With PREEMPT_RT enabled a spinlock_t becomes a sleeping lock.
This is usually not a problem with spinlocks used in IRQ context since
IRQ handlers get threaded. However, if IRQF_ONESHOT is set, the primary
handler won't be force-threaded and runs always in hardirq context. This is
a problem because spinlock_t requires a preemptible context on PREEMPT_RT.
In this particular instance, regmap mmio uses spinlock_t to protect the
register access and IRQF_ONESHOT is set on the IRQ. In this case, it is
actually better to do everything in threaded handler and it solves the
problem with PREEMPT_RT.
Reported-by: Arseniy Krasnov <avkrasnov@salutedevices.com>
Closes: https://lore.kernel.org/linux-amlogic/20240729131652.3012327-1-avkrasnov@salutedevices.com
Suggested-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Fixes: b11d26660d ("ASoC: meson: axg-fifo: use threaded irq to check periods")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Reviewed-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Link: https://patch.msgid.link/20240807162705.4024136-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This is a very big update, in large part due to extensive work the Intel
people have been doing in their drivers though it's also been busy
elsewhere. There's also a big overhaul of the DAPM documentation from
Luca Ceresoli arising from the work he did putting together his recent
ELC talk, and he also contributed a new tool for visualising the DAPM
state.
- A new tool dapm-graph for visualising the DAPM state.
- Substantial fixes and clarifications for the DAPM documentation.
- Very large updates throughout the Intel audio drivers.
- Cleanups of accessors for driver data, module labelling, and for
constification.
- Modernsation and cleanup work in the Mediatek drivers.
- Several fixes and features for the DaVinci I2S driver.
- New drivers for several AMD and Intel platforms, Nuvoton NAU8325,
Rockchip RK3308 and Texas Instruments PCM6240.
-----BEGIN PGP SIGNATURE-----
iQEzBAABCgAdFiEEreZoqmdXGLWf4p/qJNaLcl1Uh9AFAmZB2aoACgkQJNaLcl1U
h9D9YQf+K66pFTAMro/X4KWXfg3EeHnqgwbdb1pN/3zKqFgnxPfxJvhvpeM+bFSj
yZGssZZzd9jMkm6rLOZRPCycqSZtimy9DjSnNyPhQgU0jA2ZS1NRMnpN77ubMkPW
IBeWO8j5TBYaqttlmM0YBscErng9GsNqOD5a+HW9AJz5+TYbIIWt/2TVnyBRU0LV
NGkKj1x7AvJY239kitJ4cfFGZpPaGU7bxUk1HCpSPWM+asIpNSxBhKD73zZlWHZ/
kwJSbFslXdCm/TZZQtox/Z8GClxQ2dasWEfXWZK9kBJbHD3UH7VJ4Em50pBWnKis
piarddgSvu5zhVmZkhJhkmOe8jiswA==
=N5Dj
-----END PGP SIGNATURE-----
Merge tag 'asoc-v6.10' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie/sound into for-linus
ASoC: Updates for v6.10
This is a very big update, in large part due to extensive work the Intel
people have been doing in their drivers though it's also been busy
elsewhere. There's also a big overhaul of the DAPM documentation from
Luca Ceresoli arising from the work he did putting together his recent
ELC talk, and he also contributed a new tool for visualising the DAPM
state.
- A new tool dapm-graph for visualising the DAPM state.
- Substantial fixes and clarifications for the DAPM documentation.
- Very large updates throughout the Intel audio drivers.
- Cleanups of accessors for driver data, module labelling, and for
constification.
- Modernsation and cleanup work in the Mediatek drivers.
- Several fixes and features for the DaVinci I2S driver.
- New drivers for several AMD and Intel platforms, Nuvoton NAU8325,
Rockchip RK3308 and Texas Instruments PCM6240.
Merge series from Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>:
Static 'struct snd_pcm_hardware' is not modified by few drivers and its
copy is passed to the core, so it can be made const for increased code
safety.
With the AXG audio subsystem, there is a possible random channel shift on
TDM capture, when the slot number per lane is more than 2, and there is
more than one lane used.
The problem has been there since the introduction of the axg audio support
but such scenario is pretty uncommon. This is why there is no loud
complains about the problem.
Solving the problem require to make the links non-atomic and use the
trigger() callback to start FEs and BEs in the appropriate order.
This was tried in the past and reverted because it caused the block irq to
sleep while atomic. However, instead of reverting, the solution is to call
snd_pcm_period_elapsed() in a non atomic context.
Use the bottom half of a threaded IRQ to do so.
Fixes: 6dc4fa179f ("ASoC: meson: add axg fifo base driver")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20240426152946.3078805-2-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
ASoC is now unified asoc_xxx() into snd_soc_xxx().
This patch convert asoc_xxx() to snd_soc_xxx().
Signed-off-by: Kuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87pm2oqnik.wl-kuninori.morimoto.gx@renesas.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The TDM HW on the axg SoC families and derivatives actually supports
384kHz sampling rate.
Update the fifo and tdm interface constraints accordingly.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20230907090910.13546-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On the AXG family, the fifo irq is not necessary for the HW to operate.
It is just used to notify that a period has elapsed. If userpace does not
care for these wakeups (such as pipewire), we are just wasting CPU cycles.
Add support for NO_PERIOD_WAKEUP and disable irq when they are no needed.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20210407145914.311479-1-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
free_irq() is missing in case of error, fix that.
Signed-off-by: Pavel Machek (CIP) <pavel@denx.de>
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20200606153103.GA17905@amd
Signed-off-by: Mark Brown <broonie@kernel.org>
Now that the fifo depths and thresholds are properly in the axg-fifo
driver, we can relax the constraints on period. As long as the period is a
multiple of the fifo burst size (8 bytes) things should be OK.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20191218172420.1199117-5-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Let the fifo driver parse the fifo depth from DT. Eventually all DT should
have this property. Until it is actually the case, default to 256 bytes if
the property is missing. 256 bytes is the size of the smallest fifo on the
supported SoCs.
On the supported SoC, fifo A is usually bigger than the other ones. With
depth known, we can improve the usage of the fifo and adapt the setup of
request threshold.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20191218172420.1199117-4-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
On TODDR sm1, the fifo threshold register field is slightly different
compared to the other SoCs. This leads to the fifo A being flushed to
memory every 8kB. If the period is smaller than that, several periods
are pushed to memory and notified at once. This is not ideal.
Fix the register field update. With this, the fifos are flushed every
128B. We could still do better, like adapt the threshold depending on
the period size, but at least it consistent across the different
SoC/fifos
Fixes: 5ac825c3d8 ("ASoC: meson: axg-toddr: add sm1 support")
Reported-by: Alden DSouza <aldend@google.com>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20191218172420.1199117-2-jbrunet@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Clean up the driver with the new managed buffer allocation API.
The superfluous snd_pcm_lib_malloc_pages() and
snd_pcm_lib_free_pages() calls are dropped.
Cc: Jerome Brunet <jbrunet@baylibre.com>
Cc: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Acked-by: Jerome Brunet <jbrunet@baylibre.com>
Link: https://lore.kernel.org/r/20191210142614.19405-6-tiwai@suse.de
Signed-off-by: Mark Brown <broonie@kernel.org>
Use devm_platform_ioremap_resource() to simplify the code a bit.
This is detected by coccinelle.
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: YueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/r/20190727150738.54764-31-yuehaibing@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
The g12a fifos gained the ability to set the initial address of the
pointer within the buffer, instead of defaulting to the buffer start
address.
It is not very useful to us (yet) but we need to put a copy the buffer
start address in the related register for the fifo to work properly on the
g12a SoC family
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
snd_pcm_lib_preallocate_pages() and co always succeed, so the error
check is simply redundant. Drop it.
Reviewed-by: Jaroslav Kysela <perex@perex.cz>
Acked-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Return value of request_irq() was irgnored. Fix this and report
the failure if any
Fixes: 6dc4fa179f ("ASoC: meson: add axg fifo base driver")
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Use IRQ_RETVAL instead of the open coded ternary operation.
Suggested-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Amlogic's axg SoCs have two types of fifos which are the memory
interfaces of the audio subsystem. FRDDR provides the playback
interface while TODDR provides the capture interface.
The way these fifos operate is very similar. Only a few settings
are specific to each.
They implement the same pcm driver here and the specifics of each
will be dealt with the related DAI driver.
Signed-off-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
