Add support to the K3 DSP remoteproc driver to configure the C7xv
subsystem core on AM62A SoCs. The C7xv susbsytem is based on C71 DSP
with anlytics engine for deep learning purposes. The remoteproc
handling for device management is similar to the C66/C71 DSPs on K3
J7 family SoCs, even though there are additional hardware accelerators
and IP updates to C7xv subsystem.
Signed-off-by: Jai Luthra <j-luthra@ti.com>
Signed-off-by: Hari Nagalla <hnagalla@ti.com>
Link: https://lore.kernel.org/r/20221230132453.32022-3-hnagalla@ti.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Some firmwares expect the OS drivers to configure the CTABLE
entries publishing dynamically allocated memory regions. For
example, the PRU Ethernet firmwares use the C28 and C30 entries
for retrieving the Shared RAM and System SRAM (OCMC) areas
allocated by the PRU Ethernet client driver.
Provide a way for users to do that through a new API,
pru_rproc_set_ctable(). The API returns 0 on success and
a negative value on error.
NOTE:
The programmable CTABLE entries are typically re-programmed by
the PRU firmwares when dealing with a certain block of memory
during block processing. This API provides an interface to the
PRU client drivers to publish a dynamically allocated memory
block with the PRU firmware using a CTABLE entry instead of a
negotiated address in shared memory. Additional synchronization
may be needed between the PRU client drivers and firmwares if
different addresses needs to be published at run-time reusing
the same CTABLE entry.
CTABLE for stands for "constant table".
Each CTable entry just holds the upper address bits so PRU can
reference to external memory with larger address bits.
For use case please see
prueth_sw_emac_config() in "drivers/net/ethernet/ti/prueth_switch.c"
/* Set in constant table C28 of PRUn to ICSS Shared memory */
pru_rproc_set_ctable(prueth->pru0, PRU_C28, sharedramaddr);
pru_rproc_set_ctable(prueth->pru1, PRU_C28, sharedramaddr);
/* Set in constant table C30 of PRUn to OCMC memory */
pru_rproc_set_ctable(prueth->pru0, PRU_C30, ocmcaddr);
pru_rproc_set_ctable(prueth->pru1, PRU_C30, ocmcaddr);
Signed-off-by: Andrew F. Davis <afd@ti.com>
Signed-off-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Roger Quadros <rogerq@ti.com>
Signed-off-by: Grzegorz Jaszczyk <grzegorz.jaszczyk@linaro.org>
Signed-off-by: MD Danish Anwar <danishanwar@ti.com>
Link: https://lore.kernel.org/r/20230106121046.886863-6-danishanwar@ti.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
The PRU remoteproc driver is not configured for 'auto-boot' by default,
and allows to be booted either by in-kernel PRU client drivers or by
userspace using the generic remoteproc sysfs interfaces. The sysfs
interfaces should not be permitted to change the remoteproc firmwares
or states when a PRU is being managed by an in-kernel client driver.
Use the newly introduced remoteproc generic 'sysfs_read_only' flag to
provide these restrictions by setting and clearing it appropriately
during the PRU acquire and release steps.
Signed-off-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Grzegorz Jaszczyk <grzegorz.jaszczyk@linaro.org>
Signed-off-by: MD Danish Anwar <danishanwar@ti.com>
Reviewed-by: Roger Quadros <rogerq@kernel.org>
Link: https://lore.kernel.org/r/20230106121046.886863-5-danishanwar@ti.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Add two new APIs, pru_rproc_get() and pru_rproc_put(), to the PRU
driver to allow client drivers to acquire and release the remoteproc
device associated with a PRU core. The PRU cores are treated as
resources with only one client owning it at a time.
The pru_rproc_get() function returns the rproc handle corresponding
to a PRU core identified by the device tree "ti,prus" property under
the client node. The pru_rproc_put() is the complementary function
to pru_rproc_get().
Signed-off-by: Suman Anna <s-anna@ti.com>
Signed-off-by: Tero Kristo <t-kristo@ti.com>
Signed-off-by: Grzegorz Jaszczyk <grzegorz.jaszczyk@linaro.org>
Signed-off-by: MD Danish Anwar <danishanwar@ti.com>
Reviewed-by: Roger Quadros <rogerq@kernel.org>
Link: https://lore.kernel.org/r/20230106121046.886863-4-danishanwar@ti.com
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Clocks are properly reference counted and do not need to be inside the
lock range.
Right now this triggers a false-positive lockdep warning on MT8192 based
Chromebooks, through a combination of mtk-scp that has a cros-ec-rpmsg
sub-device, the (actual) cros-ec I2C adapter registration, I2C client
(not on cros-ec) probe doing i2c transfers and enabling clocks.
This is a false positive because the cros-ec-rpmsg under mtk-scp does
not have an I2C adapter, and also each I2C adapter and cros-ec instance
have their own mutex.
Move the clk operations outside of the send_lock range.
Fixes: 63c13d61ea ("remoteproc/mediatek: add SCP support for mt8183")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230104083110.736377-1-wenst@chromium.org
[Fixed "Fixes:" tag line]
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
Currently, the notification like QCOM_SSR_BEFORE_SHUTDOWN is not exactly
sent before starting shutdown activity on remote subsystem but it is
getting sent after sysmon shutdown request to remote.
On getting QCOM_SSR_BEFORE_SHUTDOWN, some client want remote subsystem
to be alive to communicate but as sysmon shutdown request is getting
sent to remote before QCOM_SSR_BEFORE_SHUTDOWN notification sent to
kernel client due to which remote is not in a condition to communicate
with kernel clients.
Fixing the subdevice ordering will fix this as ssr subdevice will be
first one to get triggered in shutdown/stop path.
Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/1671024983-22634-1-git-send-email-quic_mojha@quicinc.com
Only msm8996 and msm8998 SLPIs need the RPM_SMD_AGGR2_NOC_CLK
(as aggre2 clock). None of the other platforms do. Back when the support
for the mentioned platforms was added to the q6v5 pass driver, the
devm_clk_get_optional was not available, so the has_aggre2_clk was
necessary in order to differentiate between plaforms that need this
clock and those which do not. Now that devm_clk_get_optional is available,
we can drop the has_aggre2_clk. This makes the adsp_data more cleaner
and removes the check within adsp_init_clocks.
Signed-off-by: Abel Vesa <abel.vesa@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20220718121514.2451590-1-abel.vesa@linaro.org
Some remote processors (e.g. QCS404 CDSP, SC7180 MPSS/MSS) can be brought
to life using two different bindings:
1. PIL (Peripheral Image Loader)
2. PAS (Peripheral Authentication Service)
They still describe the same hardware - firmware load for remote
processor - but use different methods to communicate with that
processor.
For these SoCs, the qcom,adsp.yaml bindings were describing the PAS
method, however for easier customization of board DTS, the bindings
combined additional properties from PIL: reset lines, qcom,halt-regs and
additional clocks. The devices and Linux kernel drivers, when
instantiated as PAS, do not use these properties, so drop them from the
bindings.
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Acked-by: Rob Herring <robh@kernel.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20221124184333.133911-6-krzysztof.kozlowski@linaro.org
This is not a fallback compatible, it must be present in addition to
"qcom,pronto-v*". It is also not documented in qcom,wcnss-pil.txt. This
is the reason for documenting it in a separate commit.
This compatible is used in the wcn36xx driver to determine which
register must be accessed. However it isn't immediately clear why the
wcn36xx driver relies on this extra compatible string rather than just
looking for "qcom,pronto-v*".
Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20221001031345.31293-4-sireeshkodali1@gmail.com
Add the compatible for MSS as found on the MSM8953 platform.
The situation is similar to the existing bindings for MSM8974: there is
an additional "mss" power domain that must be voted for while starting
up the remote processor. The difference is that on MSM8974 the power
domain is represented as a regulator (firmware expects specific voltage)
while on MSM8953 it is represented as power domain (firmware expects
performance state instead of voltage).
Handle this difference by adding the "mss" power domain as optional
third item, and then restrict it (and make it required) only when using
the MSM8953 compatible.
Co-developed-by: Sireesh Kodali <sireeshkodali1@gmail.com>
Signed-off-by: Sireesh Kodali <sireeshkodali1@gmail.com>
Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20220908182433.466908-7-stephan.gerhold@kernkonzept.com
Port the initialization sequence necessary for booting the modem remote
processor on the MSM8909 SoC from Qualcomm's msm-3.10 release [1].
The sequence is actually similar to the existing one for MSM8996 and
MSM8998 except that there is no separate QDSP6SS_MEM_PWR_CTL register
and most of the "memories" are enabled at once instead of sequentially.
To reuse the existing code just insert some if statements where needed
and add a configuration similar to the one from MSM8916.
[1]: 56dcedc8da
Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20220908182433.466908-6-stephan.gerhold@kernkonzept.com
qcom,q6v5.txt covers multiple SoCs with quite different binding
requirements. Converting this into one DT schema would require
several if statements, making the DT schema overall harder to
read and understand.
To avoid this, follow the example of SC7180/SC7280 and split
"qcom,msm8916-mss-pil" (and the equivalent deprecated "qcom,q6v5-pil"
compatible) into a separate DT schema. The schema is somewhat based
on the one for SC7180/SC7280 but adjusted for the old platforms.
Compared to the old plain text bindings, add missing documentation for
the "bam-dmux" subnode and recommend one particular approach to specify
the MBA/MPSS "memory-region" (the other one is marked as deprecated).
Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20220908182433.466908-3-stephan.gerhold@kernkonzept.com
Due to several bugs caused by timers being re-armed after they are
shutdown and just before they are freed, a new state of timers was added
called "shutdown". After a timer is set to this state, then it can no
longer be re-armed.
The following script was run to find all the trivial locations where
del_timer() or del_timer_sync() is called in the same function that the
object holding the timer is freed. It also ignores any locations where
the timer->function is modified between the del_timer*() and the free(),
as that is not considered a "trivial" case.
This was created by using a coccinelle script and the following
commands:
$ cat timer.cocci
@@
expression ptr, slab;
identifier timer, rfield;
@@
(
- del_timer(&ptr->timer);
+ timer_shutdown(&ptr->timer);
|
- del_timer_sync(&ptr->timer);
+ timer_shutdown_sync(&ptr->timer);
)
... when strict
when != ptr->timer
(
kfree_rcu(ptr, rfield);
|
kmem_cache_free(slab, ptr);
|
kfree(ptr);
)
$ spatch timer.cocci . > /tmp/t.patch
$ patch -p1 < /tmp/t.patch
Link: https://lore.kernel.org/lkml/20221123201306.823305113@linutronix.de/
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Acked-by: Pavel Machek <pavel@ucw.cz> [ LED ]
Acked-by: Kalle Valo <kvalo@kernel.org> [ wireless ]
Acked-by: Paolo Abeni <pabeni@redhat.com> [ networking ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull spi fix from Mark Brown:
"One driver specific change here which handles the case where a SPI
device for some reason tries to change the bus speed during a message
on fsl_spi hardware, this should be very unusual"
* tag 'spi-fix-v6.2-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/broonie/spi:
spi: fsl_spi: Don't change speed while chipselect is active
