The existing code copies the hw_params pointer and reuses it later in
.prepare, specifically to re-initialize the ALH DMA channel
information that's lost in suspend-resume cycles.
This is not needed, we can directly access the information from the
substream/rtd - as done for the HDAudio DAIs in
sound/soc/sof/intel/hda-dai.c
In addition, using the saved pointer causes the suspend-resume test
cases to fail on specific platforms, depending on which version of GCC
is used. Péter Ujfalusi and I have spent long hours to root-cause this
problem that was reported by the Intel CI first with 6.2-rc1 and again
v6.3-rc1. In the latter case we were lucky that the problem was 100%
reproducible on local test devices, and found out that adding a
dev_dbg() or adding a call to usleep_range() just before accessing the
saved pointer "fixed" the issue. With errors appearing just by
changing the compiler version or minor changes in the code generated,
clearly we have a memory management Heisenbug.
The root-cause seems to be that the hw_params pointer is not
persistent. The soc-pcm code allocates the hw_params structure on the
stack, and passes it to the BE dailink hw_params and DAIs
hw_params. Saving such a pointer and reusing it later during the
.prepare stage cannot possibly work reliably, it's broken-by-design
since v5.10. It's astonishing that the problem was not seen earlier.
This simple fix will have to be back-ported to -stable, due to changes
to avoid the use of the get/set_dmadata routines this patch will only
apply on kernels older than v6.1.
Fixes: a5a0239c27 ("soundwire: intel: reinitialize IP+DSP in .prepare(), but only when resuming")
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20230321022642.1426611-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The latest Cadence IP splits some of the existing registers into two,
separated by a fixed offset. The bitfields themselves remain at the
same position, so we can use new helpers to dynamically add the fixed
offset.
For example, the existing MCP_CONFIG is now split in two with
MCP_CONFIG and IP_MCP_CONFIG (the naming comes directly from the
design document).
This patch adds helpers to access registers with the IP_ prefix. The
addition of the 'ip' prefix for helpers, registers and bitfields is
intentional to help reviewers spot any mistake.
For existing solutions, the offset is exactly zero so there's no
functional change - the MCP_CONFIG and IP_MCP_CONFIG are aliased to
the same address.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20230314015410.487311-13-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The message pointer is already stored in the bus->defer structure, not
need to pass it as an argument.
Suggested-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20230119073211.85979-5-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Copying the bus sdw_defer structure into the Cadence internals leads
to using stale pointers and kernel oopses on errors. It's just simpler
and safer to use the bus sdw_defer structure directly.
Link: https://github.com/thesofproject/linux/issues/4056
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20230119073211.85979-4-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
A previous patch removed unnecessary zeroing of the page registers
after a paged transaction, so now the reset_page_addr callback is
unused and can be removed.
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20230123164949.245898-3-rf@opensource.cirrus.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The response_buf was declared much larger (128 entries) than the number
of responses that could ever be written into it. The Cadence IP is
configurable up to a maximum of 32 entries, and the datasheet says
that RX_FIFO_AVAIL can be 2 larger than this. So allow up to 34
responses.
Also add checking in cdns_read_response() to prevent overflowing
reponse_buf if RX_FIFO_AVAIL contains an unexpectedly large number.
Signed-off-by: Richard Fitzgerald <rf@opensource.cirrus.com>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20221202161812.4186897-3-rf@opensource.cirrus.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The existing 'struct sdw_cdns_dma_data' has really nothing to do with
DMAs. The information is stored in the dai->dma_data, but this is
really private data that should be stored in a different context.
Beyond the academic elegance discussion, using dma_data is a problem
for new Intel hardware where the dma_data structure is already used
for true DMA handling performed by other parts of the code.
This patch prepares a transition away from the use of dma_data, for
now with a rename-only change.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20221101023521.2384586-2-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
While the hardware supports PDM streams, this capability has never
been tested or enabled on any product, so this is dead-code. Let's
remove all this.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Péter Ujfalusi <peter.ujfalusi@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Acked-By: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20211224021034.26635-8-yung-chuan.liao@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
This patch provides both a simplification of the suspend flows and a
better balanced operation during suspend/resume transition, as part of
the transition of Sound Open Firmware (SOF) to dynamic pipelines: the
DSP resources are only enabled when required instead of enabled on
startup.
The exiting code relies on a convoluted way of dealing with suspend
signals. Since there is no .suspend DAI callback, we used the
component .suspend and marked all the component DAI dmas as
'suspended'. The information was used in the .prepare stage to
differentiate resume operations from xrun handling, and only
reinitialize SHIM registers and DMA in the former case.
While this solution has been working reliably for about 2 years, there
is a much better solution consisting in trapping the TRIGGER_SUSPEND
in the .trigger DAI ops. The DMA is still marked in the same way for
the .prepare op to run, but in addition the callbacks sent to DSP
firmware are now balanced.
Normal operation:
hw_params -> intel_params_stream
hw_free -> intel_free_stream
suspend -> intel_free_stream
prepare -> intel_params_stream
This balanced operation was not required with existing SOF firmware
relying on static pipelines instantiated at every boot. With the
on-going transition to dynamic pipelines, it's however a requirement
to keep the use count for the DAI widget balanced across all
transitions.
The component suspend is not removed but instead modified to deal with
a corner case: when a substream is PAUSED, the ALSA core does not
throw the TRIGGER_SUSPEND. This is problematic since the refcount for
all pipelines and widgets is not balanced, leading to issues on
resume. The trigger callback keeps track of the 'paused' state with a
new flag, which is tested during the component suspend called later to
release the remaining DSP resources. These resources will be
re-enabled in the .prepare step.
The IPC used in the TRIGGER_SUSPEND to release DSP resources is not a
problem since the BE dailink is already marked as non-atomic.
Co-developed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Acked-By: Vinod Koul <vkoul@kernel.org>
Link: https://lore.kernel.org/r/20211224021034.26635-4-yung-chuan.liao@linux.intel.com
Signed-off-by: Mark Brown <broonie@kernel.org>
For debug, it's interesting to create a loopback stream for each link
and use debugfs to set a source and target PDI. The target PDI would
need to be an RX port and use the same register configurations as the
source PDI. This capability allows e.g. for the headphone playback
stream to be snooped on the headset capture stream, or alternatively
for the addition of a dedicated loopback stream, in addition of
regular capture for that link.
This patch only adds the debugfs part, the port/PDI handling will be
handled in the next patches.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20210714032209.11284-10-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The Cadence IP exposes a small number of self-clearing bits in
the MCP_CONTROL and MCP_CONFIG_UPDATE registers.
We currently do not check that those bits are indeed cleared,
e.g. during resume operations. That could lead to resuming peripheral
devices too early.
In addition, if we happen to read these registers, update one of the
fields and write the register back, we may be writing stale data that
might have been cleared in hardware. These sort of race conditions
could lead to e.g. doing a hw_reset twice or stopping a clock that
just restarted. There is no clear way of avoiding these potential race
conditions other than making sure that these registers fields are
cleared before any read-modify-write sequence. If we detect this sort
of errors, we only log them since there is no clear recovery
possible. The only way out is likely to restart the IP with a
suspend/resume cycle.
Note that the checks are performed before updating the registers, as
well as after the Intel 'sync go' sequence in multi-link mode. That
should cover both the start and end of suspend/resume hardware
configurations. The Multi-Master mode gates the configuration updates
until the 'sync go' signal is asserted, so we only check on init and
after the end of the 'sync go' sequence.
The duration of the usleep_range() was defined by the GSYNC frequency
used in multi-master mode. With a 4kHz frequency, any configuration
change might be deferred by up to 250us. Extending the range to
1000-1500us should guarantee that the configuration change is
completed without any significant impact on the overall resume
time.
Suggested-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Link: https://lore.kernel.org/r/20210714051349.13064-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
In system suspend stress cases, the SOF CI reports timeouts. The root
cause is that an alert is generated while the system suspends. The
interrupt handling generates transactions on the bus that will never
be handled because the interrupts are disabled in parallel.
As a result, the transaction never completes and times out on resume.
This error doesn't seem too problematic since it happens in a work
queue, and the system recovers without issues.
Nevertheless, this race condition should not happen. When doing a
system suspend, or when disabling interrupts, we should make sure the
current transaction can complete, and prevent new work from being
queued.
BugLink: https://github.com/thesofproject/linux/issues/2344
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Ranjani Sridharan <ranjani.sridharan@linux.intel.com>
Reviewed-by: Rander Wang <rander.wang@linux.intel.com>
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Acked-by: Jaroslav Kysela <perex@perex.cz>
Link: https://lore.kernel.org/r/20200817222340.18042-1-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The .prepare() callback is invoked for normal streaming, underflows or
during the system resume transition. In the latter case, the context
for the ALH PDIs is lost, and the DSP is not initialized properly
either, but the bus parameters don't need to be recomputed.
Conversely, when doing a regular .prepare() during an underflow, the
ALH/SHIM registers shall not be changed as the hardware cannot be
reprogrammed after the DMA started (hardware spec requirement).
This patch adds storage of PDI and hw_params in the DAI dma context,
and the difference between the types of .prepare() usages is handled
via a simple boolean, updated when suspending, and tested for in the
.prepare() case.
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200817152923.3259-6-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The existing code uses one pair of interrupt handler/thread per link
but at the hardware level the interrupt is shared. This works fine for
legacy PCI interrupts, but leads to timeouts in MSI (Message-Signaled
Interrupt) mode, likely due to edges being lost.
This patch unifies interrupt handling for all links. The dedicated
handler is removed since we use a common one for all shared interrupt
sources, and the thread function takes care of dealing with interrupt
sources. This partition follows the model used for the SOF IPC on
HDaudio platforms, where similar timeout issues were noticed and doing
all the interrupt handling/clearing in the thread improved
reliability/stability.
Validation results with 4 links active in parallel show a night-and-day
improvement with no timeouts noticed even during stress tests. Latency
and quality of service are not affected by the change - mostly because
events on a SoundWire link are throttled by the bus frame rate
(typically 8..48kHz).
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200716150947.22119-8-yung-chuan.liao@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
In multi-master mode, the IP will only accept SSP intervals with
integer relationships between the frame rate and the gsync frequency.
E.g for a 48kHz frame rate and 4 kHz gsync signal, the SSP interval
can only be 1, 2, 3, 4, 6, 12.
To simplify we only allow one SSP per gsync interval.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200317163329.25501-11-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Add support for clock stop and restart, with two configuration
parameters:
1) when entering the ClockStop mode, Slave-initiated wakes can be
prevented.
2) When exiting the ClockStop mode, the caller can request a Bus Reset
(either if all Slaves were configured in ClockStopMode1 or the Master
IP lost context and enumeration is required)
The code handles the case where no Slaves are present by configuring
the IP to treat COMMAND_IGNORED as success.
The exit_reset part can be dealt with in the caller, along with the
required syncArm/syncGo sequence in multi-link mode.
Signed-off-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200317163329.25501-6-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
If master is in clock stop state, driver can't modify registers
in master except the registers for clock stop setting.
Signed-off-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200317163329.25501-4-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
There is no need for the clock_stop_exit argument with the latest
implementation
Signed-off-by: Rander Wang <rander.wang@intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20200317163329.25501-3-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The use of clock stop is not a requirement, the IP can e.g. be
completely power gated and not detect any wakes while in s2idle/deep
sleep.
For now clock-stop is not supported anyways so the control parameter
is always false. This will be revisited when we add clock stop.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20191022235448.17586-6-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Multiple changes squashed in single patch to avoid tick-tock effect
and avoid breaking compilation/bisect
1. Per the hardware documentation, all changes to MCP_CONFIG,
MCP_CONTROL, MCP_CMDCTRL and MCP_PHYCTRL need to be validated with a
self-clearing write to MCP_CONFIG_UPDATE. Add a helper and do the
update when the CONFIG is changed.
2. Move interrupt enable after interrupt handler registration
3. Add a new helper to start the hardware bus reset with maximum duration
to make sure the Slave(s) correctly detect the reset pattern and to
ensure electrical conflicts can be resolved.
4. flush command FIFOs
Better error handling will be provided after interrupt disable is
provided in follow-up patches.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20191022235448.17586-2-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
PDI number should match dai->id, there is no need to track if a PDI is
allocated or not.
Signed-off-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190916192348.467-7-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
The existing Linux code uses a 1:1 mapping between ports and PDIs, but
still has an independent allocation of ports and PDIs.
Let's simplify the code and remove the port layer by only using PDIs.
This patch does not change any functionality, just removes unnecessary
code.
This will also allow for further simplifications where the PDIs are
not dynamically allocated but instead described in a topology file.
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190916192348.467-5-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Add debugfs file to dump the Cadence master registers.
Credits: this patch is based on an earlier internal contribution by
Vinod Koul, Sanyog Kale, Shreyas Nc and Hardik Shah.
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Sanyog Kale <sanyog.r.kale@intel.com>
Signed-off-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20190821185821.12690-3-pierre-louis.bossart@linux.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Add support for Cadence port management and implement
master port ops.
Signed-off-by: Sanyog Kale <sanyog.r.kale@intel.com>
Signed-off-by: Shreyas NC <shreyas.nc@intel.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
There can be instances where drivers using Cadence IP might want
to set sdw_master_ops differently per instance of it's use, so
remove the cdns_master_ops and export the APIs.
Signed-off-by: Shreyas NC <shreyas.nc@intel.com>
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Implement sdw_master_ops with support for xfer_msg, xfer_msg_defer
and reset_page_addr. Since Cadence module doesn't know the systems
it will be used, set the read_prop to the bus helper.
Signed-off-by: Hardik T Shah <hardik.t.shah@intel.com>
Signed-off-by: Sanyog Kale <sanyog.r.kale@intel.com>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Acked-By: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Vinod Koul <vinod.koul@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
