Currently we have structrues comp (which is empty) and comp_info being
used to register and deregister the component. This mismatch in naming
occurred from a previous commit that renamed aim_info to comp. Fix this
to use consistent component naming in line with most/net, most/sound etc.
This fixes the message two issues, one with a null empty name when
loading the module:
[ 1485.269515] most_core: registered new core component (null)
and an Oops when removing the module:
[ 1485.277971] BUG: unable to handle kernel NULL pointer dereference at 0000000000000008
[ 1485.278648] PGD 0 P4D 0
[ 1485.279253] Oops: 0002 [#2] SMP PTI
[ 1485.279847] CPU: 1 PID: 32629 Comm: modprobe Tainted: P D WC OE 4.18.0-8-generic #9
[ 1485.280442] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 0.0.0 02/06/2015
[ 1485.281040] RIP: 0010:most_deregister_component+0x3c/0x70 [most_core]
.. etc
Fixes: 1b10a0316e ("staging: most: video: remove aim designators")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Don't populate the array 'broadcast' on the stack but instead make it
static. Makes the object code smaller by 53 bytes:
Before:
text data bss dec hex filename
10511 1521 448 12480 30c0 drivers/staging/most/net/net.o
After:
text data bss dec hex filename
10394 1585 448 12427 308b drivers/staging/most/net/net.o
(gcc version 8.2.0 x86_64)
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of creating a dummy child device to manage the card object,
just use devm stuff directly for releasing with snd_card_free().
This results in a lot of code reduction.
Since the dummy child devices are gone, the device object to be passed
to the memory allocator needs to be adjusted as well.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Just a minor code refactoring and adding some const prefix.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All three functions to create PCM objects are fairly resemble, and can
be unified to a single common helper.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The recent ALSA PCM core supports the SNDRV_PCM_INFO_SYNC_APPLPTR flag
indicating that the driver needs the ack call at each appl_ptr
update. This is requirement for the indirect PCM implementations like
bcm2835-audio driver, too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The memory access to the pages allocated with
SNDRV_DMA_TYPE_CONTINUOUS are basically non-coherent, and it becomes a
problem when a process accesses via mmap.
For the more consistent access, use the device coherent memory, just
by replacing the call pattern in the allocator helpers.
The only point we need to be careful for is the device object passed
there; since bcm2835-audio driver creates fake devices and each card
is created on top of that, we need to pass its parent device as the
real device object.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For more consistency, move the module parameter description right
after its variable definition.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Yet a few header files are included unnecessarily. Drop them.
Also remove trivial comments.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
For making the whole code more consistent, replace the home-made debug
print macros with the standard dev_err() & co.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Instead of allocating a separate snd_device object, let snd_card_new()
allocate the private resource. This simplifies the code.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is the most significant part in the patch series.
The bcm2835-audio driver used to queue the commands to vc04 core via
workqueue, but basically the whole accesses to vc04 core are done in
the sleepable context, including the callback calls. In such a case,
rewriting the code using non-atomic PCM ops will simplify the logic a
lot.
This patch does it: all workqueue are gone and each former-work
implementation is now directly called from PCM ops like trigger and
write transfer.
Along with it, the DMA position updater, bcm2835_playback_fifo(), was
also rewritten to use a simpler logic. Now it handles the XRUN and
draining properly by calling snd_pcm_stop() conditionally.
The current position is kept in atomic_t value so that it can be read
concurrently from the pointer callback.
Also, the bcm2835_audio_instance object is allocated at the beginning
of bcm2835_audio_open(). This makes the resource management clearer.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
This is a cleanup and code refactoring in bcm2835-vchiq.c.
The major code changes are to provide local helpers for easier use of
lock / unlock, and message passing with/without response wait. This
allows us to reduce lots of open codes.
Also, the max packet is set at opening the stream, not at each time
when the write gets called.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The bcm2835_audio_instance object contains the array of
VCHI_SERVICE_HANDLE_T, while the code assumes and uses only the first
element explicitly. Let's reduce to a single vchi handle for
simplifying the code.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
It seems that the resolution of vc04 callback is in 10 msec; i.e. the
minimal period size is also 10 msec.
This patch adds the corresponding hw constraint.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The chip->audio_mutex is used basically for protecting the opened
stream assignment, and the prepare callback is irrelevant with it.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
These debug messages worsen the code readability a lot while they give
little debuggability (which we already have via tracing, in anyway).
Let's clean them up. This allows us to reduce the
snd_bcm2835_pcm_lib_ioctl() function to be a direct call of the
snd_pcm_lib_ioctl callback (like most other drivers do), too.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
When the parameter setup fails, the driver should propagate the error
code instead of silently ignoring it.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
alsa_stream->chip can be never NULL.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The hw_queue_size of PCM indirect helper doesn't need to be set up if
you use the whole given buffer size. Drop the useless
initialization, which just confuses readers.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Some fields in alsa_stream are the values we keep already in PCM
runtime object, hence they are redundant. Use the standard PCM
runtime values instead of the private copies.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The alsa_stream->lock is never used. Kill it.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The handling of SNDRV_PCM_TRIGGER_STOP at the trigger callback is
incorrect: when the STOP is issued, the driver is supposed to drop the
stream immediately. Meanwhile bcm2835 driver checks the DRAINING
state and tries to issue some different command.
This patch straightens things a bit, dropping the incorrect state
checks. The draining behavior would be still not perfect at this
point, but will be improved in a later patch.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The running flag of alsa_stream is basically useless. The running
state is strictly controlled in ALSA PCM core side, hence the check in
PCM trigger and close callbacks are superfluous.
Also, the prefill ack at trigger start became superfluous nowadays
with the ALSA PCM core update.
Let's rip them off.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
All the alsa_stream->open flag checks in the current code are
redundant, and they cannot be racy. For the code simplification,
let's remove the flag and its check.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
bcm2835_audio_setup(), bcm2835_audio_flush_buffers() and
bcm2835_audio_flush_playback_buffers() functions do implement
nothing.
Also, bcm2835_audio_set_ctls() is already called inside
bcm2835_audio_set_params(), so the later call is superfluous.
This patch removes these superfluous implementations.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
In the current code, the mute control is dealt in a special manner,
modifying the current volume and saving the old volume, etc. This is
inconsistent (e.g. change the volume while muted, then unmute), and
way too complex.
Also, the whole volume handling code has conversion between ALSA
volume and raw volume values, which can lead to another
inconsistency and complexity.
This patch simplifies these points:
- The ALSA volume value is saved in chip->volume
- volume->mute saves the mute state
- The mute state is evaluated only when the actual volume is passed to
the hardware, bcm2835_audio_set_ctls()
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The avail_substreams bit mask is checked for the possible racy
accesses, but this cannot happen in practice; i.e. the assignment and
the check are superfluous.
Let's rip them off.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Only a few of them are really needed.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
The "IEC958 Playback Stream" control does basically the very same
thing as "IEC958 Playback Default" redundantly. The former should
have been stream-specific and restored after closing the stream, but
we don't do in that way.
Since it's nothing but confusion, remove this fake.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
snd-bcm2835 driver takes the lock with mutex_lock_interruptible() in
all places, which don't make sense. Replace them with the simple
mutex_lock().
Also taking a mutex lock right after creating it for each PCM object
is nonsense, too. It cannot be racy at that point. We can get rid of
it.
Last but not least, initializing chip->audio_mutex at each place is
error-prone. Initialize properly at creating the chip object in
snd_bcm2835_create() instead.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Replace all the memcpy() for copying name strings from the firmware with
strlcpy() to make sure we are bounded by the source buffer size and we
also always have NULL-terminated strings.
This is needed to avoid out of bounds accesses if the firmware returns
a non-terminated string.
Reported-by: Olof Johansson <olof@lixom.net>
Acked-by: Olof Johansson <olof@lixom.net>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
Add a driver for the Innosilicon hdmi phy used on rk3228/rk3229
and rk3328 socs from Rockchip.
Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Tested-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
The phy is used so far in two Rockchip socs the rk3228 and the rk3328.
Signed-off-by: Zheng Yang <zhengyang@rock-chips.com>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Reviewed-by: Rob Herring <robh@kernel.org>
Signed-off-by: Kishon Vijay Abraham I <kishon@ti.com>
include/trace/events/sched.h includes <linux/sched.h> (via
<linux/sched/numa_balancing.h>) and so knows about the TASK_* constants
used to interpret .prev_state. So instead of duplicating the magic
numbers make use of the defined macros to ease understanding the
mapping from state bits to letters which isn't completely intuitive for
an outsider.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: kernel@pengutronix.de
Link: http://lkml.kernel.org/r/20180905093636.24068-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Ingo Molnar <mingo@kernel.org>
nr_running in struct numa_stats is not used anywhere in the code.
Remove it.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1535548752-4434-3-git-send-email-vincent.guittot@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
With:
commit 2d4056fafa ("sched/numa: Remove numa_has_capacity()")
the local variables 'smt', 'cpus' and 'capacity' and their results are not used
anymore in numa_has_capacity()
Remove this unused code.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Srikar Dronamraju <srikar@linux.vnet.ibm.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Rik van Riel <riel@surriel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/1535548752-4434-2-git-send-email-vincent.guittot@linaro.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
LLVM has a warning that tags expressions like:
if (foo && non-bool-const)
This pattern triggers for CONFIG_SCHED_DEBUG=n where sched_feat() ends
up being whatever bit we select. Avoid the warning with an explicit
cast to bool.
Reported-by: Philipp Klocke <philipp97kl@gmail.com>
Tested-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The 'prefer sibling' sched_domain flag is intended to encourage
spreading tasks to sibling sched_domain to take advantage of more caches
and core for SMT systems. It has recently been changed to be on all
non-NUMA topology level. However, spreading across domains with CPU
capacity asymmetry isn't desirable, e.g. spreading from high capacity to
low capacity CPUs even if high capacity CPUs aren't overutilized might
give access to more cache but the CPU will be slower and possibly lead
to worse overall throughput.
To prevent this, we need to remove SD_PREFER_SIBLING on the sched_domain
level immediately below SD_ASYM_CPUCAPACITY.
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-13-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
When lower capacity CPUs are load balancing and considering to pull
something from a higher capacity group, we should not pull tasks from a
CPU with only one task running as this is guaranteed to impede progress
for that task. If there is more than one task running, load balance in
the higher capacity group would have already made any possible moves to
resolve imbalance and we should make better use of system compute
capacity by moving a task if we still have more than one running.
Signed-off-by: Chris Redpath <chris.redpath@arm.com>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-11-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Idle balance is a great opportunity to pull a misfit task. However,
there are scenarios where misfit tasks are present but idle balance is
prevented by the overload flag.
A good example of this is a workload of n identical tasks. Let's suppose
we have a 2+2 Arm big.LITTLE system. We then spawn 4 fairly
CPU-intensive tasks - for the sake of simplicity let's say they are just
CPU hogs, even when running on big CPUs.
They are identical tasks, so on an SMP system they should all end at
(roughly) the same time. However, in our case the LITTLE CPUs are less
performing than the big CPUs, so tasks running on the LITTLEs will have
a longer completion time.
This means that the big CPUs will complete their work earlier, at which
point they should pull the tasks from the LITTLEs. What we want to
happen is summarized as follows:
a,b,c,d are our CPU-hogging tasks _ signifies idling
LITTLE_0 | a a a a _ _
LITTLE_1 | b b b b _ _
---------|-------------
big_0 | c c c c a a
big_1 | d d d d b b
^
^
Tasks end on the big CPUs, idle balance happens
and the misfit tasks are pulled straight away
This however won't happen, because currently the overload flag is only
set when there is any CPU that has more than one runnable task - which
may very well not be the case here if our CPU-hogging workload is all
there is to run.
As such, this commit sets the overload flag in update_sg_lb_stats when
a group is flagged as having a misfit task.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-10-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This variable can be read and set locklessly within update_sd_lb_stats().
As such, READ/WRITE_ONCE() are added to make sure nothing terribly wrong
can happen because of the compiler.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-9-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
sizeof(_Bool) is implementation defined, so let's just go with 'int' as
is done for other structures e.g. sched_domain_shared->has_idle_cores.
The local 'overload' variable used in update_sd_lb_stats can remain
bool, as it won't impact any struct layout and can be assigned to the
root_domain field.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-8-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This variable is entirely local to update_sd_lb_stats, so we can
safely change its type and slightly clean up its initialisation.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-7-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
There already are a few conditions in nohz_kick_needed() to ensure
a nohz kick is triggered, but they are not enough for some misfit
task scenarios. Excluding asym packing, those are:
- rq->nr_running >=2: Not relevant here because we are running a
misfit task, it needs to be migrated regardless and potentially through
active balance.
- sds->nr_busy_cpus > 1: If there is only the misfit task being run
on a group of low capacity CPUs, this will be evaluated to False.
- rq->cfs.h_nr_running >=1 && check_cpu_capacity(): Not relevant here,
misfit task needs to be migrated regardless of rt/IRQ pressure
As such, this commit adds an rq->misfit_task_load condition to trigger a
nohz kick.
The idea to kick a nohz balance for misfit tasks originally came from
Leo Yan <leo.yan@linaro.org>, and a similar patch was submitted for
the Android Common Kernel - see:
https://lists.linaro.org/pipermail/eas-dev/2016-September/000551.html
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-6-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
On asymmetric CPU capacity systems load intensive tasks can end up on
CPUs that don't suit their compute demand. In this scenarios 'misfit'
tasks should be migrated to CPUs with higher compute capacity to ensure
better throughput. group_misfit_task indicates this scenario, but tweaks
to the load-balance code are needed to make the migrations happen.
Misfit balancing only makes sense between a source group of lower
per-CPU capacity and destination group of higher compute capacity.
Otherwise, misfit balancing is ignored. group_misfit_task has lowest
priority so any imbalance due to overload is dealt with first.
The modifications are:
1. Only pick a group containing misfit tasks as the busiest group if the
destination group has higher capacity and has spare capacity.
2. When the busiest group is a 'misfit' group, skip the usual average
load and group capacity checks.
3. Set the imbalance for 'misfit' balancing sufficiently high for a task
to be pulled ignoring average load.
4. Pick the CPU with the highest misfit load as the source CPU.
5. If the misfit task is alone on the source CPU, go for active
balancing.
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-5-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The current sg->min_capacity tracks the lowest per-CPU compute capacity
available in the sched_group when rt/irq pressure is taken into account.
Minimum capacity isn't the ideal metric for tracking if a sched_group
needs offloading to another sched_group for some scenarios, e.g. a
sched_group with multiple CPUs if only one is under heavy pressure.
Tracking maximum capacity isn't perfect either but a better choice for
some situations as it indicates that the sched_group definitely compute
capacity constrained either due to rt/irq pressure on all CPUs or
asymmetric CPU capacities (e.g. big.LITTLE).
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-4-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
To maximize throughput in systems with asymmetric CPU capacities (e.g.
ARM big.LITTLE) load-balancing has to consider task and CPU utilization
as well as per-CPU compute capacity when load-balancing in addition to
the current average load based load-balancing policy. Tasks with high
utilization that are scheduled on a lower capacity CPU need to be
identified and migrated to a higher capacity CPU if possible to maximize
throughput.
To implement this additional policy an additional group_type
(load-balance scenario) is added: 'group_misfit_task'. This represents
scenarios where a sched_group has one or more tasks that are not
suitable for its per-CPU capacity. 'group_misfit_task' is only considered
if the system is not overloaded or imbalanced ('group_imbalanced' or
'group_overloaded').
Identifying misfit tasks requires the rq lock to be held. To avoid
taking remote rq locks to examine source sched_groups for misfit tasks,
each CPU is responsible for tracking misfit tasks themselves and update
the rq->misfit_task flag. This means checking task utilization when
tasks are scheduled and on sched_tick.
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-3-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
The existing asymmetric CPU capacity code should cause minimal overhead
for others. Putting it behind a static_key, it has been done for SMT
optimizations, would make it easier to extend and improve without
causing harm to others moving forward.
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: gaku.inami.xh@renesas.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1530699470-29808-2-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Asymmetric CPU capacity can not necessarily be determined accurately at
the time the initial sched_domain hierarchy is built during boot. It is
therefore necessary to be able to force a full rebuild of the hierarchy
later triggered by the arch_topology driver. A full rebuild requires the
arch-code to implement arch_update_cpu_topology() which isn't yet
implemented for arm. This patch points the arm implementation to
arch_topology driver to ensure that full hierarchy rebuild happens when
needed.
Signed-off-by: Morten Rasmussen <morten.rasmussen@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: dietmar.eggemann@arm.com
Cc: valentin.schneider@arm.com
Cc: vincent.guittot@linaro.org
Link: http://lkml.kernel.org/r/1532093554-30504-5-git-send-email-morten.rasmussen@arm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>