| Age | Commit message (Collapse) | Author |
|---|
|
Generalize existing fast channel support used in the perf protocol and
make it available to possibly any protocol refactoring the common code
into a couple of new scmi_proto_helpers_ops routines.
Make perf protocol use this new infrastructure.
No functional change.
Link: https://lore.kernel.org/r/20220704102241.2988447-4-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add support for SCMI v3.1 powercap protocol, with the exception of powercap
fast channels, exposing all the new related powercap protocol operations as
usual in include/linux/scmi_protocol.h.
Link: https://lore.kernel.org/r/20220704102241.2988447-3-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add a method to get hold of a protocol, causing it to be initialized and
its resource accounting updated, without getting access to its operations
and handle.
Some protocols, like SCMI SystemPower, do not expose any protocol ops to
the kernel OSPM agent but still need to be at least initialized. This
helper avoids the need to invoke a full devm_get_protocol() only to get
the protocol initialized while throwing away unused the protocol ops and
handle.
Link: https://lore.kernel.org/r/20220704101933.2981635-4-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
In order to minimize SCMI platform fw-side complexity, only one single SCMI
platform should be in charge of SCMI SystemPower protocol communications
with the OSPM. Enforce the existence of one single unique device associated
with SystemPower protocol across any possible number of SCMI platforms, and
warn if a system tries to register different SystemPower devices from
multiple platforms.
Link: https://lore.kernel.org/r/20220704101933.2981635-2-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add full message tracing for all transmitted and successfully received SCMI
commands, replies and notifications.
Link: https://lore.kernel.org/r/20220630173135.2086631-3-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
A reply to CLOCK_DESCRIBE_RATES issued against a non rate-discrete clock
should be composed of a triplet of rates descriptors (min/max/step)
returned all in one reply message.
This is not always the case when dealing with some SCMI server deployed in
the wild: relax such constraint while maintaining memory safety by checking
carefully the returned payload size.
While at that cleanup a stale debug printout.
Link: https://lore.kernel.org/r/20220616170347.2800771-1-cristian.marussi@arm.com
Fixes: 7bc7caafe6b1 ("firmware: arm_scmi: Use common iterators in the clock protocol")
Tested-by: Robin Murphy <robin.murphy@arm.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
A few dereferences could happen before the iterator pointer argument was
checked for NULL, causing the following smatch warnings:
drivers/firmware/arm_scmi/driver.c:1214 scmi_iterator_run() warn: variable
dereferenced before check 'i' (see line 1210)
Fix by moving the checks early and dropping some unneeded local references.
No functional change.
Link: https://lore.kernel.org/r/20220503121047.3590340-1-cristian.marussi@arm.com
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
SCMI specification defines some commands as optionally issued over multiple
messages in order to overcome possible limitations in payload size enforced
by the configured underlyinng transport.
Introduce some common protocol helpers to provide a unified solution for
issuing such SCMI multi-part commands.
Link: https://lore.kernel.org/r/20220330150551.2573938-14-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Introduce a new set of common protocol operations bound to the protocol
handle structure so that can be invoked by the protocol implementation code
even when protocols are built as distinct loadable kernel module without
the need of exporting new symbols, like already done with scmi_xfer_ops.
Add at first, as new common protocol helper, an .extended_name_get helper
which will ease implementation and will avoid code duplication when adding
new SCMIv3.1 per-protocol _NAME_GET commands.
Link: https://lore.kernel.org/r/20220330150551.2573938-11-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Move away from a statically allocated array for holding the current set of
protocols implemented by the platform in favour of allocating it
dynamically based on the number of protocols effectively advertised by the
platform via BASE protocol exchanges.
While at that, rectify the BASE_DISCOVER_LIST_PROTOCOLS loop iterations to
terminate only when a number of protocols equal to the advertised ones has
been received, instead of looping till the platform returns no more
protocols descriptors. This new behaviour is better compliant with the
specification and it has been tested to work equally well against an SCMI
stack running on top of an official SCP firmware on a JUNO board.
Link: https://lore.kernel.org/r/20220330150551.2573938-6-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
On SCMI transports whose channels are based on a shared resource the TX
channel area has to be acquired by the agent before placing the desired
command into the channel and it will be then relinquished by the platform
once the related reply has been made available into the channel.
On an RX channel the logic is reversed with the platform acquiring the
channel area and the agent reliquishing it once done by calling the
scmi_clear_channel() helper.
As a consequence, even in case of error, the agent must never try to clear
a TX channel from its side: restrict the existing clear channel call on the
the reply path only to delayed responses since they are indeed coming from
the RX channel.
Link: https://lore.kernel.org/r/20220224152404.12877-1-cristian.marussi@arm.com
Fixes: e9b21c96181c ("firmware: arm_scmi: Make .clear_channel optional")
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc
Pull ARM driver updates from Arnd Bergmann:
"There are a few separately maintained driver subsystems that we merge
through the SoC tree, notable changes are:
- Memory controller updates, mainly for Tegra and Mediatek SoCs, and
clarifications for the memory controller DT bindings
- SCMI firmware interface updates, in particular a new transport
based on OPTEE and support for atomic operations.
- Cleanups to the TEE subsystem, refactoring its memory management
For SoC specific drivers without a separate subsystem, changes include
- Smaller updates and fixes for TI, AT91/SAMA5, Qualcomm and NXP
Layerscape SoCs.
- Driver support for Microchip SAMA5D29, Tesla FSD, Renesas RZ/G2L,
and Qualcomm SM8450.
- Better power management on Mediatek MT81xx, NXP i.MX8MQ and older
NVIDIA Tegra chips"
* tag 'arm-drivers-5.18' of git://git.kernel.org/pub/scm/linux/kernel/git/soc/soc: (154 commits)
ARM: spear: fix typos in comments
soc/microchip: fix invalid free in mpfs_sys_controller_delete
soc: s4: Add support for power domains controller
dt-bindings: power: add Amlogic s4 power domains bindings
ARM: at91: add support in soc driver for new SAMA5D29
soc: mediatek: mmsys: add sw0_rst_offset in mmsys driver data
dt-bindings: memory: renesas,rpc-if: Document RZ/V2L SoC
memory: emif: check the pointer temp in get_device_details()
memory: emif: Add check for setup_interrupts
dt-bindings: arm: mediatek: mmsys: add support for MT8186
dt-bindings: mediatek: add compatible for MT8186 pwrap
soc: mediatek: pwrap: add pwrap driver for MT8186 SoC
soc: mediatek: mmsys: add mmsys reset control for MT8186
soc: mediatek: mtk-infracfg: Disable ACP on MT8192
soc: ti: k3-socinfo: Add AM62x JTAG ID
soc: mediatek: add MTK mutex support for MT8186
soc: mediatek: mmsys: add mt8186 mmsys routing table
soc: mediatek: pm-domains: Add support for mt8186
dt-bindings: power: Add MT8186 power domains
soc: mediatek: pm-domains: Add support for mt8195
...
|
|
An SCMI agent can be configured system-wide with a well-defined atomic
threshold: only SCMI synchronous command whose latency has been advertised
by the SCMI platform to be lower or equal to this configured threshold will
be considered for atomic operations, when requested and if supported by the
underlying transport at all.
Link: https://lore.kernel.org/r/20220217131234.50328-6-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add support for .mark_txdone and .poll_done transport operations to SCMI
VirtIO transport as pre-requisites to enable atomic operations.
Add a Kernel configuration option to enable SCMI VirtIO transport polling
and atomic mode for selected SCMI transactions while leaving it default
disabled.
Link: https://lore.kernel.org/r/20220217131234.50328-4-cristian.marussi@arm.com
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Igor Skalkin <igor.skalkin@opensynergy.com>
Cc: Peter Hilber <peter.hilber@opensynergy.com>
Cc: virtualization@lists.linux-foundation.org
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
modprobe can't handle spaces in aliases. Get rid of it to fix the issue.
Link: https://lore.kernel.org/r/20220211102704.128354-1-sudeep.holla@arm.com
Fixes: aa4f886f3893 ("firmware: arm_scmi: add basic driver infrastructure for SCMI")
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Alyssa Ross <hi@alyssa.is>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add a new xfer parameter to mark_txdone transport operation which enables
the SCMI core to optionally pass back into the transport layer a reference
to the xfer descriptor that is being handled.
Link: https://lore.kernel.org/r/20211220195646.44498-9-cristian.marussi@arm.com
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
An SCMI transport can be configured as .atomic_enabled in order to signal
to the SCMI core that all its TX path is executed in atomic context and
that, when requested, polling mode should be used while waiting for command
responses.
When a specific platform configuration had properly configured such a
transport as .atomic_enabled, the SCMI core will also take care not to
sleep in the corresponding RX path while waiting for a response if that
specific command transaction was requested as atomic using polling mode.
Asynchronous commands should not be used in an atomic context and so a
warning is emitted if polling was requested for an asynchronous command.
Add also a method to check, from the SCMI drivers, if the underlying SCMI
transport is currently configured to support atomic transactions: this will
be used by upper layers to determine if atomic requests can be supported at
all on this SCMI instance.
Link: https://lore.kernel.org/r/20211220195646.44498-7-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add a flag to let the transport signal to the core if its handling of sync
command implies that, after .send_message has returned successfully, the
requested command can be assumed to be fully and completely executed on
SCMI platform side so that any possible response value is already
immediately available to be retrieved by a .fetch_response: in other words
the polling phase can be skipped in such a case and the response values
accessed straight away.
Note that all of the above applies only when polling mode of operation was
selected by the core: if instead a completion IRQ was found to be available
the normal response processing path based on completions will still be
followed.
Link: https://lore.kernel.org/r/20211220195646.44498-4-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
SCMI communications along TX channels can optionally be provided of a
completion interrupt; when such interrupt is not available, command
transactions should rely on polling, where the SCMI core takes care to
repeatedly evaluate the transport-specific .poll_done() function, if
available, to determine if and when a request was fully completed or
timed out.
Such mechanism is already present and working on a single transfer base:
SCMI protocols can indeed enable hdr.poll_completion on specific commands
ahead of each transfer and cause that transaction to be handled with
polling.
Introduce a couple of flags to be able to enforce such polling behaviour
globally at will:
- scmi_desc.force_polling: to statically switch the whole transport to
polling mode.
- scmi_chan_info.no_completion_irq: to switch a single channel dynamically
to polling mode if, at runtime, is determined that no completion
interrupt was available for such channel.
Link: https://lore.kernel.org/r/20211220195646.44498-2-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Use new trace event to mark start of waiting for response section.
Link: https://lore.kernel.org/r/20211129191156.29322-6-cristian.marussi@arm.com
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Refactor code path waiting for message responses into a dedicated helper
function.
No functional change.
Link: https://lore.kernel.org/r/20211129191156.29322-4-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Use transport specific transmission timeout (max_rx_timeout_ms) also for
polling transactions.
Initially when polling mode was added, it was intended to be used only
in scheduler context and hence the choice of 100us for the polling timeout.
However the only user for that was dropped for other SCMI concurrency
issues, so it shouldn't cause any issue to increase this timeout value now.
Link: https://lore.kernel.org/r/20211129191156.29322-3-cristian.marussi@arm.com
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
[sudeep.holla: Updated commit message with historical facts about 100us timeout]
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Lookup cinfo data early in do_xfer so as to avoid any further init work
on xfer structure in case of error.
No functional change.
Link: https://lore.kernel.org/r/20211129191156.29322-2-cristian.marussi@arm.com
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add a new transport channel to the SCMI firmware interface driver for
SCMI message exchange based on optee transport channel. The optee
transport is realized by connecting and invoking OP-TEE SCMI service
interface PTA.
Optee transport support (CONFIG_ARM_SCMI_TRANSPORT_OPTEE) is default
enabled when optee driver (CONFIG_OPTEE) is enabled. Effective optee
transport is setup upon OP-TEE SCMI service discovery at optee
device initialization. For this SCMI UUID is registered to the optee
bus for probing. This is done from the link_supplier operator of the
SCMI optee transport.
The optee transport can use a statically defined shared memory in
which case SCMI device tree node defines it using an "arm,scmi-shmem"
compatible phandle through property shmem. Alternatively, optee transport
allocates the shared memory buffer from the optee driver when no shmem
property is defined.
The protocol used to exchange SCMI message over that shared memory is
negotiated between optee transport driver and the OP-TEE service through
capabilities exchange.
OP-TEE SCMI service is integrated in OP-TEE since its release tag 3.13.0.
The service interface is published in [1].
Link: [1] https://github.com/OP-TEE/optee_os/blob/3.13.0/lib/libutee/include/pta_scmi_client.h
Link: https://lore.kernel.org/r/20211028140009.23331-2-etienne.carriere@linaro.org
Cc: Cristian Marussi <cristian.marussi@arm.com>
Cc: Sudeep Holla <sudeep.holla@arm.com>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Etienne Carriere <etienne.carriere@linaro.org>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Use a WARN_ON() when SCMI stack is loaded to check the consistency of
configured SCMI transports instead of the current compile-time check
BUILD_BUG_ON() to avoid breaking bot-builds on random bad configs.
Bail-out early and noisy during SCMI stack initialization if no transport
was enabled in configuration since SCMI cannot work without at least one
enabled transport and such constraint cannot be enforced in Kconfig due to
circular dependency issues.
Link: https://lore.kernel.org/r/20210809092245.8730-1-cristian.marussi@arm.com
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Mailbox channels for the base protocol are setup during probe.
There can be a scenario where probe fails to acquire the base
protocol due to a timeout leading to cleaning up of all device
managed memory including the scmi_mailbox structure setup during
mailbox_chan_setup function.
| arm-scmi soc:qcom,scmi: timed out in resp(caller: version_get+0x84/0x140)
| arm-scmi soc:qcom,scmi: unable to communicate with SCMI
| arm-scmi: probe of soc:qcom,scmi failed with error -110
Now when a message arrives at cpu slightly after the timeout, the mailbox
controller will try to call the rx_callback of the client and might end
up accessing freed memory.
| rx_callback+0x24/0x160
| mbox_chan_received_data+0x44/0x94
| __handle_irq_event_percpu+0xd4/0x240
This patch frees the mailbox channels setup during probe and adds some more
error handling in case the probe fails.
Link: https://lore.kernel.org/r/1628111999-21595-1-git-send-email-rishabhb@codeaurora.org
Tested-by: Cristian Marussi <cristian.marussi@arm.com>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Rishabh Bhatnagar <rishabhb@codeaurora.org>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
This transport enables communications with an SCMI platform through virtio;
the SCMI platform will be represented by a virtio device.
Implement an SCMI virtio driver according to the virtio SCMI device spec
[1]. Virtio device id 32 has been reserved for the SCMI device [2].
The virtio transport has one Tx channel (virtio cmdq, A2P channel) and
at most one Rx channel (virtio eventq, P2A channel).
The following feature bit defined in [1] is not implemented:
VIRTIO_SCMI_F_SHARED_MEMORY.
The number of messages which can be pending simultaneously is restricted
according to the virtqueue capacity negotiated at probing time.
As soon as Rx channel message buffers are allocated or have been read
out by the arm-scmi driver, feed them back to the virtio device.
Since some virtio devices may not have the short response time exhibited
by SCMI platforms using other transports, set a generous response
timeout.
SCMI polling mode is not supported by this virtio transport since deemed
meaningless: polling mode operation is offered by the SCMI core to those
transports that could not provide a completion interrupt on the TX path,
which is never the case for virtio whose core callbacks can easily call
into core scmi_rx_callback upon messages reception.
[1] https://github.com/oasis-tcs/virtio-spec/blob/master/virtio-scmi.tex
[2] https://www.oasis-open.org/committees/ballot.php?id=3496
Link: https://lore.kernel.org/r/20210803131024.40280-16-cristian.marussi@arm.com
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Co-developed-by: Peter Hilber <peter.hilber@opensynergy.com>
Co-developed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Igor Skalkin <igor.skalkin@opensynergy.com>
[ Peter: Adapted patch for submission to upstream. ]
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
[ Cristian: simplified driver logic, changed link_supplier and channel
available/setup logic, removed dummy callbacks ]
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add a new opaque void *priv parameter to scmi_rx_callback which can be
optionally provided by the transport layer when invoking scmi_rx_callback
and that will be passed back to the transport layer in xfer->priv.
This can be used by transports that needs to keep track of their specific
data structures together with the valid xfers.
Link: https://lore.kernel.org/r/20210803131024.40280-15-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Some transports are also effectively registered with other kernel subsystem
in order to be properly probed and initialized; as a consequence such kind
of transports, and their related devices, might still not have been probed
and initialized at the time the main SCMI core driver is probed.
Add an optional .link_supplier() transport operation which can be used by
the core SCMI stack to dynamically check if the transport is ready and
dynamically link its device to the SCMI platform instance device.
Link: https://lore.kernel.org/r/20210803131024.40280-13-cristian.marussi@arm.com
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
[ Cristian: reworded commit message ]
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
The maximum number of simultaneously pending messages is a transport
specific quantity that is usually described statically in struct scmi_desc.
Some transports, though, can calculate such number only at run-time after
some initial transport specific setup and probing is completed; moreover
the resulting max message numbers could also be different between rx and
tx channels.
Add an optional get_max_msg() operation so that a transport can report more
accurate max message numbers for each channel type.
The value in scmi_desc.max_msg is still used as default when transport does
not provide any get_max_msg() method.
Link: https://lore.kernel.org/r/20210803131024.40280-11-cristian.marussi@arm.com
Co-developed-by: Peter Hilber <peter.hilber@opensynergy.com>
Co-developed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Igor Skalkin <igor.skalkin@opensynergy.com>
[ Peter: Adapted patch for submission to upstream. ]
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
[ Cristian: refactored how get_max_msg() is used to minimize core changes ]
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add configuration options to be able to select which SCMI transports have
to be compiled into the SCMI stack.
Mailbox and SMC are by default enabled if their related dependencies are
satisfied.
While doing that move all SCMI related config options in their own
dedicated submenu.
Link: https://lore.kernel.org/r/20210803131024.40280-9-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add a check for the presence of .poll_done transport operation so that
transports that do not need to support polling mode have no need to provide
a dummy .poll_done callback either and polling mode can be disabled in the
SCMI core for that tranport.
Link: https://lore.kernel.org/r/20210803131024.40280-8-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Make transport operation .clear_channel optional since some transports
do not need it and so avoid to have them implement dummy callbacks.
Link: https://lore.kernel.org/r/20210803131024.40280-7-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Even though in case of asynchronous commands an SCMI platform is
constrained to emit the delayed response message only after the related
message response has been sent, the configured underlying transport could
still deliver such messages together or in inverted order, causing races
due to the concurrent or out-of-order access to the underlying xfer.
Introduce a mechanism to grant exclusive access to an xfer in order to
properly serialize concurrent accesses to the same xfer originating from
multiple correlated messages.
Add additional state information to xfer descriptors so as to be able to
identify out-of-order message deliveries and act accordingly:
- when a delayed response is expected but delivered before the related
response, the synchronous response is considered as successfully
received and the delayed response processing is carried on as usual.
- when/if the missing synchronous response is subsequently received, it
is discarded as not congruent with the current state of the xfer, or
simply, because the xfer has been already released and so, now, the
monotonically increasing sequence number carried by the late response
is stale.
Link: https://lore.kernel.org/r/20210803131024.40280-6-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Tokens are sequence numbers embedded in the each SCMI message header: they
are used to correlate commands with responses (and delayed responses), but
their usage and policy of selection is entirely up to the caller (usually
the OSPM agent), while they are completely opaque to the callee (i.e. SCMI
platform) which merely copies them back from the command into the response
message header.
This also means that the platform does not, can not and should not enforce
any kind of policy on received messages depending on the contained sequence
number: platform can perfectly handle concurrent requests carrying the same
identifiying token if that should happen.
Moreover the platform is not required to produce in-order responses to
agent requests, the only constraint in these regards is that in case of
an asynchronous message the delayed response must be sent after the
immediate response for the synchronous part of the command transaction.
Currenly the SCMI stack of the OSPM agent selects a token for the egressing
commands picking the lowest possible number which is not already in use by
an existing in-flight transaction, which means, in other words, that we
immediately reuse any token after its transaction has completed or it has
timed out: this policy indeed does simplify management and lookup of tokens
and associated xfers.
Under the above assumptions and constraints, since there is really no state
shared between the agent and the platform to let the platform know when a
token and its associated message has timed out, the current policy of early
reuse of tokens can easily lead to the situation in which a spurious or
late received response (or delayed_response), related to an old stale and
timed out transaction, can be wrongly associated to a newer valid in-flight
xfer that just happens to have reused the same token.
This misbehaviour on such late/spurious responses is more easily exposed on
those transports that naturally have an higher level of parallelism in
processing multiple concurrent in-flight messages.
This commit introduces a new policy of selection of tokens for the OSPM
agent: each new command transfer now gets the next available, monotonically
increasing token, until tokens are exhausted and the counter rolls over.
Such new policy mitigates the above issues with late/spurious responses
since the tokens are now reused as late as possible (when they roll back
ideally) and so it is much easier to identify such late/spurious responses
to stale timed out transactions: this also helps in simplifying the
specific transports implementation since stale transport messages can be
easily identified and discarded early on in the rx path without the need
to cross check their actual state with the core transport layer.
This mitigation is even more effective when, as is usually the case, the
maximum number of pending messages is capped by the platform to a much
lower number than the whole possible range of tokens values (2^10).
This internal policy change in the core SCMI transport layer is fully
transparent to the specific transports so it has not and should not have
any impact on the transports implementation.
Link: https://lore.kernel.org/r/20210803131024.40280-5-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Some SCMI transport could need to perform some transport specific setup
before they can be used by the SCMI core transport layer: typically this
early setup consists in registering with some other kernel subsystem.
Add the optional capability for a transport to provide a couple of init
and exit functions that are assured to be called early during the SCMI
core initialization phase, well before the SCMI core probing step.
[ Peter: Adapted RFC patch by Cristian for submission to upstream. ]
Link: https://lore.kernel.org/r/20210803131024.40280-4-cristian.marussi@arm.com
Signed-off-by: Peter Hilber <peter.hilber@opensynergy.com>
[ Cristian: Fixed scmi_transports_exit point of invocation ]
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Being a while that we have SCMI trace events in the SCMI stack, remove
this debug helper and its call sites.
Link: https://lore.kernel.org/r/20210803131024.40280-3-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Add SCMI type handling to pack/unpack_scmi_header common helper functions.
Initialize hdr.type properly when initializing a command xfer.
Link: https://lore.kernel.org/r/20210803131024.40280-2-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
SCMI message headers carry a sequence number and such field is sized to
allow for MSG_TOKEN_MAX distinct numbers; moreover zero is not really an
acceptable maximum number of pending in-flight messages.
Fix accordingly the checks performed on the value exported by transports
in scmi_desc.max_msg
Link: https://lore.kernel.org/r/20210712141833.6628-3-cristian.marussi@arm.com
Reported-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
[sudeep.holla: updated the patch title and error message]
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Kernel doc validation script still complains about the following:
|No description found for return value of 'scmi_get_protocol_device'
|No description found for return value of 'scmi_devm_notifier_register'
|No description found for return value of 'scmi_devm_notifier_unregister'
Fix adding missing Return kernel-doc statements.
Link: https://lore.kernel.org/r/20210712143504.33541-1-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
The scmi_linux_errmap buffer access index is supposed to depend on the
array size to prevent element out of bounds access. It uses SCMI_ERR_MAX
to check bounds but that can mismatch with the array size. It also
changes the success into -EIO though scmi_linux_errmap is never used in
case of success, it is expected to work for success case too.
It is slightly confusing code as the negative of the error code
is used as index to the buffer. Fix it by negating it at the start and
make it more readable.
Link: https://lore.kernel.org/r/20210707135028.1869642-1-sudeep.holla@arm.com
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Since the hdr->protocol_id is set from the scmi_protocol_instance handle
just before the transfer, there is no need to initialise the same in
scmi_xfer_get_init. Remove the unnecessary initialisations.
Link: https://lore.kernel.org/r/20210608140140.2042257-1-sudeep.holla@arm.com
Tested-by: Cristian Marussi <cristian.marussi@arm.com>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Re-using timed out xfers in a loop can lead to issue if completion was
not properly reinitialized. Move reinit_completion from scmi_xfer_get to
do_xfer to avoid the issue.
Link: https://lore.kernel.org/r/20210606221232.33768-3-cristian.marussi@arm.com
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
[sudeep.holla: moved reinit_completion instead of adding another one]
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
A successfully received delayed response could anyway report a failure at
the protocol layer in the message status field.
Add a check also for this error condition.
Link: https://lore.kernel.org/r/20210608103056.3388-1-cristian.marussi@arm.com
Fixes: 58ecdf03dbb9 ("firmware: arm_scmi: Add support for asynchronous commands and delayed response")
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
0day CI kernel test robot reported following build error with randconfig
aarch64-linux-ld: drivers/firmware/arm_scmi/driver.o:(.rodata+0x1e0):
undefined reference to `scmi_mailbox_desc'
Fix the error by adding CONFIG_MAILBOX dependency for scmi_mailbox_desc.
Link: https://lore.kernel.org/r/20210603072631.1660963-1-sudeep.holla@arm.com
Cc: Etienne Carriere <etienne.carriere@linaro.org>
Cc: Cristian Marussi <cristian.marussi@arm.com>
Reviewed-by: Etienne Carriere <etienne.carriere@linaro.org>
Reviewed-by: Cristian Marussi <cristian.marussi@arm.com>
Tested-by: Cristian Marussi <cristian.marussi@arm.com>
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
During an async commands execution the Rx buffer length is at first set
to max_msg_sz when the synchronous part of the command is first sent.
However once the synchronous part completes the transport layer waits
for the delayed response which will be processed using the same xfer
descriptor initially allocated. Since synchronous response received at
the end of the xfer will shrink the Rx buffer length to the effective
payload response length, it needs to be reset again.
Raise the Rx buffer length again to max_msg_sz before fetching the
delayed response to ensure full response is read correctly from the
shared memory.
Link: https://lore.kernel.org/r/20210601102421.26581-2-cristian.marussi@arm.com
Fixes: 58ecdf03dbb9 ("firmware: arm_scmi: Add support for asynchronous commands and delayed response")
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
[sudeep.holla: moved reset to scmi_handle_response as it could race with
do_xfer_with_response]
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Having added the support for SCMI protocols as modules in order to let
vendors extend the SCMI core with their own additions it seems odd to
then force SCMI drivers built on top to use a static device table to
declare their devices since this way any new SCMI drivers addition
would need the core SCMI device table to be updated too.
Remove the static core device table and let SCMI drivers to simply declare
which device/protocol pair they need at initialization time: the core will
then take care to generate such devices dynamically during platform
initialization or at module loading time, as long as the requested
underlying protocol is defined in the devicetree.
Link: https://lore.kernel.org/r/20210316124903.35011-39-cristian.marussi@arm.com
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Extend SCMI protocols accounting mechanism to address possible module
usage and add the support to possibly define new protocols as loadable
modules.
Keep the standard protocols built into the SCMI core.
Link: https://lore.kernel.org/r/20210316124903.35011-38-cristian.marussi@arm.com
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Notification private data is currently accessible via handle->notify_priv,
this data was indeed meant to be private to the notification core support
and not to be accessible by SCMI drivers. Make it private hiding it
inside instance descriptor struct scmi_info and accessible only via
dedicated helpers.
Link: https://lore.kernel.org/r/20210316124903.35011-36-cristian.marussi@arm.com
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
|
Remove unused core scmi_xfer wrappers now that we have migrated all
protocols to the new interface based on protocol handles.
Link: https://lore.kernel.org/r/20210316124903.35011-34-cristian.marussi@arm.com
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Cristian Marussi <cristian.marussi@arm.com>
Signed-off-by: Sudeep Holla <sudeep.holla@arm.com>
|
