Move the requesting of DMA channels further down from 'probe' to
'pcm_new'. This is to spare the allocated DMA channel resources as we
typically only ever use one or two of the clusters for PCM streaming.
Before we would request DMA channels for all clusters.
(This is prompted by a change in the Audio DMA Controller driver, which
will now be allocating cache SRAM to channels.)
Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
Link: https://lore.kernel.org/r/20220905074030.1293-1-povik+lin@cutebit.org
Signed-off-by: Mark Brown <broonie@kernel.org>
The MCA driver selects COMMON_CLK, which leads to misconfiguration
on platforms with HAVE_LEGACY_CLK (under compile test). Kconfig catches
the conflict with the following warning:
WARNING: unmet direct dependencies detected for COMMON_CLK
Depends on [n]: !HAVE_LEGACY_CLK [=y]
Selected by [m]:
- SND_SOC_APPLE_MCA [=m] && SOUND [=m] && !UML && SND [=m] &&
SND_SOC [=m] && (ARCH_APPLE || COMPILE_TEST [=y])
Eventually the build fails with errors like:
>> drivers/clk/clk.c:867:6: error: redefinition of 'clk_unprepare'
867 | void clk_unprepare(struct clk *clk)
| ^~~~~~~~~~~~~
In file included from drivers/clk/clk.c:9:
include/linux/clk.h:303:20: note: previous definition of 'clk_unprepare' with type 'void(struct clk *)'
303 | static inline void clk_unprepare(struct clk *clk)
| ^~~~~~~~~~~~~
which appears to be because COMMON_CLK is selected but HAVE_CLK_PREPARE
is not. In the end it seems we had no business selecting COMMON_CLK from
an unrelated driver like that, so remove the selection. The linux/clk.h
API is there anyway.
Fixes: 3df5d0d972 ("ASoC: apple: mca: Start new platform driver")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
Link: https://lore.kernel.org/r/20220901113415.27449-1-povik+lin@cutebit.org
Signed-off-by: Mark Brown <broonie@kernel.org>
In DAI ops, accesses to the native cluster (of the DAI), and to data of
clusters related to it by a DPCM frontend-backend link, should have
been synchronized by the 'pcm_mutex' lock at ASoC level.
What is not covered are the 'port_driver' accesses on foreign clusters
to which the current cluster has no a priori relation, so fill in
locking for that. (This should only matter in bizarre configurations of
sharing one MCA peripheral between ASoC cards.)
Signed-off-by: Martin Povišer <povik+lin@cutebit.org>
Link: https://lore.kernel.org/r/20220824160715.95779-5-povik+lin@cutebit.org
Signed-off-by: Mark Brown <broonie@kernel.org>
