With the merge window having been delayed for another week here's
another batch of updates that came in during that week. There's a few
important fixes in here, mainly a fix for I/O on a number of devices
caused by some of the component rework and a fix for a potential issue
if more than one component in a link provides compressed operations.
The I/O fixes are particularly important as the problem causes a power
regression on a number of OMAP platforms.
-----BEGIN PGP SIGNATURE-----
iQFHBAABCgAxFiEEreZoqmdXGLWf4p/qJNaLcl1Uh9AFAlp65E4THGJyb29uaWVA
a2VybmVsLm9yZwAKCRAk1otyXVSH0Oy1B/9XQd6twiSRfNKtbdujOyvjc8lpZ01n
JYHVySXyL9ZqAnIblM7Or/iXVEtsAVcXgFegF3SLHAY6VF7DQ1pDolnxUtuXOxOj
j+/2Y4wnDGCjXEr0tMoxrbNUIqlVZLCpPwsPo3vvVbr6sLLmQYVposNp2A2sK2bz
uWm9E3Nr26Q0UctzjWQM5+AFHSouyL7zDPfBCoWkEToP7163w6r4JDr991KdNGwP
Ac+5qjRUSldsn8WB2ngm8ioqbq+aOvsz2THYjG8gxrlQK+BWsyCDqF7f1d9GWse2
7k+xZLrdJrVkBMOnpvOx/Y4KRfe9BAFZBZ3KRbi2IR++7TD3902xEX27
=aIRm
-----END PGP SIGNATURE-----
Merge tag 'asoc-v4.16-5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound
Pull more ASoC updates from Mark Brown:
"With the merge window having been delayed for another week here's
another batch of updates that came in during that week.
There's a few important fixes in here, mainly a fix for I/O on a
number of devices caused by some of the component rework and a fix for
a potential issue if more than one component in a link provides
compressed operations. The I/O fixes are particularly important as the
problem causes a power regression on a number of OMAP platforms"
* tag 'asoc-v4.16-5' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound: (22 commits)
ASoC: stm32: add of dependency for stm32 drivers
ASoC: mt8173-rt5650: fix child-node lookup
ASoC: dapm: fix debugfs read using path->connected
ASoC: compress: Fixup error messages
ASoC: compress: Remove some extraneous blank lines
ASoC: compress: Correct handling of copy callback
ASoC: Intel: kbl: Enable mclk and ssp sclk early
ASoC: Intel: Skylake: Add extended I2S config blob support in Clock driver
ASoC: Intel: Skylake: Add ssp clock driver
ASoC: Fix twl4030 and 6040 regression by adding back read and write
ASoC: sun8i-codec: Add ADC support for a33
ASoC: rockchip: Use dummy_dai for rt5514 dsp dailink
ASoC: soc-pcm: rename .pmdown_time to .use_pmdown_time for Component
ASoC: ak4613: call dummy write for PW_MGMT1/3 when Playback
ASoC: soc-pcm: don't call flush_delayed_work() many times in soc_pcm_private_free()
ASoC: soc-core: snd_soc_rtdcom_lookup() cares component driver name
ASoC: sam9x5_wm8731: Drop 'ASoC' prefix from error messages
ASoC: sam9g20_wm8731: use dev_*() logging functions
ASoC: max98373 Changed SPDX header in C++ comments style
ASoC: dmic: Fix check of return value from read of 'num-channels'
...
This fix a bug in dapm_widget_power_read_file(),
where it may sent opposite order of source/sink widget
into the p->connected().
for example,
static int connected_check(source, sink);
{"w_sink", NULL, "w_source", connected_check}
the dapm_widget_power_read_file() will query p->connected()
in following case
p->conneted("w_source", "w_sink")
p->conneted("w_sink", "w_source")
we should avoid the last case, since it's the wrong order (source/sink)
as declared in snd_soc_dapm_route.
Signed-off-by: KaiChieh Chuang <kaichieh.chuang@mediatek.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The newly introduced function is declared as globally visible,
but is not declared in a header, causing a warning 'make W=1'
or 'make C=1':
sound/soc/soc-dapm.c:3782:1: warning: symbol 'snd_soc_dapm_free_kcontrol' was not declared. Should it be static?
The suggestion to make it static seems appropriate here, so let's
do that.
Fixes: 19ad683abc ("ASoC: dapm: Avoid creating kcontrol for params")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
We now allocate the array conditionally, but we always pass
the pointer to the new snd_soc_dapm_free_kcontrol() function,
which introduces a warning for the case that it is not
initialized:
sound/soc/soc-dapm.c: In function 'snd_soc_dapm_new_pcm':
sound/soc/soc-dapm.c:3940:2: error: 'w_param_text' may be used uninitialized in this function [-Werror=maybe-uninitialized]
As snd_soc_dapm_free_kcontrol() is global, it doesn't get inlined
and gcc fails to notice that we don't actually access the array
in that case, so the code is actually safe. Adding an initialization
for the array pointer shuts up the warning.
Fixes: 19ad683abc ("ASoC: dapm: Avoid creating kcontrol for params")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently in codec to codec dai link if there are multiple
params defined then dapm can use created kcontrol to
decide which param to apply at runtime.
However, in case there is only single param configuration
then there is no point in creating the kcontrol and also there
is no point in allocating memory for kcontrol.
In the snd_soc_dapm_new_pcm function, there is memory
allocation happening for kcontrol which is later used
or not used based on num_param. It is better to not
allocate memory when there is only a single configuration.
This change is to remedy that anomaly.
Signed-off-by: anish kumar <yesanishhere@gmail.com>
Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Tested-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Make snd_soc_dapm_add_routes and snd_soc_dapm_del_routes use the
SND_SOC_DAPM_CLASS_RUNTIME dapm subclass to allow dynamic dapm route
updates.
Signed-off-by: Stuart Henderson <stuarth@opensource.cirrus.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
On soc_add_dai(), it uses null_dai_ops if driver doesn't have
its own ops. This means, dai->driver->ops never been NULL.
dai->driver->ops check is not needed.
This patch removes it.
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
refactor snd_soc_dapm_new_pcm to reduce the size of
this function to facilitate further refactoring.
Signed-off-by: anish kumar <yesanishhere@gmail.com>
Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
w_param_text[count] is freed in the wrong error path.
Fix it by shifting the outfree_w_param label.
Signed-off-by: anish kumar <yesanishhere@gmail.com>
Reviewed-by: Charles Keepax <ckeepax@opensource.cirrus.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
commit 66feeec9322132689d42723df2537d60f96f8e44
"RFC: ASoC: dapm: handle probe deferrals"
forgot a to update some two sites where the call
was used. The static codechecks quickly found them.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Fixes: 66feeec93221 ("RFC: ASoC: dapm: handle probe deferrals")
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
This starts to handle probe deferrals on regulators and clocks
on the ASoC DAPM.
I came to this patch after audio stopped working on Ux500 ages
ago and I finally looked into it to see what is wrong. I had
messages like this in the console since a while back:
ab8500-codec.0: ASoC: Failed to request audioclk: -517
ab8500-codec.0: ASoC: Failed to create DAPM control audioclk
ab8500-codec.0: Failed to create new controls -12
snd-soc-mop500.0: ASoC: failed to instantiate card -12
snd-soc-mop500.0: Error: snd_soc_register_card failed (-12)!
snd-soc-mop500: probe of snd-soc-mop500.0 failed with error -12
Apparently because the widget table for the codec looks like
this (sound/soc/codecs/ab8500-codec.c):
static const struct snd_soc_dapm_widget ab8500_dapm_widgets[] = {
/* Clocks */
SND_SOC_DAPM_CLOCK_SUPPLY("audioclk"),
/* Regulators */
SND_SOC_DAPM_REGULATOR_SUPPLY("V-AUD", 0, 0),
SND_SOC_DAPM_REGULATOR_SUPPLY("V-AMIC1", 0, 0),
SND_SOC_DAPM_REGULATOR_SUPPLY("V-AMIC2", 0, 0),
SND_SOC_DAPM_REGULATOR_SUPPLY("V-DMIC", 0, 0),
So when we call snd_soc_register_codec() and any of these widgets
get a deferred probe we do not get an -EPROBE_DEFER (-517) back as
we should and instead we just fail. Apparently the code assumes
that clocks and regulators must be available at this point and
not defer.
After this patch it rather looks like this:
ab8500-codec.0: Failed to create new controls -517
snd-soc-mop500.0: ASoC: failed to instantiate card -517
snd-soc-mop500.0: Error: snd_soc_register_card failed (-517)!
(...)
abx500-clk.0: registered clocks for ab850x
snd-soc-mop500.0: ab8500-codec-dai.0 <-> ux500-msp-i2s.1 mapping ok
snd-soc-mop500.0: ab8500-codec-dai.1 <-> ux500-msp-i2s.3 mapping ok
I'm pretty happy about the patch as it it, but I'm a bit
uncertain on how to proceed: there are a lot of users of the
external functions snd_soc_dapm_new_control() (111 sites)
and that will now return an occassional error pointer, which
is not handled in the calling sites.
I want an indication from the maintainers whether I should just
go in and augment all these call sites, or if deferred probe
is frowned upon when it leads to this much overhead.
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
While DAPM is mono or single channel, its controls can be shared between
widgets, such as sharing one stereo mixer control between the left and
right channel widgets. An example such as the following routes
[Line In Left]----------<Line In Playback Switch>-------[Left Mixer]
^
^ ^ | ^
(inputs) (paths) <shared stereo mixer control> (outputs)
v v | v
v
[Line In Right]---------<Line In Playback Switch>-------[Right Mixer]
where we have separate widgets and paths for the left and right channels
from "Line In" to "Mixer", but a shared stereo mixer control for the
2 paths.
This patch introduces support for such shared mixer controls, allowing
more than 1 path to be attached to a single stereo control, and being
able to control left/right channels independently.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
To support double channel shared controls split across 2 registers, one
for each channel, we must be able to update both registers together.
Add a second set of register fields to struct snd_soc_dapm_update, and
update the DAPM control writeback (put) callbacks to support this.
For codecs that use custom events which call into DAPM to do updates,
also clear struct snd_soc_dapm_update before using it, so the second
set of fields remains clean.
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
Commit d88429a695 ("ASoC: dapm: Add output driver widget") added
the snd_soc_dapm_out_drv ID for the output driver widget, which is
the same as the PGA widget, with a later power sequence number.
Commit 19a2557b76 ("ASoC: dapm: Add kcontrol support for PGAs")
then added kcontrol support for PGA widgets, but failed to account
for output driver widgets. Attempts to use kcontrols with output
driver widgets result in silent failures, with the developer having
little idea about what went on.
Add snd_soc_dapm_out_drv to the switch/case block under snd_soc_dapm_pga
in dapm_create_or_share_kcontrol, since they are essentially the same.
Fixes: 19a2557b76 (ASoC: dapm: Add kcontrol support for PGAs)
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
The value for the second channel in _ENUM_DOUBLE (double channel) MUXs
is not correctly updated, due to using the wrong bit shift.
Use the correct bit shift, so both channels toggle together.
Fixes: 3727b49684 (ASoC: dapm: Consolidate MUXs and value MUXs)
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Reviewed-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
If soc_dapm_read() fails, val will be uninitialized, and bogus values
will be written later:
ret = soc_dapm_read(dapm, reg, &val);
val = (val >> shift) & mask;
However, the compiler does not give a warning. Return on error before
val is really used to avoid this.
This is similar to the commit 6912831623 ("ASoC: dapm: Fix
uninitialized variable in snd_soc_dapm_get_enum_double()")
Fixes: ce0fc93ae5 (ASoC: Add DAPM support at the component level)
Signed-off-by: Chen-Yu Tsai <wens@csie.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
The SND_SOC_DAPM_PRE_PMU case would call startup()/hw_params() that
might access substream->runtime through other functions.
For example:
Unable to handle kernel NULL pointer dereference at virtual address
[....]
PC is at snd_pcm_hw_rule_add+0x24/0x1b0
LR is at snd_pcm_hw_constraint_list+0x20/0x28
[....]
Process arecord (pid: 424, stack limit = 0xffffffc1ecaf0020)
Call trace:
[<ffffffc00086be68>] snd_pcm_hw_rule_add+0x24/0x1b0
[<ffffffc00086c014>] snd_pcm_hw_constraint_list+0x20/0x28
[<ffffffc0008b47a4>] cs53l30_pcm_startup+0x24/0x30
[<ffffffc0008a6260>] snd_soc_dai_link_event+0x290/0x354
[<ffffffc0008a7528>] dapm_seq_check_event.isra.31+0x134/0x2c8
[<ffffffc0008a7768>] dapm_seq_run_coalesced+0x94/0x1c8
[<ffffffc0008a7940>] dapm_seq_run+0xa4/0x404
[<ffffffc0008a8bac>] dapm_power_widgets+0x524/0x984
[<ffffffc0008ab1c4>] snd_soc_dapm_stream_event+0x8c/0xa8
[<ffffffc0008ac7f4>] soc_pcm_prepare+0x10c/0x1ec
[<ffffffc000865b9c>] snd_pcm_do_prepare+0x1c/0x38
[<ffffffc000865600>] snd_pcm_action_single+0x40/0x88
[<ffffffc0008656b8>] snd_pcm_action_nonatomic+0x70/0x90
[<ffffffc000868d28>] snd_pcm_common_ioctl1+0xb6c/0xdd8
[<ffffffc000869508>] snd_pcm_capture_ioctl1+0x200/0x334
[<ffffffc00086a084>] snd_pcm_ioctl_compat+0x648/0x95c
[<ffffffc0001ff4b4>] compat_SyS_ioctl+0xac/0xfc4
[<ffffffc000084cf0>] el0_svc_naked+0x24/0x28
---[ end trace 0dc4f99c2759c35c ]---
So this patch adds a dummy runtime for the original dummy substream
to merely avoid the NULL pointer access.
Signed-off-by: Nicolin Chen <nicoleotsuka@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
This is useful outside the core, when one dapm element is added
at a time.
Signed-off-by: Subhransu S. Prusty <subhransu.s.prusty@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
In is_connected_ep, when custom_stop_condition is true, need to
return the correct paths instead of con which is 0.
Fixes: 6742064aef7f('ASoC: dapm: support user-defined stop condition in dai_get_connected_widgets')
Signed-off-by: Jeeja KP <jeeja.kp@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Certain situations may warrant examining DAPM paths only to a certain
arbitrary point, as opposed to always following them to the end. For
instance, when establishing a connection between a front-end DAI link
and a back-end DAI link in a DPCM path, it does not make sense to walk
the DAPM graph beyond the first widget associated with a back-end link.
This patch introduces a mechanism which lets a user of
dai_get_connected_widgets supply a function which will be called for
every node during the graph walk. When invoked, this function can
execute arbitrary logic to decide whether the walk, given a DAPM widget
and walk direction, should be terminated at that point or continued
as normal.
Signed-off-by: Piotr Stankiewicz <piotrs@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The dummy component is reused for all cards so we special case and don't
bind it to any of them. This means that code like that displaying the
component widgets that tries to look at the card will crash. In the
future we will fix this by ensuring that the dummy component looks like
other components but that is invasive and so not suitable for a fix.
Instead add a special case check here.
Reported-by: Harry Pan <harry.pan@intel.com>
Suggested-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
This patch fixes coccicheck report:
Unneeded variable: "ret". Return "0" on line 2817
Remove unneccesary variable ret created to return zero.
Signed-off-by: Rajan Vaja <rajan.vaja@gmail.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
snd_soc_dapm_dai_link_get() and _put() access the associated ctl
values as value.integer.value[]. However, this is an enum ctl, and it
has to be accessed via value.enumerated.item[]. The former is long
while the latter is unsigned int, so they don't align.
Fixes: c66150824b ('ASoC: dapm: add code to configure dai link parameters')
Cc: <stable@vger.kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
When a DAPM context has a prefix the autodisable widgets get prefixed
twice, once for the control and once for the widget. To avoid this use
the un-prefixed control name to construct the autodisable widget name.
This change is purely cosmetic.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Always on widgets currently have some odd interactions with DAPM.
Enabling/disabling a widget (snd_soc_dapm_enable_pin) then connecting
it to a path works as expected, ie. when the widget is disabled the
path doesn't power up and it does when the widget is enabled. However
once in a path enabling the widget does not cause anything to power
up, dapm_widget_set_power will return the current power state of the
widget as 1, meaning we never check peer power states.
This patch updates dapm_always_on_check_power to return w->connected
such that it is effected by snd_soc_dapm_enable_pin and the like.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
DAPM models various widgets but lacks a sink widget.
DSPs can have modules which take audio data, process it and are
capable of generating events thus acting as a sink of data.
To make the dapm graph complete for such paths we need a dapm
sink widget for these modules, so add a SND_SOC_DAPM_SINK to
declare such a widget. This widget will be treated as
SND_SOC_DAPM_EP_SINK endpoint in the dapm graph
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Reviewed-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently the number of DAI links is statically defined by the machine
driver at build time using an array. This makes it difficult to shrink/
grow the number of DAI links at runtime in order to reflect any changes
in topology.
We can change the DAI link array in the core to a list so that PCMs and
FE DAI links can be added and deleted at runtime to reflect changes in
use case and DSP topology. The machine driver can still register DAI links
as an array.
As the 1st step, this patch change the PCM runtime array to a list. A new
PCM runtime is added to the list when a DAI link is bound successfully.
Later patches will further implement the DAI link list.
More:
- define snd_soc_new/free_pcm_runtime() to create/free a runtime.
- define soc_add_pcm_runtime() to add a runtime to the rtd list.
- define soc_remove_pcm_runtimes() to clean up the runtime list.
- traverse the rtd list to probe the link components and dais.
- Add a field "num" to PCM runtime struct, used to specify the device
number when creating the pcm device, and for a soc card to access
its dai_props array.
- The following 3rd party machine/platform drivers iterate the rtd list
to check the runtimes:
sound/soc/intel/atom/sst-mfld-platform-pcm.c
sound/soc/intel/boards/cht_bsw_rt5645.c
sound/soc/intel/boards/cht_bsw_rt5672.c
sound/soc/intel/boards/cht_bsw_max98090_ti.c
Signed-off-by: Mengdong Lin <mengdong.lin@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
If there is anything in damp->path_source_cache or
damp->path_sink_cache, it can not be valid after the widgets have been
freed. Without this patch a repeated remove and load of a machine
driver may cause NULL pointer reference in dapm_wcache_lookup() when a
freed widget, not belonging to any list, is haunting in the wcache.
Signed-off-by: Jyri Sarha <jsarha@ti.com>
Reported-by: Felipe Balbi <balbi@ti.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Given a kcontrol, we may want to access the parent widget
and it's associated data. So export function to return it.
Signed-off-by: Mythri P K <mythri.p.k@intel.com>
Signed-off-by: Jeeja KP <jeeja.kp@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
For DAI link events, DSPs would like to get notified for startup
and shutdown event as well apart for existing hw_params. This
helps managing DSP resource allocation and freeup on these events
So add support for startup and shutdown for
snd_soc_dai_link_event()
Signed-off-by: Jeeja KP <jeeja.kp@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
For DSPs we can define processing blocks as DAPM PGA widgets.
Some of these proceesing blocks can be configured by usermode
like EQ etc. So we need to add support of kcontrol for PGA
widgets.
Signed-off-by: Jeeja KP <jeeja.kp@intel.com>
Signed-off-by: Mythri P K <mythri.p.k@intel.com>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Incase of an unknown event we were directly returning but we missed
freeing params.
Signed-off-by: Sudip Mukherjee <sudip@vectorindia.org>
Signed-off-by: Mark Brown <broonie@kernel.org>
A number of functions and structures in the sound subsystem had incomplete
and/or obsolete DocBook comments, leading to warnings when the docs were
built. Correct those comments so that we can enjoy our audio in the
absence of warning noise.
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Mark Brown <broonie@kernel.org>
The snd_soc_dapm_input_path and snd_soc_dapm_output_path trace events are
identical except for the direction. Instead of having two events have a
single one that has a field that contains the direction.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
After the recent cleanups and generalizations of the DAPM algorithm the
handling of input and output paths is now fully symmetric. This means by
making some slight changes to the data structure and using arrays with one
entry for each direction, rather than separate fields, it is possible to
create a generic implementation that is capable of handling both input and
output paths.
Unfortunately this generalization significantly increases the code size on
the hot path of is_connected_{input,output}_ep() and
dapm_widget_invalidate_{input,output}_paths(), which has a negative impact
on the overall performance. The inner loops of those functions are quite
small and the generic implementation adds extra pointer arithmetic in a few
places.
Testing on ARM shows that the combined code size of the specialized
functions is about 50% larger than the generalized function in relative
numbers. But in absolute numbers its less than 200 bytes, which is still
quite small. On the other hand the generalized function increases the
execution time of dapm_power_one_widget() by 30%. Given that this function
is one of the most often called functions of the DAPM framework the
trade-off of getting better performance at expense of generating slightly
larger code at seems to be worth it.
To avoid this still keep two versions of these functions around, one for
input and one for output. But have a generic implementation of the
algorithm which gets inlined by those two versions. And then let the
compiler take care of optimizing it and removing he extra instructions.
This still reduces the source code size as well as the makes making changes
to the implementation more straight forward since the same change does no
longer need to be done in two separate places. Also on the slow paths we
can use a generic implementations that handle both input and output paths.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Make sure to unlock the DAPM mutex when dapm_widget_list_create() fails.
This means the function will now generate a trace_snd_soc_dapm_connected
event, even if the creation of the list fails. But that was the behavior
before the patch that introduced the unlock issue, so that should be fine.
Fixes: 1ce43acff0 ("ASoC: dapm: Simplify list creation in dapm_dai_get_connected_widgets()")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Add helper iterator macros for iterating over the source and sink paths of
widget. This will make it easier to change the implementation later on.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
When running dapm_dai_get_connected_widgets() currently in
is_connected_{input,output}_ep() for each widget that gets added the array
is resized and the code also loops over all existing entries to avoid
adding a widget multiple times.
The former can be avoided by collecting the widgets in a linked list and
only once we have all widgets allocate the array.
The later can be avoided by changing when the widget is added. Currently it
is added when walking the neighbor lists of a widget. Since a widget can be
neighbors with multiple other widgets it could get added twice and hence
the check is necessary. But the main body of is_connected_{input,output}_ep
is guaranteed to be only executed at most once per widget. So adding the
widget to the list at the beginning of the function automatically makes
sure that each widget gets only added once. The only difference is that
using this method the starting point itself will also end up on the list,
but it can easily be skipped when creating the array.
Overall this reduces the code size and speeds things slightly up.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
list_first_entry() always returns non NULL and since the code previously
checked that list is not empty it will also be a valid pointer. Furthermore
a path has always a sink or a source widget. So both checks are redundant
and can be removed.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
When creating a new widget from a template the name string of the template
is duplicated for the newly created widget. This is necessary because in
some cases the string might be stored on the stack or other volatile
memory locations.
But most of the time the string is static const data, which means it is
possible to use it directly without having to worry that it might get freed
or changed.
Use kstrdup_const() to handle duplicating the string. This function is
capable of detecting whether a string is immutable and if it is returns the
input without duplicating it. This will slightly reduce the runtime memory
footprint of DAPM and also speed up initialization.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
snd_soc_tplg_widget_remove_all() has a verbatim copy of an older version of
the widget freeing code from dapm_free_widgets(). Add a new helper function
that takes care of freeing a widget and use it in both places.
This removes the duplicated code and also makes sure that future changes to
the widget freeing code only have to be made in one location.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Commit fdb6eb0a12 ("ASoC: dapm: Modify widget stream name according to
prefix") fixed the case where a DAPM route between a DAI widget and a
DAC/ADC/AIF widget with a matching stream name was not created when the
DAPM context was using a prefix.
Unfortunately the patch introduced a few issues on its own like leaking the
dynamically allocated stream name memory and also not checking whether the
allocation succeeded in the first place.
It is also incomplete in that it still does not handle the case where
stream name of the widget is a substring of the stream name of the DAI,
which is explicitly allowed and works fine if no DAPM prefix is used.
Revert the commit and take a slightly different approach to solving the
issue. Instead of comparing the widget's stream name to the name of the DAI
widget compare it to the stream name of the DAI widget. The stream name of
the DAI widget is identical to the name of the DAI widget except that it
wont have the DAPM prefix added. So this approach behaves identical
regardless to whether the DAPM context uses a prefix or not.
We don't have to worry about potentially matching with a widget with the
same stream name, but from a different DAPM context with a different
prefix, since the code already makes sure that both the DAI widget and the
matched widget are from the same DAPM context.
Fixes: fdb6eb0a12 ("ASoC: dapm: Modify widget stream name according to prefix")
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
The name field of the widget template is only used inside
snd_soc_dapm_new_control_unlocked() which allocates a copy for the actual
widget. This means we need to free the name allocated for the template in
dapm_kcontrol_data_alloc() and not the name of the actual widget in
dapm_kcontrol_free(). Otherwise we get a double free on the widget name and
a memory leak on the template name.
Fixes: 773da9b358 ("ASoC: dapm: Append "Autodisable" to autodisable widget names")
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Dummy dai can be used by multiple sound card. But it only belong to one
card's dapm list. If another card use it, there will be dapm_assert_locked
warning.
[ 20.015782] WARNING: CPU: 1 PID: 661 at sound/soc/soc-dapm.c:124 dapm_assert_locked.isra.36+0x4c/0x58()
[ 20.025249] Modules linked in:
[ 20.028349] CPU: 1 PID: 661 Comm: aplay Not tainted 4.1.0-rc6-next-20150605-00004-gaee05d8-dirty #92
[ 20.037528] Hardware name: Freescale i.MX6 Quad/DualLite (Device Tree)
[ 20.044110] Backtrace:
[ 20.046614] [<80012e00>] (dump_backtrace) from [<80012fa0>] (show_stack+0x18/0x1c)
[ 20.054229] r6:809e8060 r5:00000000 r4:00000000 r3:00000000
[ 20.060002] [<80012f88>] (show_stack) from [<807a0f74>] (dump_stack+0x80/0x9c)
[ 20.067293] [<807a0ef4>] (dump_stack) from [<8002b144>] (warn_slowpath_common+0x7c/0xb4)
[ 20.075427] r5:0000007c r4:00000000
[ 20.079065] [<8002b0c8>] (warn_slowpath_common) from [<8002b1a0>] (warn_slowpath_null+0x24/0x2c)
[ 20.087898] r8:00000001 r7:88007c28 r6:ed94a680 r5:809e83e4 r4:ed83d6c0
[ 20.094747] [<8002b17c>] (warn_slowpath_null) from [<8058403c>] (dapm_assert_locked.isra.36+0x4c/0x58)
[ 20.104101] [<80583ff0>] (dapm_assert_locked.isra.36) from [<805842ec>] (dapm_mark_dirty+0x64/0xa4)
[ 20.113165] [<80584288>] (dapm_mark_dirty) from [<805853a8>] (soc_dapm_dai_stream_event.isra.42+0x30/0xc8)
[ 20.122863] r8:ed9b5dbc r7:00000000 r6:00000001 r5:00000001 r4:ed83d6c0
[ 20.129706] [<80585378>] (soc_dapm_dai_stream_event.isra.42) from [<80587e28>] (snd_soc_dapm_stream_event+0x78/0xa0)
[ 20.140264] r5:ee2ee62c r4:00000001
[ 20.143918] [<80587db0>] (snd_soc_dapm_stream_event) from [<8058957c>] (soc_pcm_prepare+0x138/0x21c)
[ 20.153058] r8:ed8d9480 r7:00000000 r6:ed9b0e00 r5:00000001 r4:ee2ee62c r3:00000000
...
This patch is to not probe the dummy component in soc_probe_component. Then
there is no widget created for dummy DAI, and also don't need to check the
dummy dai in dapm_connect_dai_link_widgets().
Signed-off-by: Shengjiu Wang <shengjiu.wang@freescale.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
There is nothing snd_soc_codec specific in this function, just use
snd_soc_component instead.
With this also the last reference of snd_soc_codec form soc-dapm.c.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The DAPM lock must be held when accessing the DAPM graph status through
sysfs or debugfs, otherwise concurrent changes to the graph can result in
undefined behaviour.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
The commit 02aa78abec ("ASoC: DAPM: Add APIs to create individual DAPM
controls.") added locking to the snd_soc_dapm_new_control function but
did not update the call to snd_soc_dapm_new_control in the auto-disable
mux code, this appears to be because the patches were sent at fairly
similar times.
This patch change the call in the auto-disable mux code to use the new
snd_soc_dapm_new_control_unlocked function instead.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Fix the following error:-
All error/warnings (new ones prefixed by >>):
>
> sound/built-in.o: In function `soc_tplg_dapm_widget_create':
> >> :(.text+0x25a90): undefined reference to `snd_soc_dapm_new_control'
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The change fixes the following compilation problem:
sound/soc/soc-dapm.c: In function 'dapm_kcontrol_data_alloc':
sound/soc/soc-dapm.c:388:4: error: implicit declaration of function
'snd_soc_dapm_new_control' [-Werror=implicit-function-declaration]
data->widget = snd_soc_dapm_new_control(widget->dapm,
^
sound/soc/soc-dapm.c:387:17: warning: assignment makes pointer
from integer without a cast [enabled by default]
data->widget = snd_soc_dapm_new_control(widget->dapm,
^
sound/soc/soc-dapm.c: At top level:
sound/soc/soc-dapm.c:3269:1: error: conflicting types for
'snd_soc_dapm_new_control'
snd_soc_dapm_new_control(struct snd_soc_dapm_context *dapm,
^
In addition to the fix add static qualifier to
snd_soc_dapm_new_control() function to silence checkpatch.
Fixes: 02aa78abec ("ASoC: DAPM: Add APIs to create individual DAPM controls.")
Signed-off-by: Vladimir Zapolskiy <vz@mleia.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The topology core needs to be able to create individual widget controls at
runtime and driver init. Add a regular locked and unlocked API calls to
facilitate this requirement.
The unlocked call is used by the topology core during component driver
probing where the card dapm_mutex is held by the ASoC core and the locked
version at non component driver probe time.
Signed-off-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
When there is prefix specified, currently we will add this prefix in
widget->name, but not in widget->sname.
it causes failure at snd_soc_dapm_link_dai_widgets:
if (!w->sname || !strstr(w->sname, dai_w->name))
because dai_w->name has prefix added, but w->sname does not.
We should also add prefix for stream name
Signed-off-by: Koro Chen <koro.chen@mediatek.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Cc: stable@vger.kernel.org
Some CODECs have a significant number of DAPM routes and for each route,
when it is added to the card, the entire card widget list must be
searched. When adding routes it is very likely, however, that adjacent
routes will require adjacent widgets. For example all the routes for a
mux are likely added in a block and the sink widget will be the same
each time and it is also quite likely that the source widgets are
sequential located in the widget list.
This patch adds a cache to the DAPM context, this cache will hold the
source and sink widgets from the last call to snd_soc_dapm_add_route for
that context. A small search of the widget list will be made from those
points for both the sink and source. Currently this search only checks
both the last widget and the one adjacent to it.
On wm8280 which has approximately 500 widgets and 30000 routes (one of
the largest CODECs in mainline), the number of paths that hit the cache
is 24000, which significantly improves probe time.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently snd_soc_dapm_add_route will continue to search the widget list
even after both the source and sink for the route have been located.
This patch breaks out of the search when both are located giving a
small improvement in probe time for drivers.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
The is a missing mutex unlock on the error path in
snd_soc_dapm_get_enum_double. This was introduced in commit
561ed680b7 ("ASoC: dapm: Add support for autodisable mux controls").
This patch adds the missing unlock.
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
A demux is conceptually similar to a mux. Where a mux has multiple input
and one output and selects one of the inputs to be connected to the output,
the demux has one input and multiple outputs and selects one of the outputs
to which the input gets connected.
This similarity makes it straight forward to support them in DAPM using the
existing mux support, we only need to swap sinks and sources when initially
setting up the paths.
The only slightly tricky part is that there can only be one control per
path. Since mixers/muxes are at the sink of a path and a demux is at the
source and both types want a control it is not possible to directly connect
a demux output to a mixer/mux input. The patch adds some sanity checks to
make sure that this does not happen.
Drivers who want to model hardware which directly connects a demux output
to a mixer/mux input can do this by inserting a dummy widget between the
two. E.g.:
{ "Dummy", "Demux Control", "Demux" },
{ "Mixer", "Mixer Control", "Dummy" },
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently new widgets are appended to the beginning of the cards widget
list. This has the effect that widgets that are created while iterating
over the widget list in snd_soc_dapm_new_widgets() (like e.g. the
auto-disable widgets) are not covered during that invocation of the
function. If no further invocations of snd_soc_dapm_new_widgets() happen
these widgets will not be fully initialized and e.g. no debugfs entries are
created for them.
By adding new widgets to the end of the widget list we make sure that
widgets that are created in snd_soc_dapm_new_widgets() will still be
handled during the same snd_soc_dapm_new_widgets() invocation and are
always fully initialized.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Commit 57295073b6 ("ASoC: dapm: Implement mixer input auto-disable")
added support for autodisable controls, controls whose values are only
written to the hardware when their respective widgets are powered up.
But it only added support for controls based on the mixer abstraction.
This patch add support for mux controls (DAPM controls based on the
enum abstraction) to be auto-disabled as well. As each mux can only have
a single control, there is no need to tie the autodisable widget to the
inputs (as is done for the mixer controls) it can be tided directly to
the mux widget itself.
Note that it is assumed that the first entry in a autodisable mux
control will always represent the off state for the mux and is what the
mux will be set to whilst it is disabled.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Reviewed-by: Lars-Peter Clausen <lars@metafoo.de>
Tested-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
This makes it a little easier to follow what is happening in debugfs.
Additionally is also useful in facilitating work to add autodisable
muxes because the control name is already used for the mux widget and
thus shouldn't be reused for the autodisable widget.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Reviewed-by: Lars-Peter Clausen <lars@metafoo.de>
Tested-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The memory subsystem is pretty chatty on failure no need to have local
OOM messages as well.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Reviewed-by: Lars-Peter Clausen <lars@metafoo.de>
Tested-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
All drivers have the same line at the end of the set_bias_level callback to
update the bias_level state. Move this update into
snd_soc_dapm_force_bias_level() and remove them from the drivers.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently drivers are responsible for managing the bias_level field of
their DAPM context. The DAPM state itself is managed by the DAPM core
though and the core has certain expectations on how and when the bias_level
field should be updated. If drivers don't adhere to these undefined
behavior can occur.
This patch adds a few helper functions for manipulating the DAPM context
state, each function with a description on when it should be used and what
its effects are. This will also help us to move more of the bias_level
management from drivers to the DAPM core.
For convenience also add snd_soc_codec_* wrappers around these helpers.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
If the registration of a debugfs directory fails this is treated as a
non-fatal error in ASoC and operation continues as normal. This means we
need to be careful and check if the parent debugfs directory exists if we
try to register a debugfs file or sub-directory. Otherwise we might end up
passing NULL for the parent and the file or directory will be registered in
the top-level debugfs directory.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
commit c66150824b ("ASoC: dapm: add code to configure dai link
parameters") introduced the following build warning:
sound/soc/soc-dapm.c: In function 'snd_soc_dapm_new_pcm':
sound/soc/soc-dapm.c:3389:4: warning: passing argument 1 of 'snprintf'
discards 'const' qualifier from pointer target type
snprintf(w_param_text[count], len,
This patch fixes this by switching to using devm_kasprintf. This also
saves a couple of lines of code.
Signed-off-by: Charles Keepax <ckeepax@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
dai-link params for codec-codec links were fixed. The fixed
link between codec and another chip which may be another codec,
baseband, bluetooth codec etc may require run time configuaration
changes. This change provides an optional alsa control to select
one of the params from a list of params.
Signed-off-by: Nikesh Oswal <nikesh@opensource.wolfsonmicro.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
snd_soc_dapm_kcontrol_codec() is a extremely simple function and inlining it
typically results in less code than necessary for calling the non-inlined
version of the function.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Instead of calling device_create_file() manually, assign the static
attribute group entries at the device registration. This simplifies
the error handling and avoids the possible races.
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
There are no more users of this field left so it can finally be removed.
New users should use snd_soc_dapm_to_codec(w->dapm);
The reason why it is removed is because it doesn't fit to well anymore in
the componentized ASoC hierarchy, where DAPM works on the snd_soc_component
level. And the alternative of snd_soc_dapm_to_codec(w->dapm) typically
generates the same amount of code, so there is really no reason to keep it.
For automatic conversion the following coccinelle semantic patch can be used:
// <smpl>
@@
struct snd_soc_dapm_widget *w;
@@
-w->codec
+snd_soc_dapm_to_codec(w->dapm)
// </smpl>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The only user of the async I/O support in ASoC is SPI which was using it to
avoid needless context thrashing and minimise controller runtime PM bounces.
The SPI framework has now been enhanced so that even normal spi_sync() calls
won't suffer these effects so we don't need to handle this in ASoC and in
fact it can be more efficient not to since we don't need to set up and tear
down the buffers needed to manage asynchronous I/O.
The async completions that DAPM does are left in place so drivers can use
them, they are very cheap if there is no asynchronous work queued.
Signed-off-by: Mark Brown <broonie@kernel.org>
For legacy reasons the ASoC framework assumes that a CODEC INPUT or OUTPUT
widget that is not explicitly connected to a external source or sink is
potentially connected to a source or a sink and hence the framework treats
the widget itself as source (for INPUT) or sink (for OUTPUT). For this
reason a INPUT or OUTPUT widget that is really not connected needs to be
explicitly marked as so.
Setting the card's fully_routed flag will cause the ASoC core, once that all
widgets and routes have been registered, to go through the list of all
widgets and mark all INPUT and OUTPUT that are not externally connected as
non-connected. This essentially negates the default behaviour of treating
INPUT or OUTPUT widgets without external routes as sources or sinks.
This patch takes a different approach while getting the same result. Instead
of first marking INPUT and OUTPUT widgets as sinks/sources and then later
marking them as non-connected, just never mark them as a sink or a source if
the fully_routed flag is set on a card.
This requires a lot less code and also results in a slightly faster card
initialization since there is no need to iterate over all widgets and check
whether the INPUT and OUTPUT widgets are connected or not.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The connected flag of a widget only affects widgets that are either a source
or a sink. The MICBIAS widget is a simple pass-through widget though and
hence its behavior is the same regardless of whether the connected flag is
set or not.
Hence there is not much point in trying to automatically mark MICBIAS
widgets as non-connected, so just remove it.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The card field of the snd_soc_dai field is very rarely used. We can use
dai->component->card instead and remove the card field from the snd_soc_dai
struct.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Currently we cache the number of input and output paths going to/from a
widget only within a power update sequence. But not in between power update
sequences.
But we know how changes to the DAPM graph affect the number of input (form a
source) and output (to a sink) paths of a widget and only need to
recalculate them if a operation has been performed that might have changed
them.
* Adding/removing or connecting/disconnecting a path means that the for
the source of the path the number of output paths can change and for
the sink the number of input paths can change.
* Connecting/disconnecting a widget has the same effect has connecting/
disconnecting all paths of the widget. So for the widget itself the
number of inputs and outputs can change, for all sinks of the widget
the number of inputs can change and for all sources of the widget the
number of outputs can change.
* Activating/Deactivating a stream can either change the number of
outputs on the sources of the widget associated with the stream or the
number of inputs on the sinks.
Instead of always invalidating all cached numbers of input and output paths
for each power up or down sequence this patch restructures the code to only
invalidate the cached numbers when a operation that might change them has
been performed. This can greatly reduce the number of DAPM power checks for
some very common operations.
Since per DAPM operation typically only either change the number of inputs
or outputs the number of path checks is reduced by at least 50%. The number
of neighbor checks is also reduced about the same percentage, but since the
number of neighbors encountered when walking from sink to source is not the
same as when walking from source to sink the actual numbers will slightly
vary from card to card (e.g. for a mixer we see 1 neighbor when walking from
source to sink, but the number of inputs neighbors when walking from source
to sink).
Bigger improvements can be observed for widgets with multiple connected
inputs and output (e.g. mixers probably being the most widespread form of
this). Previously we had to re-calculate the number of inputs and outputs
on all input and output paths. With this change we only have to re-calculate
the number of outputs on the input path that got changed and the number of
inputs on the output paths.
E.g. imagine the following example:
A --> B ----.
v
M --> N --> Z <-- S <-- R
|
v
X
Widget Z has multiple input paths, if any change was made that cause Z to be
marked as dirty the power state of Z has to be re-computed. This requires to
know the number of inputs and outputs of Z, which requires to know the
number of inputs and outputs of all widgets on all paths from or to Z.
Previously this meant re-computing all inputs and outputs of all the path
going into or out of Z. With this patch in place only paths that actually
have changed need to be re-computed.
If the system is idle (or the part of the system affected by the changed
path) the number of path checks drops to either 0 or 1, regardless of how
large or complex the DAPM context is. 0 if there is no connected sink and no
connected source. 1 if there is either a connected source or sink, but not
both. The number of neighbor checks again will scale accordingly and will be
a constant number that is the number of inputs or outputs of the widget for
which we did the path check.
When loading a state file or switching between different profiles typically
multiple mixer and mux settings are changed, so we see the benefit of this
patch multiplied for these kinds of operations.
Testing with the ADAU1761 shows the following changes in DAPM stats for
changing a single Mixer switch for a Mixer with 5 inputs while the DAPM
context is idle.
Power Path Neighbour
Before: 2 12 30
After: 2 1 2
For the same switch, but with a active playback stream the stat changed are
as follows.
Power Path Neighbour
Before: 10 20 54
After: 10 7 21
Cumulative numbers for switching the audio profile which changes 7 controls
while the system is idle:
Power Path Neighbour
Before: 16 80 170
After: 16 7 23
Cumulative numbers for switching the audio profile which changes 7 controls
while playback is active:
Power Path Neighbour
Before: 51 123 273
After: 51 29 109
Starting (or stopping) the playback stream:
Power Path Neighbour
Before: 34 34 117
After: 34 17 69
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Supply widgets are somewhat special and not all kinds of paths to or from
supply widgets make sense. This patch adds a few sanity checks that errors
out during the path instantiation for those invalid paths. This will prevent
drivers to depend on weird behavior resulting from such paths as well as
will allow the DAPM algorithms to assume that they never see such paths.
This patch adds checks for the following three invalid types of paths:
* A path with a non-supply widget as a source connected to a supply
widget as a sink. Such a path has no effect on either of the two
connected widgets.
* Paths with a connected() callback that have a non-supply widget as the
source. The DAPM algorithm only uses the conneceted() callback for
supply widget power checks. And since it prevents caching of the DAPM
state there is no intention to make it more generic as it has
negative performance implications.
* Paths which connect a supply to a mixer or mux via a control. Controls
are only meant to affect the routing of audio data.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The state of endpoint widgets is affected by that card's power state.
Endpoint widgets that do no have the ignore_suspend flag set will be
considered inactive during suspend. So they have to be re-checked and marked
dirty after the card's power state changes. Currently the input and output
widgets are marked dirty instead, this works most of the time since
typically a path from one endpoint to another will go via a input or output
widget. But marking the endpoints dirty is technically more correct and will
also work for odd corner cases.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Supply widgets do not count towards the input and output widgets of their
neighbors and for supply widgets themselves we do not care for the number
of input or output paths. This means that a path that connects to a supply
widget effectively behaves the same as a path that as the weak property set.
This patch adds a new path flag that gets set to true when the path is
connected to at least one supply widget. If a path with the flag set is
encountered in is_connected_{input,output}_ep() is is skipped in the same
way that weak paths are skipped. This slightly brings down the number of
path checks.
Since both the weak and the supply flag are implemented as bitfields which
are stored in the same word there is no runtime overhead due to checking
both rather than just one and also the size of the path struct is not
increased by this patch. Another advantage is that we do not have to handle
supply widgets in is_connected_{input,output}_ep() anymore since it will
never be called for supply widgets. The only exception is from
dapm_widget_power_read_file() where a check is added to special case supply
widgets.
Testing with the ADAU1761, which has a handful of supply widgets, shows the
following changes in the DAPM stats for a playback stream start.
Power Path Neighbour
Before: 34 78 117
After: 34 48 117
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
DAPM widgets can be classified into four categories:
* supply: Supply widgets do not affect the power state of their
non-supply widget neighbors and unlike other widgets a
supply widget is not powered up when it is on an active
path, but when at least on of its neighbors is powered up.
* source: A source is a widget that receives data from outside the
DAPM graph or generates data. This can for example be a
microphone, the playback DMA or a signal generator. A source
widget will be considered powered up if there is an active
path to a sink widget.
* sink: A sink is a widget that transmits data to somewhere outside
of the DAPM graph. This can e.g. be a speaker or the capture
DMA. A sink widget will be considered powered up if there is
an active path from a source widget.
* normal: Normal widgets are widgets not covered by the categories
above. A normal widget will be considered powered up if it
is on an active path between a source widget and a sink
widget.
The way the number of input and output paths for a widget is calculated
depends on its category. There are a bunch of factors which decide which
category a widget is. Currently there is no formal classification of these
categories and we calculate the category of the widget based on these
factors whenever we want to know it. This is at least once for every widget
during each power update sequence. The factors which determine the category
of the widgets are mostly static though and if at all change rather seldom.
This patch introduces three new per widget flags, one for each of non-normal
widgets categories. Instead of re-computing the category each time we want
to know them the flags will be checked. For the majority of widgets the
category is solely determined by the widget id, which means it never changes
and only has to be set once when the widget is created. The only widgets
with dynamic categories are:
snd_soc_dapm_dai_out: Is considered a sink iff the capture stream is
active, otherwise normal.
snd_soc_dapm_dai_in: Is considered a source iff the playback stream
is active, otherwise normal.
snd_soc_dapm_input: Is considered a sink iff it has no outgoing
paths, otherwise normal.
snd_soc_dapm_output: Is considered a source iff it has no incoming
paths, otherwise normal.
snd_soc_dapm_line: Is considered a sink iff it has no outgoing paths
and is considered a source iff it has no incoming paths,
otherwise normal.
For snd_soc_dapm_dai_out/snd_soc_dapm_dai_in widgets the category will be
updated when a stream is started or stopped. For the other dynamic widgets
the category will be updated when a path connecting to it is added or
removed.
Introducing those new widget categories allows to make
is_connected_{output,input}_ep, which are among the hottest paths of the
DAPM algorithm, more generic and significantly shorter.
The before and after sizes for is_connected_{output,input}_ep are:
On ARM (defconfig + CONFIG_SND_SOC):
function old new delta
is_connected_output_ep 480 340 -140
is_connected_input_ep 456 352 -104
On amd64 (defconfig + CONFIG_SND_SOC):
function old new delta
is_connected_output_ep 579 427 -152
is_connected_input_ep 563 427 -136
Which is about a 25%-30% decrease, other architectures are expected to have
similar numbers. At the same time the size of the snd_soc_dapm_widget struct
does not change since the new flags are stored in the same word as the
existing flags.
Note: that since the per widget 'ext' flag was only used to decide whether a
snd_soc_dapm_input or snd_soc_dapm_output widget was a source or a sink it
is now unused and can be removed.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Controls on a path only have an effect if the sink on the path is either a
mixer or mux widget. Currently we sort of silently ignore controls on other
paths, but since they don't do anything having them on other paths does not
make much sense and it is probably safe to assume that if we see such a path
it is a mistake in the driver that registered the path. This patch modifies
snd_soc_dapm_add_path() to report an error if a path with and control is
encountered where we didn't expect a control. This also allows to simplify
the code quite a bit.
The patch also moves the connecting of the path lists out of
dapm_connect_mux() and dapm_connect_mixer() into snd_soc_dapm_add_path().
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Paths that are directly connected to a MUX widget are not affected by
changes to the MUX's control. Rather than checking if a path is directly
connected each time the MUX is updated do it only once when MUX is created.
We can also remove the check for e->texts[mux] != NULL, since if that
condition was true the code would have had already crashed much earlier (And
generally speaking if a enum's 'texts' entry is NULL it's a bug in the
driver).
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Rework soc_dapm_{mixer,mux}_update_power() to only mark a path dirty if the
connect state if the path has actually changed. This avoids unnecessary
power state checks for the widgets involved.
Also factor out the common code that is involved in this into a helper
function.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
The 'walked' flag was used to avoid walking paths that have already been
walked. But since we started caching the number of inputs and outputs of a
path we never actually get into a situation where we try to walk a path that
has the 'walked' flag set.
There are two cases in which we can end up walking a path multiple times
within a single run of is_connected_output_ep() or is_connected_input_ep().
1) If a path splits up and rejoins later:
.--> C ---v
A -> B E --> F
'--> D ---^
When walking from A to F we'll end up at E twice, once via C and once via D.
But since we do a depth first search we'll fully discover the path and
initialize the number of outputs/inputs of the widget the first time we get
there. The second time we get there we'll use the cached value and not
bother to check any of the paths again. So we'll never see a path where
'walked' is set in this case.
2) If there is a circle:
A --> B <-- C <-.--> F
'--> D ---'
When walking from A to F we'll end up twice at B. But since there is a
circle the 'walking' flag will still be set on B once we get there the
second time. This means we won't look at any of it's outgoing paths. So in
this case we won't ever see a path where 'walked' is set either.
So it is safe to remove the flag. This on one hand means we remove some
always true checks from one of the hottest paths of the DAPM algorithm and
on the other hand means we do not have to do the tedious clearing of the
flag after checking the number inputs or outputs of a widget.
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
dapm_adc_check_power() checks if the widget is active, if yes it only checks
whether there are any connected input paths. Otherwise it calls
dapm_generic_check_power() which will check for both connected input and
output paths. But the function that checks for connected output paths will
return true if the widget is a active sink. Which means the generic power
check function will work just fine and there is no need for a special power
check function.
The same applies for dapm_dac_check_power(), but with input and output paths
reversed.
This patch removes both dapm_adc_check_power() and dapm_dac_check_power()
and replace their usage with dapm_generic_check_power().
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@kernel.org>
Linus Torvalds