnect: Convert qcom,sdm845 to DT schema
Since snprintf() returns the would-be-output size instead of the
actual output size, the succeeding calls may go beyond the given
buffer limit. Fix it by replacing with scnprintf().
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Commit 1f27dbd826 ("platform/x86: GPD pocket fan: Allow somewhat
lower/higher temperature limits") changed the module-param sanity check
to accept temperature limits between 20 and 90 degrees celcius.
But the error message printed when the module params are outside this
range was not updated. This commit updates the error message to match
the new min and max value for the temp-limits.
Reported-by: Pavel Machek <pavel@denx.de>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Acked-by: Pavel Machek <pavel@denx.de>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
The MMIO driver is not unregistering with the correct type with the ISST
common core during module removal. This should be unregistered with
ISST_IF_DEV_MMIO instead of ISST_IF_DEV_MBOX.
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Function fan_boost_mode_store returns 0 if store is successful,
this leads to infinite loop after any write to it's sysfs entry:
# echo 0 >/sys/devices/platform/asus-nb-wmi/fan_boost_mode
This command never ends, one CPU core is at 100% utilization.
This patch fixes this by returning size of written data.
Fixes: b096f626a6 ("platform/x86: asus-wmi: Switch fan boost mode")
Signed-off-by: Leonid Maksymchuk <leonmaxx@gmail.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
The WMI method to set the charge threshold does not provide a
way to specific a battery, so we assume it is the first/primary
battery (by checking if the name is BAT0).
On some newer ASUS laptops (Zenbook UM431DA) though, the
primary/first battery isn't named BAT0 but BATT, so we need
to support that case.
Fixes: 7973353e92 ("platform/x86: asus-wmi: Refactor charge threshold to use the battery hooking API")
Cc: stable@vger.kernel.org
Signed-off-by: Kristian Klausen <kristian@klausen.dk>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
nmi_enter() does lockdep_off() and hence lockdep ignores everything.
And NMI context makes it impossible to do full IN-NMI tracking like we
do IN-HARDIRQ, that could result in graph_lock recursion.
However, since look_up_lock_class() is lockless, we can find the class
of a lock that has prior use and detect IN-NMI after USED, just not
USED after IN-NMI.
NOTE: By shifting the lockdep_off() recursion count to bit-16, we can
easily differentiate between actual recursion and off.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Link: https://lkml.kernel.org/r/20200221134215.090538203@infradead.org
There were two patterns for lockdep_recursion:
Pattern-A:
if (current->lockdep_recursion)
return
current->lockdep_recursion = 1;
/* do stuff */
current->lockdep_recursion = 0;
Pattern-B:
current->lockdep_recursion++;
/* do stuff */
current->lockdep_recursion--;
But a third pattern has emerged:
Pattern-C:
current->lockdep_recursion = 1;
/* do stuff */
current->lockdep_recursion = 0;
And while this isn't broken per-se, it is highly dangerous because it
doesn't nest properly.
Get rid of all Pattern-C instances and shore up Pattern-A with a
warning.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200313093325.GW12561@hirez.programming.kicks-ass.net
Qian Cai reported a bug when PROVE_RCU_LIST=y, and read on /proc/lockdep
triggered a warning:
[ ] DEBUG_LOCKS_WARN_ON(current->hardirqs_enabled)
...
[ ] Call Trace:
[ ] lock_is_held_type+0x5d/0x150
[ ] ? rcu_lockdep_current_cpu_online+0x64/0x80
[ ] rcu_read_lock_any_held+0xac/0x100
[ ] ? rcu_read_lock_held+0xc0/0xc0
[ ] ? __slab_free+0x421/0x540
[ ] ? kasan_kmalloc+0x9/0x10
[ ] ? __kmalloc_node+0x1d7/0x320
[ ] ? kvmalloc_node+0x6f/0x80
[ ] __bfs+0x28a/0x3c0
[ ] ? class_equal+0x30/0x30
[ ] lockdep_count_forward_deps+0x11a/0x1a0
The warning got triggered because lockdep_count_forward_deps() call
__bfs() without current->lockdep_recursion being set, as a result
a lockdep internal function (__bfs()) is checked by lockdep, which is
unexpected, and the inconsistency between the irq-off state and the
state traced by lockdep caused the warning.
Apart from this warning, lockdep internal functions like __bfs() should
always be protected by current->lockdep_recursion to avoid potential
deadlocks and data inconsistency, therefore add the
current->lockdep_recursion on-and-off section to protect __bfs() in both
lockdep_count_forward_deps() and lockdep_count_backward_deps()
Reported-by: Qian Cai <cai@lca.pw>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200312151258.128036-1-boqun.feng@gmail.com
The IMC uncore unit in Ice Lake server can only be accessed by MMIO,
which is similar as Snow Ridge.
Factor out __snr_uncore_mmio_init_box which can be shared with Ice Lake
server in the following patch.
No functional changes.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/1584470314-46657-2-git-send-email-kan.liang@linux.intel.com
The offset between uncore boxes of free-running counters varies, e.g.
IIO free-running counters on Ice Lake server.
Add box_offsets, an array of offsets between adjacent uncore boxes.
Signed-off-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/1584470314-46657-1-git-send-email-kan.liang@linux.intel.com
This NULL check is reversed so it leads to a Smatch warning and
presumably a NULL dereference.
kernel/events/core.c:1598 perf_event_groups_less()
error: we previously assumed 'right->cgrp->css.cgroup' could be null
(see line 1590)
Fixes: 95ed6c707f ("perf/cgroup: Order events in RB tree by cgroup id")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200312105637.GA8960@mwanda
Kan and Andi reported that we fail to kill rotation when the flexible
events go empty, but the context does not. XXX moar
Fixes: fd7d55172d ("perf/cgroups: Don't rotate events for cgroups unnecessarily")
Reported-by: Andi Kleen <ak@linux.intel.com>
Reported-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200305123851.GX2596@hirez.programming.kicks-ass.net
While looking at an objtool UACCESS warning, it suddenly occurred to me
that it is entirely possible to have an OPTPROBE right in the middle of
an UACCESS region.
In this case we must of course clear FLAGS.AC while running the KPROBE.
Luckily the trampoline already saves/restores [ER]FLAGS, so all we need
to do is inject a CLAC. Unfortunately we cannot use ALTERNATIVE() in the
trampoline text, so we have to frob that manually.
Fixes: ca0bbc70f147 ("sched/x86_64: Don't save flags on context switch")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Link: https://lkml.kernel.org/r/20200305092130.GU2596@hirez.programming.kicks-ass.net
In update_sg_wakeup_stats(), the comment says:
Computing avg_load makes sense only when group is fully
busy or overloaded.
But, the code below this comment does not check like this.
From reading the code about avg_load in other functions, I
confirm that avg_load should be calculated in fully busy or
overloaded case. The comment is correct and the checking
condition is wrong. So, change that condition.
Fixes: 57abff067a ("sched/fair: Rework find_idlest_group()")
Signed-off-by: Tao Zhou <ouwen210@hotmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Link: https://lkml.kernel.org/r/Message-ID:
If we failed to find a fitting CPU, in cpupri_find(), we only fallback
to the level we found a hit at.
But Steve suggested to fallback to a second full scan instead as this
could be a better effort.
https://lore.kernel.org/lkml/20200304135404.146c56eb@gandalf.local.home/
We trigger the 2nd search unconditionally since the argument about
triggering a full search is that the recorded fall back level might have
become empty by then. Which means storing any data about what happened
would be meaningless and stale.
I had a humble try at timing it and it seemed okay for the small 6 CPUs
system I was running on
https://lore.kernel.org/lkml/20200305124324.42x6ehjxbnjkklnh@e107158-lin.cambridge.arm.com/
On large system this second full scan could be expensive. But there are
no users outside capacity awareness for this fitness function at the
moment. Heterogeneous systems tend to be small with 8cores in total.
Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Qais Yousef <qais.yousef@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/20200310142219.syxzn5ljpdxqtbgx@e107158-lin.cambridge.arm.com
when we create a kthread with ktrhead_create_on_cpu(),the child thread
entry is ktread.c:ktrhead() which will be preempted by the parent after
call complete(done) while schedule() is not called yet,then the parent
will call wait_task_inactive(child) but the child is still on the runqueue,
so the parent will schedule_hrtimeout() for 1 jiffy,it will waste a lot of
time,especially on startup.
parent child
ktrhead_create_on_cpu()
wait_fo_completion(&done) -----> ktread.c:ktrhead()
|----- complete(done);--wakeup and preempted by parent
kthread_bind() <------------| |-> schedule();--dequeue here
wait_task_inactive(child) |
schedule_hrtimeout(1 jiffy) -|
So we hope the child just wakeup parent but not preempted by parent, and the
child is going to call schedule() soon,then the parent will not call
schedule_hrtimeout(1 jiffy) as the child is already dequeue.
The same issue for ktrhead_park()&&kthread_parkme().
This patch can save 120ms on rk312x startup with CONFIG_HZ=300.
Signed-off-by: Liang Chen <cl@rock-chips.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/20200306070133.18335-2-cl@rock-chips.com
