Add helpers to program SPA/CPA bits, using a mutex to access the
shared LCTL register if required.
All links are managed with the same LCTLx.SPA bits. However there are
quite a few implementation details to be aware of:
Legacy HDaudio multi-links are powered-up when exiting reset, which
requires the ref_count to be manually set to one when initializing the
link.
Alternate links for SoundWire/DMIC/SSP need to be explicitly
powered-up before accessing the SHIM/IP/Vendor-Specific SHIM space for
each sublink. DMIC/SSP/SoundWire are all different cases with a
different device/dai/hlink relationship.
SoundWire will handle power management with the auxiliary device
resume/suspend routine. The ref_count is not necessary in this case.
The DMIC/SSP will by contrast handle the power management from DAI
.startup and .shutdown callbacks.
The SSP has a 1:1 mapping between sublink and DAI, but it's
bidirectional so the ref_count will help avoid turning off the sublink
when one of the two directions is still in use.
The DMIC has a single link but two DAIs for data generated at
different sampling frequencies, again the ref_count will make sure the
two DAIs can be used concurrently.
And last the SoundWire Intel require power-up/down and bank switch to
be handled with a lock already taken, so the 'eml_lock' is made
optional with the _unlocked versions of the helpers.
Note that the _check_power_active() implementation is similar to
previous helpers in sound/hda/ext, with sleep duration and timeout
aligned with hardware recommendations. If desired, this helper could
be modified in a second step with .e.g. readl_poll_timeout()
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20230404104127.5629-9-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Extend hdac_ext_link to store information needed for ALT
links. Follow-up patches will include more functional patches for
power-up and down.
Note that this patch suggests the use of an 'eml_lock' to serialize
access to shared registers. SoundWire-specific sequence require the
lock to be taken at a higher level, as a result the helpers added in
follow-up patches will provide 'unlocked' versions when needed.
Also note that the low-level sequences with the 'hdaml_' prefix are
taken directly from the hardware specifications - naming conventions
included. The code will be split in two, with locking and linked-list
management handled separately to avoid mixing required hardware setup
and Linux-based resource management.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20230404104127.5629-7-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from Srinivas Kandagatla <srinivas.kandagatla@linaro.org>:
Click/Pop Noise was a long pending issue with WSA Codecs which are prone
to accumlate DC when ports are active but without any data streams.
There are multiple places in the current setup, where this could happen
in both startup as well as shutdown path.
Smatch Warns:
sound/firewire/tascam/tascam-stream.c:493 snd_tscm_stream_start_duplex()
warn: missing unwind goto?
The direct return will cause the stream list of "&tscm->domain" unemptied
and the session in "tscm" unfinished if amdtp_domain_start() returns with
an error.
Fix this by changing the direct return to a goto which will empty the
stream list of "&tscm->domain" and finish the session in "tscm".
The snd_tscm_stream_start_duplex() function is called in the prepare
callback of PCM. According to "ALSA Kernel API Documentation", the prepare
callback of PCM will be called many times at each setup. So, if the
"&d->streams" list is not emptied, when the prepare callback is called
next time, snd_tscm_stream_start_duplex() will receive -EBUSY from
amdtp_domain_add_stream() that tries to add an existing stream to the
domain. The error handling code after the "error" label will be executed
in this case, and the "&d->streams" list will be emptied. So not emptying
the "&d->streams" list will not cause an issue. But it is more efficient
and readable to empty it on the first error by changing the direct return
to a goto statement.
The session in "tscm" has been begun before amdtp_domain_start(), so it
needs to be finished when amdtp_domain_start() fails.
Fixes: c281d46a51 ("ALSA: firewire-tascam: support AMDTP domain")
Signed-off-by: Xu Biang <xubiang@hust.edu.cn>
Reviewed-by: Dan Carpenter <error27@gmail.com>
Acked-by: Takashi Sakamoto <o-takashi@sakamocchi.jp>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20230406132801.105108-1-xubiang@hust.edu.cn
Signed-off-by: Takashi Iwai <tiwai@suse.de>
dapm_connect_dai_pair() handles
"Normal/Codec2Codec" x "CPU/Codec" x "Playback/Capture".
(A) is "Codec2Codec" case of "CPU" widget x "Playback/Capture",
(B) is "Normal" case of "CPU" widget x "Playback/Capture",
(C) is each case of "Codec" widget.
(X) is handling "Playback" case DAI connecting,
(Y) is handling "Capture" case DAI connecting.
static void dapm_connect_dai_pair(...)
{
...
(A) if (dai_link->params) {
playback_cpu = ...
capture_cpu = ...
(B) } else {
playback_cpu = ...
capture_cpu = ...
}
^ /* connect BE DAI playback if widgets are valid */
| stream = SNDRV_PCM_STREAM_PLAYBACK;
| (C) codec = codec_dai->playback_widget;
|
| if (playback_cpu && codec) {
(X) if (dai_link->params && !rtd->c2c_widget[stream]) {
| ...
| }
|
| (z) dapm_connect_dai_routes(...);
v }
capture:
^ /* connect BE DAI capture if widgets are valid */
| stream = SNDRV_PCM_STREAM_CAPTURE;
| (C) codec = codec_dai->capture_widget;
|
| if (codec && capture_cpu) {
(Y) if (dai_link->params && !rtd->c2c_widget[stream]) {
| ...
| }
|
| (z) dapm_connect_dai_routes(...);
v }
}
(X) part and (Y) part are almost same.
Main purpose of these parts (and this function) is calling
dapm_connect_dai_routes() (= z) on each cases.
The difference is "parameter"
(= Normal/Codec2Codec x CPU/Codec x Playback/Capture).
This patch cleanup these, but nothing changed for meaning.
Link: https://lore.kernel.org/r/87ilen6ni4.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/877cuqvswc.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
It could have never worked, as snd_emu10k1_fx8010_playback_prepare() and
snd_emu10k1_fx8010_playback_hw_free() assume the emu10k1 offset for the
ETRAM, and the default DSP code includes no handler for it. It also
wouldn't make a lot of sense to make it work, as Audigy has an own, much
simpler, pass-through mechanism. So just skip creation of the device.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20230405201220.2197938-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Due to two copy/pastos, closing the MIC or EFX capture device would
make a running ADC capture hang due to unsetting its interrupt handler.
In principle, this would have also allowed dereferencing dangling
pointers, but we're actually rather thorough at disabling and flushing
the ints.
While it may sound like one, this actually wasn't a hypothetical bug:
PortAudio will open a capture stream at startup (and close it right
away) even if not asked to. If the first device is busy, it will just
proceed with the next one ... thus killing a concurrent capture.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20230405201220.2197923-1-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
The BIOS botches this one completely - it says the 2nd S/PDIF output is
used, while in fact it's the 1st one. This is tested on DP45SG, but I'm
assuming it's valid for the other boards in the series as well.
Also add some comments regarding the pins.
FWIW, the codec is apparently still sold by Tempo Semiconductor, Inc.,
where one can download the documentation.
Signed-off-by: Oswald Buddenhagen <oswald.buddenhagen@gmx.de>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20230405201220.2197826-2-oswald.buddenhagen@gmx.de
Signed-off-by: Takashi Iwai <tiwai@suse.de>
With PCI if the device was suspended it is brought back to full
power and then suspended again.
This doesn't happen when device is described via DT.
We need to make sure that we tear down pipelines only if the device
was previously active (thus the pipelines were setup).
Otherwise, we can break the use_count:
[ 219.009743] sof-audio-of-imx8m 3b6e8000.dsp:
sof_ipc3_tear_down_all_pipelines: widget PIPELINE.2.SAI3.IN is still in use: count -1
and after this everything stops working.
Fixes: d185e0689a ("ASoC: SOF: pm: Always tear down pipelines before DSP suspend")
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Daniel Baluta <daniel.baluta@nxp.com>
Link: https://lore.kernel.org/r/20230405092655.19587-1-daniel.baluta@oss.nxp.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Merge series from Peter Ujfalusi <peter.ujfalusi@linux.intel.com>:
This series will add support for SOF Linux stack to run without using the DSP.
DSPless mode provides a good tool for verification that the hardware itself
works correctly by taking the DSP use out from the picture.
It can only work with interfaces which supports this mode: Intel HDA at the
moment but with LNL it could be possible to support other audio interfaces.
The main driver for this mode is to be able to test programming sequences,
low-level code and for low-level verification of a platform.
The feature is not targetted for end-users and it will not make the SOF stack
to work on hardware without DSP, but it is giving us a tool to debug and enable
platforms earlier (when for example t he firmware is not mature enough).
Smatch Warns:
sound/soc/tegra/tegra20_ac97.c:321 tegra20_ac97_platform_probe()
warn: missing unwind goto?
The goto will set the "soc_ac97_ops" and "soc_ac97_bus" operations to
NULL. But they are already NULL at this point so it is a no-op.
However, just for consistency, change the direct return to a goto. No
functional change.
Signed-off-by: Zihao Wang <u202012060@hust.edu.cn>
Reviewed-by: Dan Carpenter <error27@gmail.com>
Link: https://lore.kernel.org/r/20230404084622.1202-1-u202012060@hust.edu.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
When the DSP is disabled in the BIOS, the DSP_BAR and PP_BAR cannot be
accessed.
One possible objection noted in initial reviews is that this patch
adds a number of branches. However the number of branches is actually
limited in probe/suspend/resume routines mostly, so there isn't really
a degradation in terms of readability and maintainability. Adding yet
another level of abstraction/ops/callbacks would increase complexity
and not really help in terms of code reuse or readability and
maintainability. A split between controller and DSP driver would be
even more invasive.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Signed-off-by: Peter Ujfalusi <peter.ujfalusi@linux.intel.com>
Link: https://lore.kernel.org/r/20230404092115.27949-7-peter.ujfalusi@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
ACP SOF driver supports different audio configurations.
Explicit condition check for I2S configuration will break
other audio endpoint configurations.
acp_dai_probe() function is not required as we have
machine select logic to select the exact machine.
Remove acp_dai_probe() from existing AMD PCI driver code base.
Signed-off-by: Vijendar Mukunda <Vijendar.Mukunda@amd.com>
Link: https://lore.kernel.org/r/20230403071651.919027-2-Vijendar.Mukunda@amd.com
Signed-off-by: Mark Brown <broonie@kernel.org>
