ASoC: SOF: Intel: don't ignore IOC interrupts for non-audio transfers
The HDaudio stream interrupts are ignored unless the stream is PCM or
compressed audio. For alternate non-audio usages, such as code loader
or SoundWire BPT case, the IOC interrupt on the last buffer
transferred is silently ignored.
This patch adds a 'struct completion' for each HDaudio stream. This
capability helps detect if the non-audio data transfers
completed. There is no performance impact for audio streams.
In the code loader case, the code currently starts the DMA and
directly checks if the firmware status changes, without checking if
the DMA succeeded. With a first pass waiting for the DMA to complete,
system validation engineers can gather more precise timing information
on firmware boot time or root-cause boot failures more accurately.
A timeout of 500ms was selected for the code loader DMA. This is an
experimental value which should be more than enough - higher values
would certainly be problematic from a usage/latency perspective.
ASoC: SOF: Intel: hda-loader: change signature of code loader helpers
We need to reuse cl_prepare, cl_trigger and cl_cleanup helpers from a
SoundWire context where only the device information is
available. Rather than pass the 'sdev' argument, use get_drvdata() to
retrieve the required structure.
For consistency, rename hda_cl_stream_prepare() as hda_cl_prepare().
These three helpers are also exported so that they can be referenced
from another module.
ASoC: SOF: Intel: hda-loader: add flag to tag ICCMAX stream
The existing code conflates capture stream and ICCMAX stream. This
isn't going to be true any longer when we add support for SDW BPT RX
streams. Add a boolean tag to flag ICCMAX streams.
No functionality change, only future-proofing change.
The only known users of the soc-topology ABI v4 are Chromebook
configurations. Starting from kernel v5.4, all of them are making use of
soc-topology ABI v5. The patchset first removes obsolete code from the
Intel's skylake-driver - the driver of choice for the mentioned
Chromebooks - and then proceeds with removal of relevant soc-topology.c
and uapi bits.
Cezary Rojewski (4):
ASoC: Intel: Skylake: Remove soc-topology ABI v4 support
ASoC: topology: Remove ABI v4 support
ASoC: topology: Cleanup after ABI v4 support removal
ASoC: topology: Remove obsolete ABI v4 structs
Dan Carpenter [Thu, 4 Apr 2024 07:38:45 +0000 (10:38 +0300)]
ASoC: SOF: Disable pointless writes to debugfs file
The permissions on this debugfs file are 0444 so it can't be written to.
And writing to the file hasn't done anything since commit 6e9548cdb30e
("ASoC: SOF: Convert the generic IPC flood test into SOF client").
Delete the write function.
The imx8_*_clocks API requires keeping track of all of the clocks used
by the IMX SOF driver via an array. This is unnecessary and doesn't
scale well. As such, remove it altogether and replace it with
devm_clk_bulk_get_all() and friends.
ASoC: codecs: rk3308: fix building without CONFIG_OF
The newly added codec has an extraneous of_match_ptr() annotation for
the ID table:
sound/soc/codecs/rk3308_codec.c:956:34: error: 'rk3308_codec_of_match' defined but not used [-Werror=unused-const-variable=]
956 | static const struct of_device_id rk3308_codec_of_match[] = {
Remove it to avoid introducing a warning when -Wunused-const-variable
gets enabled by default:
This set of patches factors out some repeated code to clean up
firmware control read/write functions, and removes some redundant
control notification code.
This small series fixes a build warning reported by kernel test robot
<[email protected]> and improves the Kconfig entry.
Signed-off-by: Luca Ceresoli <[email protected]>
---
Changes in v2:
- Replaced v1 implementation with __maybe_unused
- Added patch to depend on || COMPILE_TEST
- Link to v1: https://lore.kernel.org/r/20240329-rk3308-audio-codec-fix-warning-v1-1-b9d177fcd6c9@bootlin.com
---
Luca Ceresoli (2):
ASoC: codecs: rk3308: fix "defined but not used" warning on !OF
ASoC: codecs: rk3308: depend on ARM64 || COMPILE_TEST
Simon Trimmer [Mon, 25 Mar 2024 11:31:23 +0000 (11:31 +0000)]
firmware: cs_dsp: Add locked wrappers for coeff read and write
It is a common pattern for functions to take and release the DSP
pwr_lock over the cs_dsp calls to read and write firmware controls.
Add wrapper functions to do this sequence so that the calling code can
be simplified to a single function call..
ASoC: SOF: imx: drop usage of the imx8_*_clocks API
Currently, the driver has to keep track of all the clocks
it uses via an array of "struct clk_bulk_data", which doesn't
scale well and is unnecessary. As such, replace the usage of
the imx8_*_clocks with "devm_clk_bulk_get_all()" and friends.
ASoC: codecs: rk3308: depend on ARM64 || COMPILE_TEST
This codec is only known to exist in the RK3308 ARM64 SoC, so depend on it
except for compile test cases. Note that the driver won't probe without
CONFIG_OF, but ARM64 selects OF already so it is not needed.
ASoC: codecs: rk3308: fix "defined but not used" warning on !OF
Building with CONFIG_OF=n triggers:
warning: 'rk3308_codec_of_match' defined but not used [-Wunused-const-variable=]
warning: unused variable 'rk3308_codec_of_match' [-Wunused-const-variable]
Even though OF is needed for probing, fix by declaring as __maybe_unused to
still allow building on non-OF configurations for build testing.
Peter Ujfalusi [Wed, 3 Apr 2024 10:52:10 +0000 (13:52 +0300)]
ASoC: SOF: Intel: mtl: Correct the mtl_dsp_dump output
The ROM/firmware state handling has changed between CAVS and ACE
architecture:
CAVS: ROM and firmware uses the SRAM window for the state and status/error
code reporting
ACE: ROM code is using two registers to report the state and error while
the firmware is using the SRAM window to report states and status/error
codes.
Use the generic hda_dsp_get_state() to decode ROM state and error codes and
print out the firmware state and status/error code only if the SRAM
window is accessible - the firmware is booted and the Status readout is
not 0xffffffff.
Peter Ujfalusi [Wed, 3 Apr 2024 10:52:09 +0000 (13:52 +0300)]
ASoC: SOF: Intel: hda-dsp/mtl: Add support for ACE ROM state codes
The ROM state codes differ between CAVS and ACE architecture, there is a
slight overlap.
Add the ACE related state defines to mtl.h, introduce new table and
use it on case the function is called when running on ACE architecture.
Peter Ujfalusi [Wed, 3 Apr 2024 10:52:08 +0000 (13:52 +0300)]
ASoC: SOF: Intel: mtl: Implement firmware boot state check
With the corrected rom_status_reg values we can now add a check for target
boot status for firmware booting.
With the check now we can identify failed firmware boots (IMR boots) and
we can use the fallback to purge boot the DSP.
Peter Ujfalusi [Wed, 3 Apr 2024 10:52:06 +0000 (13:52 +0300)]
ASoC: SOF: Intel: lnl: Correct rom_status_reg
ACE2 architecture changed the place where the ROM updates the status code
from the shared SRAM window (and HFFLGP1QW0 in ACE1) to HFDSC register for
the status and HFDEC (HFDSC + 4) for the error code.
The rom_status_reg is not used on LNL because it was wrongly assigned based
on older platform convention (SRAM window) and it was giving inconsistent
readings.
Add new header file for lnl specific register definitions.
Peter Ujfalusi [Wed, 3 Apr 2024 10:52:05 +0000 (13:52 +0300)]
ASoC: SOF: Intel: mtl: Correct rom_status_reg
ACE1 architecture changed the place where the ROM updates the status code
from the shared SRAM window to HFFLGP1QW0 register for the status and
HFFLGP1QW0 + 4 for the error code.
The rom_status_reg is not used on MTL because it was wrongly assigned based
on older platform convention (SRAM window) and it was giving inconsistent
readings.
Peter Ujfalusi [Wed, 3 Apr 2024 10:52:04 +0000 (13:52 +0300)]
ASoC: SOF: Intel: hda: Create debugfs file to force a clean DSP boot
When IMR boot is supported on a platform it is always going to be used to
boot the DSP unless some catastrophic event happens.
There is no way for a developer to force a clean DSP boot without removing
and re-inserting the modules.
Create a 'skip_imr_boot' debugfs file which can be used to force the
next DSP boot as clean (prune) boot.
This patchset contains important updates for SoundWire support.
We initially implemented support for multiple amplifiers on different
links using a single HDaudio DMA transfer. To align with the other OS,
the 'aggregation' is now supported by the firmware. This change in
directions has kernel impacts, since we now have multiple HDaudio DMAs
to program and start, but since there are no platforms released so far
there's no end-user impact.
In addition, the behavior in case of xruns is improved by clearing the
PCM states and better handling of the hw_free case.
Note that the hw_free support will compile but will only be functional
with the companion patch "soundwire: intel: add intel_free_stream() back"
already applied in the SoundWire tree.
The first two patches change snd_soc_card_get_kcontrol() to use the
core snd_ctl_find_id_mixer() functionality instead of open-coding its
own list walk.
The last patch adds a KUnit test for this, which was tested on the
original and modified code.
Also correct the 'dai-format' to 'format' in document.
For 'audio-routing', the items are not listed. Because
this fsl-asoc-card is generic driver, which supports several
codecs, if list all the items, there will be a long list.
ASoC: SOF: ipc4-topology: Save the ALH DAI index during hw_params
The node_id for multi-gateway ALH DAI's get overwritten with the
group_id during the DAI copier's ipc_prepare op. So, save the ALH
dai_index during the BE DAI hw_params in the dai_index field of struct
ipc4_copier and use that to set the device ID in the configuration blob.
This will avoid errors during copier init after an xrun.
Note that the dai_index is typically set in topology for DMIC/SSP, but
it's not used for ALH. Reclaiming this dai_index field to store the
node_id does not generate a conflict with topology-defined values.
ASoC: SOF: pcm: reset all PCM sources in case of xruns
With the delayed stops, the xrun handling is problematic: the
applications expects everything to be reset but the firmware and DMA
are still in a PAUSED state.
This patch makes sure the prepare while pending_stop is set is
special-cased.
Add a state variable to keep track of delayed stops, in case
pcm_ops->platform_stop_during_hw_free is set.
This patch should be iso-functionality, possibly removing no-op
cases. The main purpose of this new state variable is to prepare a
follow-up patch to reset all PCM and DMAs in case of stop/prepare xrun
sequences.
ASoC: SOF: Intel: hda: Clear Soundwire node ID during BE DAI hw_free
When an xrun happens, the BE DAI hw_params doesn't get invoked before
the stream restarts with a prepare. In this case, clearing the node ID
when the DAI widget is freed and unprepared will result in an error when
it is re-initialized. In order to avoid this, move the code to clear the
node ID to the BE DAI hw_free op to keep it balanced with the BE DAI
hw_params.
ASoC: SOF: Intel: hda-dai: set dma_stream_channel_map device
sof_ipc4_dma_config_tlv{} is required for ACE2.x. The patch follow the
convention to set the dma_stream_channel_map.mapping device as
"link_id << 8 | pdi_id".
And the mapping in sof_ipc4_alh_configuration_blob{} should be the same
as dma_stream_channel_map.mapping in sof_ipc4_dma_config{}.
The purposes of device id is to map DMA tlv.
The SoundWire aggregated solution was to use one DMA on multiple links.
But, the solution changed to use one DMA for each link. It means that
we should assign HDaudio stream_tag for each cpu_dai.
The existing code derives the channel map used to program the HDaudio
link DMA from the hw_params, but that is not quite right in the case
of aggregation. The code in soc-pcm.c splits the hw_params depending
on the codec_ch_map, and we need to reconstruct the channel-map to
insert the data in the right places.
This issue is seen only on amplifier feedback capture where the data
from the second amplifier was replaced by that of the first amplifier.
Note that the loop iterator of the macro for_each_rtd_cpu_dais() is
reused in a following loop. This is different to all existing usages
of that macro, hence the use of a boolean flag to avoid an access to
an uninitialized variable.
ASoC: SOF: amd: acp-loader: abort firmware download on write error
cppcheck reports this issue:
sound/soc/sof/amd/acp-loader.c:299:6: style: Variable 'ret' is
reassigned a value before the old one has been
used. [redundantAssignment]
ret = request_firmware(&adata->fw_dbin, fw_filename, sdev->dev);
^
sound/soc/sof/amd/acp-loader.c:289:6: note: ret is assigned
ret = snd_sof_dsp_block_write(sdev, SOF_FW_BLK_TYPE_IRAM, 0,
^
sound/soc/sof/amd/acp-loader.c:299:6: note: ret is overwritten
ret = request_firmware(&adata->fw_dbin, fw_filename, sdev->dev);
^
This behavior is probably unintentional, there's no reason to return
an error for the DRAM but not the IRAM.
R-Car Gen1 and Gen4 doesn't have SCU, and current driver checks Gen1
(only) on each devices (A). But these drivers are assuming it might be
not enabled after that (B).
/* This driver doesn't support Gen1 at this point */
(A) if (rsnd_is_gen1(priv))
return 0;
(B) node = rsnd_src_of_node(priv);
if (!node)
return 0; /* not used is not error */
Gen4 DT doesn't have these device settings. (A) check doesn't work for
it, but (B) check handling it.
Gen1 DT doesn't have these device settings either, and (A) is handling
it, but (B) only is very enough. (A) is no longer needed.
Renesas Sound driver is using rsnd_is_genX() macro to handling
difference. We can use "grep rsnd_is_gen" to find-out difference for
each SoC today.
In general, driver uses flag to handling SoC difference. But in Renesas
Sound case, the differences are very specific, if you try to make it
general, the code will become be very complicated.
Some of them can use flag to handling, but we can find-out all diffence
by using "grep rsnd_is_gen" today. It is not so good, but not so bad so
far. So we will keep current style for a while.
This patch cares Gen1/Gen2 exception by using negative handling to makes no
more update for this part when we support Gen5 or later.
- if (rsnd_is_gen3() || rsnd_is_gen4() || rsnd_is_gen5() || ...)
+ if (!rsnd_is_gen1() && !rsnd_is_gen2())
ASoC: rsnd: rename rsnd_is_e3() to rsnd_is_gen3_e3()
Renesas Sound driver is using rsnd_is_genX() macro to handling
difference. We can use "grep rsnd_is_gen" to find-out difference for
each SoC except rsnd_is_e3(). Let's put same rule for E3 as well.
All SoC has reg-name, no need to support getting resource from ID
any more. Remove it. To get physical address for DMA settings,
it still need to know the index, but it is no longer fixed.
This patch tidyup it.
Gen4 calls rsnd_gen_regmap_init() for SDMC, but the address is calculated
by rsnd_gen4_dma_addr(). It needs physical address, but other info are
not needed. use NULL conf for it.
ASoC: soc-card: Use snd_ctl_find_id_mixer() instead of open-coding
Use the snd_ctl_find_id_mixer[_locked]() wrapper in
snd_soc_card_get_kcontrol[_locked]() instead of open-coding a custom
list walk of the card controls list.
Adds wrapper function snd_ctl_find_id_mixer_locked(). This is
identical to snd_ctl_find_id_mixer() except that it can be called
from code that is already holding controls_rwsem.
Zhang Yi [Tue, 2 Apr 2024 06:20:42 +0000 (14:20 +0800)]
ASoC: codecs: ES8326: Solve a headphone detection issue after suspend and resume
We got a headphone detection issue after suspend and resume.
And we fixed it by modifying the configuration at es8326_suspend
and invoke es8326_irq at es8326_resume.
We got an error report about headphone type detection and button detection.
We fixed the headphone type detection error by adjusting the debounce timer
configuration. And we fixed the button detection error by disabling the
button detection feature when the headphone are unplugged and enabling it
when headphone are plugged in.
This last part is the continuation of Brent Lu's cleanups. Multiple
quirks have been removed to use "default" configurations and
ACPI-based detection of codecs and selection of topology files.
This cleanup has been done in multiple steps/phases since Fall 2023,
thanks Brent for this contribution!
Stephen Lee [Tue, 26 Mar 2024 01:01:31 +0000 (18:01 -0700)]
ASoC: ops: Fix wraparound for mask in snd_soc_get_volsw
In snd_soc_info_volsw(), mask is generated by figuring out the index of
the most significant bit set in max and converting the index to a
bitmask through bit shift 1. Unintended wraparound occurs when max is an
integer value with msb bit set. Since the bit shift value 1 is treated
as an integer type, the left shift operation will wraparound and set
mask to 0 instead of all 1's. In order to fix this, we type cast 1 as
`1ULL` to prevent the wraparound.
The drivers for Realtek SoundWire codecs use similar logs, which is
problematic to analyze problems reported by CI tools, e.g. "Failed to
get private value: 752001 => 0000 ret=-5". It's not uncommon to have
several Realtek devices on the same platform, having the same log
thrown makes support difficult.
This patch adds __func__ to all error logs which didn't already
include it.
No functionality change, only error logs are modified.
Frank Li [Thu, 28 Mar 2024 14:51:35 +0000 (10:51 -0400)]
ASoC: dt-bindings: fsl-sai: allow only one dma-names
Some sai only connect one direction dma (rx/tx) in SOC. For example:
imx8qxp sai5 only connect tx dma channel. So allow only one "rx" or "tx"
for dma-names.
Remove description under dmas because no user use index to get dma channel.
All user use 'dma-names' to get correct dma channel. dma-names already in
'required' list.
projects / linux.git / log