When devices are instantiated from device-tree, pdev->id is set to -1.
Rework the driver so it doesn't depend on the ID.
Tegra I2S instantiated from board files are configured with pdev
name "tegra-i2s" and ID 0 or 1. The driver core then names the
device "tegra-i2s.0" or "tegra-i2s.1". This is not changing.
When a device is instantiated from device-tree, it will have
pdev->name="" and pdev->id=-1. For this reason, the pdev->id value is
not something we can rely on.
This patch doesn't actually change any names though:
When a device is instantiated from device-tree, the overall device name
will be "${unit_address}.${node_name}". This causes issues such as
clk_get() failures due to lack of a device-name match. To solve that,
AUXDATA was invented, to force a specific device name, thus allowing
dev_name() to return the same as the non-device-tree case. Tegra
currently uses AUXDATA for the I2S controllers. Eventually, AUXDATA will
go away, most likely replaced by phandle-based references within the
device tree.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Commit 1ee46ebd("ASoC: Make the DAI ops constant in the DAI structure")
introduced the possibility to have constant DAI ops structures, yet this is
barley used in both existing drivers and also new drivers being submitted,
although none of them modifies its DAI ops structure. The later is not
surprising since existing drivers are often used as templates for new drivers.
So this patch just constifies all existing snd_soc_dai_ops structs to eliminate
the issue altogether.
The patch was generated with the following coccinelle semantic patch:
// <smpl>
@@
identifier ops;
@@
-struct snd_soc_dai_ops ops =
+const struct snd_soc_dai_ops ops =
{ ... };
// </smpl>
Signed-off-by: Lars-Peter Clausen <lars@metafoo.de>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
module_platform_drive saves some boiler-plate code.
The devm_ APIs remove the need to manually clean up allocations,
thus removing some code.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
This removes potentially machine-specific routing knowledge from the
I2S driverinto the machine drivers, which is better equipped to know
what the appropriate routing configuration is.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The clock needed by the I2S driver is associated with the I2S device name
in the standard fashion. Hence, use clk_get(dev) instead of clk_get_sys(clk_name).
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Liam Girdwood <lrg@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The I2S controller needs a clock to respond to register writes. Without
this, register writes will at worst hang the CPU. In practice, I've only
observed writes being dropped.
Luckily, the dropped register writes historically had no effect:
TEGRA_I2S_TIMING: The value we wrote was the reset default.
TEGRA_I2S_FIFO_SCR: The default was for the FIFOs to request more data
when one slot was empty. The requested value was for the FIFOs to request
when four slots were empty. The DMA controller in the mainline kernel is
configured to burst a single entry at a time into the FIFO, hence there
was no issue. The only negative effect was on bus efficiency losses due
to an increased number of arbitration attempts.
However, in various non-upstream changes, the DMA controller now bursts
four entries at a time into the FIFO. If there is only space for one
entry, the data is simply dropped. In practice, this resulted in 3/4 of
samples being dropped, and playback at 4x the expected rate and pitch.
By fixing the clocking issue, this is solved.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Liam Girdwood <lrg@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
The prototype of the inline dummy version of tegra_i2s_debug_add
was not consistent with the real version.
Reported-by: Rhyland-Klein <rklein@nvidia.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Liam Girdwood <lrg@ti.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
With the appropriate MODULE_ALIAS in place, the audio modules will be
automatically loaded; there is no longer a need for manual modprobes.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
A recent discussion on linux-arm-kernel noted that the value returned by
clk_get_sys is an opaque token, and not strictly a pointer; it is
meaningful only to the clock API, clients should not dereference the value,
and the clock API must accept any non-IS_ERR value it returned.
Hence, only IS_ERR is appropriate to interpret the result, not
IS_ERR_OR_NULL.
I checked that clk_get_sys in both ASoC's for-next and Tegra's for-next
do behave as described; NULL is not returned in the case of error.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Everything else is using snd_soc_ so we should use it here too.
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
This provides an ASoC DAI interface for Tegra's I2S controller.
Signed-off-by: Stephen Warren <swarren@nvidia.com>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
Stephen Warren
Lars-Peter Clausen
Axel Lin
Mark Brown