PCM buffers might be allocated dynamically when the buffer
preallocation failed or a larger buffer is requested, and it's not
guaranteed that substream->dma_buffer points to the actually used
buffer.  The driver needs to refer to substream->runtime->dma_addr
instead for the buffer address.

Cc: <stable@vger.kernel.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20210731084331.32225-1-tiwai@suse.deSigned-off-by: NMark Brown <broonie@kernel.org>

Now we can use asoc_substream_to_rtd() macro,
let's use it.
Signed-off-by: NKuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87ft9n0ytv.wl-kuninori.morimoto.gx@renesas.comSigned-off-by: NMark Brown <broonie@kernel.org>

ASoC PCM core deals the empty ioctl field now as default.
Let's kill the redundant lines.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20191210145406.21419-2-tiwai@suse.deSigned-off-by: NMark Brown <broonie@kernel.org>

Clean up the drivers with the new managed buffer allocation API.
The superfluous snd_pcm_lib_malloc_pages() and
snd_pcm_lib_free_pages() calls are dropped.
Signed-off-by: NTakashi Iwai <tiwai@suse.de>
Link: https://lore.kernel.org/r/20191210142614.19405-2-tiwai@suse.deSigned-off-by: NMark Brown <broonie@kernel.org>

snd_pcm_ops is no longer needed.
Let's use component driver callback.
Signed-off-by: NKuninori Morimoto <kuninori.morimoto.gx@renesas.com>
Link: https://lore.kernel.org/r/87imp7af1t.wl-kuninori.morimoto.gx@renesas.comSigned-off-by: NMark Brown <broonie@kernel.org>

Use devm_platform_ioremap_resource() to simplify the code a bit.
This is detected by coccinelle.
Reported-by: NHulk Robot <hulkci@huawei.com>
Signed-off-by: NYueHaibing <yuehaibing@huawei.com>
Link: https://lore.kernel.org/r/20190904074833.23572-1-yuehaibing@huawei.comSigned-off-by: NMark Brown <broonie@kernel.org>

Based on 1 normalized pattern(s):

  this program is free software you can redistribute it and or modify
  it under the terms and conditions of the gnu general public license
  version 2 as published by the free software foundation this program
  is distributed in the hope it will be useful but without any
  warranty without even the implied warranty of merchantability or
  fitness for a particular purpose see the gnu general public license
  for more details

extracted by the scancode license scanner the SPDX license identifier

  GPL-2.0-only

has been chosen to replace the boilerplate/reference in 263 file(s).
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
Reviewed-by: NAllison Randal <allison@lohutok.net>
Reviewed-by: NAlexios Zavras <alexios.zavras@intel.com>
Cc: linux-spdx@vger.kernel.org
Link: https://lkml.kernel.org/r/20190529141901.208660670@linutronix.deSigned-off-by: NGreg Kroah-Hartman <gregkh@linuxfoundation.org>

snd_pcm_lib_preallocate_pages() and co always succeed, so the error
check is simply redundant.  Drop it.
Reviewed-by: NJaroslav Kysela <perex@perex.cz>
Acked-by: NMark Brown <broonie@kernel.org>
Signed-off-by: NTakashi Iwai <tiwai@suse.de>

AMD platform device acp_audio_dma can only be created by parent PCI
device driver (drivers/gpu/drm/amd/amdgpu/amdgpu_acp.c). Pass struct
device of the parent to snd_pcm_lib_preallocate_pages() so
dma_alloc_coherent() can use correct dma_ops. Otherwise, it will
use default dma_ops which is nommu_dma_ops on x86_64 even when
IOMMU is enabled and set to non passthrough mode.

Though platform device inherits some dma related fields during its
creation in mfd_add_device(), we can't simply pass its struct device
to snd_pcm_lib_preallocate_pages() because dma_ops is not among the
inherited fields. Even it were, drivers/iommu/amd_iommu.c would
ignore it because get_device_id() doesn't handle platform device.

This change shouldn't give us any trouble even struct device of the
parent becomes null or represents some non PCI device in the future,
because get_dma_ops() correctly handles null struct device or uses
the default dma_ops if struct device doesn't have it set.
Signed-off-by: NYu Zhao <yuzhao@google.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

We shouldn't assume CPU physical address we get from page_to_phys()
is same as DMA address we get from dma_alloc_coherent(). On x86_64,
we won't run into any problem with the assumption when dma_ops is
nommu_dma_ops. However, DMA address is IOVA when IOMMU is enabled.
And it's most likely different from CPU physical address when AMD
IOMMU is not in passthrough mode.
Signed-off-by: NYu Zhao <yuzhao@google.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

alsa_conformance_test -C hw:0,4 -p 1024 --debug
would sometime show:
TIME_DIFF(s)    HW_LEVEL       READ              RATE
0.000095970         1024       1024    10670001.041992
0.042609555         1024       2048       24032.168372
0.021330364         1024       3072       48006.681930
0.021339559         1024       4096       47985.996337
The issue is that in dma pointer function we can have stale value
of the register for current descriptor of channel.
The register retains the number of the last descriptor that
was transferred.

Fix ensures that we report position, 0, till the one period worth of
data is transferred.  After one period of data, in handler of period
completion interrupt we update the config and correct value of descriptor
starts reflecting.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

If capture and playback are started on different channel (I2S/BT)
there is a possibilty that channel information passed from machine driver
is overwritten before the configuration is done in dma driver.
Example:
113.597588: cz_max_startup: ---playback sets BT channel
113.597694: cz_dmic1_startup: ---capture sets I2S channel
113.597979: acp_dma_hw_params: ---configures capture for I2S channel
113.598114: acp_dma_hw_params: ---configures playback for I2S channel

This is fixed by having 2 separate instance for playback and capture.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

HW register descriptions says:
"DMA Channel Reset...Software must confirm that this bit is
cleared before reprogramming any of the channel configuration registers."
There could be cases where dma stop errored out leaving dma channel
in reset state. We need to ensure that before the start of another dma,
channel is out of the reset state.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

ACP->SYSMEM DMA happens at every I2S->SYSMEM period
completion. Thus, there is delay of x frames till
I2S->SYSMEM reaches a period length. This delay is
communicated to user space.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Give position on ACP->SYSMEM DMA channel for
the number of bytes that have been transferred on
the basis of current descriptor under service.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

In capture case we don't want ACP to SYSMEM dma
to be circular. This is because if an in place DSP
filter is applied to captured output then circular DMA
can overwrite the filter value with stale data.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Having interrupts enabled for ACP<->SYSMEM DMA transfer, we are in
for an interrupt storm.
For both playback and capture interrupts should be enabled for
I2S<->ACP DMA.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Earlier, ch1 was used to define ACP-SYSMEM transfer and ch2 for
ACP-I2S transfer. With recent patches ch1 is used to define channel
order number 1 and ch2 as channel order number 2. Thus,
Playback:
ch1:SYSMEM->ACP
ch2:ACP->I2S
Capture:
ch1:I2S->ACP
ch1:ACP->SYSMEM
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Now that the I2S channel names are fixed, and DMA data flow order is
consistent (ch1 then ch2), we can simplify channel start order:
start the upstream channel and then the downstream channel for both
playback and capture cases.
Signed-off-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

The pointer() callback gets its value by reading the I2S BYTE_COUNT
register.  This is a 64-bit runnning transaction counter. If a
transaction was aborted in the middle of a sample buffer, the counter will
stop counting on a number divisible by the buffer size.  Since we actually
use it as a pointer into an aligned buffer, however, we do want to ensure
that it always starts at a number divisible by the buffer size when
starting a transaction, hence we reset it whenever starting a transaction.

To accomplish this, it wasn't necessary to zero bytescount at the
termination of each transaction, so remove this unnecessary code.
Signed-off-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

On capture, audio data is first copied from I2S to ACP memory, and then
from ACP to SYSRAM.  The I2S_TO_ACP_DMA interrupt fires on every sample
transferred from I2S to ACP memory.  That is it fires ~48000 times per
second when capturing @ 48 kHz.  Since we don't do anything on this
interrupt anyway, disable it to save quite a few unnecessary interrupts.
The real "work" (calling snd_pcm_period_elapsed()) is done when transfer
from ACP to SYSRAM is complete.
Signed-off-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

On capture, audio data is first copied from I2S to ACP memory, and then
to SYSRAM.  For each step the channel number increases, so the names in
the driver were wrong.
Signed-off-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

It is always correct to subtract out the starting bytescount value.  Even
in the case of 2^64 byte rollover (292 Million Years in the future
@ 48000 Hz) the math still works out.
Signed-off-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Commit 6b116dfb ("ASoC: AMD: make channel 1 dma as circular") made
both channels circular, so this comment and logic no longer applies.  Always
stop ch2 (the channel closest to the output) before ch1.  This ensures
that the downstream circular DMA channel does not continue to play/capture
repeated samples after the upstream circular DMA channel has already
stopped.
Signed-off-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

ST/CZ SoC have 2 channels for capture in the I2SSP path.
The DMA though these channels is done using the same dma
descriptors.
We configure the channel and enable it on the basis of
channel selected by machine driver. Machine driver knows
which codec sits on which channel and thus sends the information
to dma driver.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Fix crash in those platforms whose machine driver does not expose
platform_info. For those platforms we rely on default value and
select I2SSP channel.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

channel 1: SYSMEM<->ACP
channel 2: ACP<->I2S
Instead of waiting on period interrupt of ch 2 and then starting
dma on ch1, we make ch1 dma as circular.
This removes dependency of period granularity on hw pointer.
Signed-off-by: NAkshu Agrawal <akshu.agrawal@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Tested-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

With in ACP, There are three I2S controllers can be
configured/enabled ( I2S SP, I2S MICSP, I2S BT).
Default enabled I2S controller instance is I2S SP.
This patch provides required changes to support I2S BT
controller Instance.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

rtd structure freed early may result in kernel panic in dma close
call back. moved releasing memory for rtd structure to the end of
dma close callback.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Added sram bank variable to audio_substream_data structure.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Added pte offset variable in audio_substream_data structure.
Added Stoney related PTE offset macros in acp header file.
Modified hw_params callback to assign the pte offset value
based on asic_type.
PTE Offset macros used to calculate no of PTE entries
need to be programmed when memory allocated for audio buffer.
Depending upon allocated audio buffer size, PTE offset values
will change.
Compared to CZ, Stoney has SRAM memory limitation i.e 48k
It is required to define separate PTE Offset macros for
Stoney.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Removed separate byte count variables for playback and capture.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Added byte count register offset variables to audio_substream_data
structure. Modified dma pointer callback.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Added dma configuration parameters to rtd structure.
Moved dma configuration parameters initialization to
hw_params callback.
Removed hard coding in prepare and trigger callbacks.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NDaniel Kurtz <djkurtz@chromium.org>
Signed-off-by: NMark Brown <broonie@kernel.org>

Trivial fix to spelling mistake in dev_err error message
Signed-off-by: NColin Ian King <colin.king@canonical.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

In order to make audio_substream_data structure variable
consistent throughout the code, changed the name from
audio_config to rtd wherever applicable.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

fixed checkpatch pl warnings.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

As per design, non-circular dma also need to be
stopped explicitly for both playback and capture
scenarios.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Moved 16bit resolution condition check for stoney platform
to acp_hw_params.Depending upon substream required register
value need to be programmed rather than enabling 16bit resolution
support all time in acp init.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>

Removed hardcoding in dma descriptor programming api's.
These changes are required to extend the logic to support
dma descriptor programming for multiple i2s controller
instances.
Signed-off-by: NVijendar Mukunda <Vijendar.Mukunda@amd.com>
Reviewed-by: NAlex Deucher <alexander.deucher@amd.com>
Signed-off-by: NMark Brown <broonie@kernel.org>