| Age | Commit message (Collapse) | Author |
|---|
|
The mailbox framework has a single inflight request at a time. If
a request is sent while another is still active, it will be queued
to the mailbox core ring buffer.
ACPM protocol did not serialize the calls to the mailbox subsystem so we
could start the timeout ticks in parallel for multiple requests, while
just one was being inflight.
Consider a hypothetical case where the xfer timeout is 100ms and an ACPM
transaction takes 90ms:
| 0ms: Message #0 is queued in mailbox layer and sent out, then sits
| at acpm_dequeue_by_polling() with a timeout of 100ms
| 1ms: Message #1 is queued in mailbox layer but not sent out yet.
| Since send_message() doesn't block, it also sits at
| acpm_dequeue_by_polling() with a timeout of 100ms
| ...
| 90ms: Message #0 is completed, txdone is called and message #1 is sent
| 101ms: Message #1 times out since the count started at 1ms. Even though
| it has only been inflight for 11ms.
Fix the problem by moving mbox_send_message() and mbox_client_txdone()
immediately after the message has been written to the TX queue and while
still keeping the ACPM TX queue lock. We thus tie together the TX write
with the doorbell ring and mark the TX as done after the doorbell has
been rung. This guarantees that the doorbell has been rang before
starting the timeout ticks. We should also see some performance
improvement as we no longer wait to receive a response before ringing
the doorbell for the next request, so the ACPM firmware shall be able to
drain faster the TX queue. Another benefit is that requests are no
longer able to ring the doorbell one for the other, so it eases
debugging. Finally, the mailbox software queue will always contain a
single doorbell request due to the serialization done at the ACPM TX
queue level. Protocols like ACPM, that handle their own hardware queues
need a passthrough mailbox API, where they are able to just ring the
doorbell or flip a bit directly into the mailbox controller. The mailbox
software queue mechanism, the locking done into the mailbox core is not
really needed, so hopefully this lays the foundation for a passthrough
mailbox API.
Reported-by: André Draszik <andre.draszik@linaro.org>
Fixes: a88927b534ba ("firmware: add Exynos ACPM protocol driver")
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20250606-acpm-timeout-v2-1-306b1aa07a6c@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Pull SoC driver updates from Arnd Bergmann:
"Updates are across the usual driver subsystems with SoC specific
drivers:
- added soc specicific drivers for sophgo cv1800 and sg2044, qualcomm
sm8750, and amlogic c3 and s4 chips.
- cache controller updates for sifive chips, plus binding changes for
other cache descriptions.
- memory controller drivers for mediatek mt6893, stm32 and cleanups
for a few more drivers
- reset controller drivers for T-Head TH1502, Sophgo sg2044 and
Renesas RZ/V2H(P)
- SCMI firmware updates to better deal with buggy firmware, plus
better support for Qualcomm X1E and NXP i.MX specific interfaces
- a new platform driver for the crypto firmware on Cznic Turris
Omnia/MOX
- cleanups for the TEE firmware subsystem and amdtee driver
- minor updates and fixes for freescale/nxp, qualcomm, google,
aspeed, wondermedia, ti, nxp, renesas, hisilicon, mediatek,
broadcom and samsung SoCs"
* tag 'soc-drivers-6.16' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (133 commits)
soc: aspeed: Add NULL check in aspeed_lpc_enable_snoop()
soc: aspeed: lpc: Fix impossible judgment condition
ARM: aspeed: Don't select SRAM
docs: firmware: qcom_scm: Fix kernel-doc warning
soc: fsl: qe: Consolidate chained IRQ handler install/remove
firmware: qcom: scm: Allow QSEECOM for HP EliteBook Ultra G1q
dt-bindings: mfd: qcom,tcsr: Add compatible for ipq5018
dt-bindings: cache: add QiLai compatible to ax45mp
memory: stm32_omm: Fix error handling in stm32_omm_disable_child()
dt-bindings: cache: Convert marvell,tauros2-cache to DT schema
dt-bindings: cache: Convert marvell,{feroceon,kirkwood}-cache to DT schema
soc: samsung: exynos-pmu: enable CPU hotplug support for gs101
MAINTAINERS: Add google,gs101-pmu-intr-gen.yaml binding file
dt-bindings: soc: samsung: exynos-pmu: gs101: add google,pmu-intr-gen phandle
dt-bindings: soc: google: Add gs101-pmu-intr-gen binding documentation
bus: fsl-mc: Use strscpy() instead of strscpy_pad()
soc: fsl: qbman: Remove const from portal->cgrs allocation type
bus: fsl_mc: Fix driver_managed_dma check
bus: fsl-mc: increase MC_CMD_COMPLETION_TIMEOUT_MS value
bus: fsl-mc: drop useless cleanup
...
|
|
Correct kerneldoc warnings after commit a8dc26a0ec43 ("firmware:
exynos-acpm: introduce devm_acpm_get_by_node()") changed the function
prototype:
exynos-acpm.c:672: warning: Function parameter or struct member 'acpm_np' not described in 'acpm_get_by_node'
exynos-acpm.c:672: warning: expecting prototype for acpm_get_by_phandle(). Prototype was for acpm_get_by_node() instead
While touching the lines, change the name of device_node pointer to
'np' to match convention.
Fixes: a8dc26a0ec43 ("firmware: exynos-acpm: introduce devm_acpm_get_by_node()")
Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/oe-kbuild-all/202504222051.7TqaSQ48-lkp@intel.com/
Reviewed-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20250424203308.402168-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
To allow ACPM clients to simply be children of the ACPM node in DT,
they need to be able to get the ACPM handle based on that ACPM node
directly.
Add an API to allow them to do so, devm_acpm_get_by_node().
At the same time, the previous approach of acquiring the ACPM handle
via a DT phandle is now obsolete and we can remove
devm_acpm_get_by_phandle(), which was there to facilitate that. There
are no existing or anticipated upcoming users of that API, because all
clients should be children of the ACPM node going forward.
Note that no DTs have been merged that use the old approach, so doing
this API change in this driver now will not affect any existing DTs or
client drivers.
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20250327-acpm-children-v1-2-0afe15ee2ff7@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
ACPM clients (PMIC, clocks, etc.) will be modeled as children of the
ACPM interface. Populate children platform_devices from device tree.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20250327-acpm-children-v1-1-0afe15ee2ff7@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
This driver emits error messages when client drivers are trying to get
an interface handle to this driver here before this driver has
completed _probe().
Given this driver returns -EPROBE_DEFER in that case, this is not an
error and shouldn't be emitted to the log, similar to how
dev_err_probe() behaves, so just remove them.
This change also allows us to simplify the logic around releasing of
the acpm_np handle.
Fixes: a88927b534ba ("firmware: add Exynos ACPM protocol driver")
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20250319-acpm-fixes-v2-2-ac2c1bcf322b@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
We need to access the PMIC during late system shutdown and at that time
we are not allowed to sleep anymore.
To make this case work, and since we can't detect this case in a
non-racy way, switch to using udelay() unconditionally, instead of
usleep_range().
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20250325-acpm-atomic-v3-2-c66aae7df925@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
acpm_dequeue_by_polling() uses a loop counter and assumes that each
iteration of the loop takes 20us. It may take longer, though, because
usleep_range() may sleep a different amount.
Switch to using ktime_get() / ktime_before() to detect the timeout
condition more reliably.
This change also makes the code easier to follow and it allows us to
adjust the sleep if necessary, without having to adjust the loop
counter exit condition.
Reviewed-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Signed-off-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20250325-acpm-atomic-v3-1-c66aae7df925@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
When we're polling for responses and get a response that corresponds to
another request, we save the RX data in order to drain the RX queue.
If the response for the current request is not found in the request's
iteration of the queue, or if the queue is empty, we must check whether
the RX data was saved by a previous request when it drained the RX queue.
We failed to check for already saved responses when the queue was empty,
and requests could time out. Check saved RX before bailing out on empty
RX queue.
Fixes: a88927b534ba ("firmware: add Exynos ACPM protocol driver")
Reported-by: André Draszik <andre.draszik@linaro.org>
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Reviewed-by: André Draszik <andre.draszik@linaro.org>
Tested-by: André Draszik <andre.draszik@linaro.org>
Link: https://lore.kernel.org/r/20250324-acpm-drained-rx-queue-v1-1-577774335151@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
There is a spelling mistake in a dev_err_probe message. Fix it.
Signed-off-by: Colin Ian King <colin.i.king@gmail.com>
Link: https://lore.kernel.org/r/20250217091341.297401-1-colin.i.king@gmail.com
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
|
Alive Clock and Power Manager (ACPM) Message Protocol is defined for
the purpose of communication between the ACPM firmware and masters
(AP, AOC, ...). ACPM firmware operates on the Active Power Management
(APM) module that handles overall power activities.
ACPM and masters regard each other as independent hardware component and
communicate with each other using mailbox messages and shared memory.
This protocol driver provides the interface for all the client drivers
making use of the features offered by the APM. Add ACPM protocol support.
Signed-off-by: Tudor Ambarus <tudor.ambarus@linaro.org>
Link: https://lore.kernel.org/r/20250213-gs101-acpm-v9-2-8b0281b93c8b@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
|
