Introduce for_each_rtd_dais_rollback macro which behaves exactly like
for_each_codec_dais_rollback and its cpu_dais equivalent but for all
dais instead.

Use newly added macro to fix soc_pcm_open error path and prevent
uninitialized dais from being cleaned-up.

Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Fixes: 5d9fa03e6c35 ("ASoC: soc-pcm: tidyup soc_pcm_open() order")
Acked-by: Liam Girdwood <liam.r.girdwood@linux.intel.com>
Acked-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/20200907111939.16169-1-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>

commit 25612477d20b52 ("ASoC: soc-dai: set dai_link dpcm_ flags with a helper")
added snd_soc_dai_link_set_capabilities().
But it is using snd_soc_find_dai() (A) which is required client_mutex (B).
And client_mutex is soc-core.c local.

	struct snd_soc_dai *snd_soc_find_dai(xxx)
	{
		...
(B)		lockdep_assert_held(&client_mutex);
		...
	}

	void snd_soc_dai_link_set_capabilities(xxx)
	{
		...
		for_each_pcm_streams(direction) {
			...
			for_each_link_cpus(dai_link, i, cpu) {
(A)				dai = snd_soc_find_dai(cpu);
				...
			}
			...
			for_each_link_codecs(dai_link, i, codec) {
(A)				dai = snd_soc_find_dai(codec);
				...
			}
		}
		...
	}

Because of these background, we will get WARNING if .config has CONFIG_LOCKDEP.

	WARNING: CPU: 2 PID: 53 at sound/soc/soc-core.c:814 snd_soc_find_dai+0xf8/0x100
	CPU: 2 PID: 53 Comm: kworker/2:1 Not tainted 5.7.0-rc1+ #328
	Hardware name: Renesas H3ULCB Kingfisher board based on r8a77951 (DT)
	Workqueue: events deferred_probe_work_func
	pstate: 60000005 (nZCv daif -PAN -UAO)
	pc : snd_soc_find_dai+0xf8/0x100
	lr : snd_soc_find_dai+0xf4/0x100
	...
	Call trace:
	 snd_soc_find_dai+0xf8/0x100
	 snd_soc_dai_link_set_capabilities+0xa0/0x16c
	 graph_dai_link_of_dpcm+0x390/0x3c0
	 graph_for_each_link+0x134/0x200
	 graph_probe+0x144/0x230
	 platform_drv_probe+0x5c/0xb0
	 really_probe+0xe4/0x430
	 driver_probe_device+0x60/0xf4

snd_soc_find_dai() will be used from (X) CPU/Codec/Platform driver with
mutex lock, and (Y) Card driver without mutex lock.
This snd_soc_dai_link_set_capabilities() is for Card driver,
this means called without mutex.
This patch adds snd_soc_find_dai_with_mutex() to solve it.

Fixes: 25612477d20b52 ("ASoC: soc-dai: set dai_link dpcm_ flags with a helper")
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87blixvuab.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>

Modify snd_soc_add_component so it calls snd_soc_component_initialize
no longer and thus providing true two-step registration. Drivers may
choose to change component's fields before actually adding it to ASoC
subsystem.

Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20200731144146.6678-4-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>

Move 'name' field initialization responsibility back to
snd_soc_component_initialize to prepare snd_soc_add_component function
for being called separatelly as a second registration step.

Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20200731144146.6678-3-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>

To allow for two-step component registration, expose
snd_soc_component_initialize function and move it back to soc-core.c.

Signed-off-by: Cezary Rojewski <cezary.rojewski@intel.com>
Link: https://lore.kernel.org/r/20200731144146.6678-2-cezary.rojewski@intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>

Current soc-xxx are getting rtd from substream by

	rtd = substream->private_data;

But, getting data from "private_data" is very unclear.
This patch adds asoc_substream_to_rtd() macro which is
easy to understand that rtd from substream.

Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87wo2z0yve.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>

While experimenting and introducing errors in Baytrail topology files
until I got them right, I encountered multiple kernel oopses and
memory leaks. This is a first batch to harden the code, but we should
probably think of a tool to fuzz the topology...

Pierre-Louis Bossart (5):
  ASoC: topology: fix kernel oops on route addition error
  ASoC: topology: fix tlvs in error handling for widget_dmixer
  ASoC: topology: use break on errors, not continue
  ASoC: topology: factor kfree(se) in error handling
  ASoC: topology: add more logs when topology load fails.

 sound/soc/soc-topology.c | 97 ++++++++++++++++++++++++----------------
 1 file changed, 58 insertions(+), 39 deletions(-)

base-commit: a5911ac5790acaf98c929b826b3f7b4a438f9759
--
2.25.1

The ASoC devm_ functions that register a component
(devm_snd_soc_register_component and devm_snd_dmaengine_pcm_register) will
clean their component by running snd_soc_unregister_component.

snd_soc_unregister_component will then remove all the components for the
device that was used to register the component in the first place.

However, some drivers register several components (such as a DAI and a
dmaengine PCM) on the same device, and if the dmaengine PCM is registered
first, then the DAI will be cleaned up first and
snd_dmaengine_pcm_unregister will be called next.

snd_dmaengine_pcm_unregister will then lookup the dmaengine PCM component
on the device, and if there's one unregister that component and release its
dmaengine channels. That doesn't happen in practice though since the first
call to snd_soc_unregister_component removed all the components, so we
never get the chance to release the dmaengine channels.

In order to fix this, instead of removing all the components for a given
device, we can simply remove the component that was registered in the first
place. We should have the same number of component registration than we
have components, so it should work just fine.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://lore.kernel.org/r/20200707074237.287171-1-maxime@cerno.tech
Signed-off-by: Mark Brown <broonie@kernel.org>

To fix compilation error:

- error: field 'XXX' has incomplete type

Moves definition of enum snd_soc_bias_level from soc.h to soc-dapm.h.

Signed-off-by: Tzung-Bi Shih <tzungbi@google.com>
Link: https://lore.kernel.org/r/20200625153543.85039-2-tzungbi@google.com
Signed-off-by: Mark Brown <broonie@kernel.org>

Some machine drivers allocate or request resources with
snd_soc_link_init() phase of the card probe. These resources need to
be properly released when removing a card, and this patch suggests a
dual exit() callback.

The exit() is invoked in soc_remove_pcm_runtime(), which is not
completely symmetric with the init() invoked in soc_init_pcm_runtime().

Alternate solutions were considered, e.g. adding a .remove() callback
for the platform driver, but that's not symmetrical at all and would
be difficult to handle if there are more than one dailink implementing
an .init(). We looked also into using .remove_dai_link() callback, but
that would also be imbalanced.

Note that because of the error handling in snd_soc_bind_card(), which
jumps to probe_end, there is no way to guarantee the exit() is invoked
with resources allocated in the init(). Prior to releasing those
resources, implementations of the exit() callback shall check the
resources are valid.

Suggested-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Guennadi Liakhovetski <guennadi.liakhovetski@linux.intel.com>
Reviewed-by: Curtis Malainey <curtis@malainey.com>
Link: https://lore.kernel.org/r/20200622154241.29053-2-pierre-louis.bossart@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>