On Fri Oct 9, 2026 at 8:54 PM JST, Alexandre Courbot wrote:
> `receive_msg` currently handles both the transport and RPC layers.
> Introduce the `MessageElement` trait to define how messages are
> processed, independently of their type.
>
> Keep the transport layer in `wait_for_msg`, and introduce
> `consume_element`, a transport-level method which reads the message's
> contents using an implementation of the `MessageElement` trait before
> advancing the CPU read pointer past it. RPC messages are handled by the
> `RpcMessageElement` wrapper, and the RPC part of `receive_msg` is moved
> to its implementation of `MessageElement`. This sets things up for
> moving the RPC code into its own module, leaving the transport agnostic
> of the message type.
>
> As a result of the new abstraction layer, RPC messages whose size is
> unexpected are consumed instead of remaining on the queue.
>
> Signed-off-by: Alexandre Courbot <[email protected]>
> ---
> drivers/gpu/nova-core/gsp/cmdq.rs | 102
> ++++++++++++++++++++++----------------
> 1 file changed, 58 insertions(+), 44 deletions(-)
>
> diff --git a/drivers/gpu/nova-core/gsp/cmdq.rs
> b/drivers/gpu/nova-core/gsp/cmdq.rs
> index a059ce74ba2d..65e1a0b6bb44 100644
> --- a/drivers/gpu/nova-core/gsp/cmdq.rs
> +++ b/drivers/gpu/nova-core/gsp/cmdq.rs
> @@ -199,6 +199,15 @@ fn write(&self, dev: &device::Device, seq: u32, dst:
> &mut GspCommand<'_>) -> Res
> }
> }
>
> +/// Trait implemented by types that can be received as single command queue
> elements.
> +///
> +/// The command queue validates the element header before calling `read()`
> to interpret the
> +/// contents.
> +trait MessageElement: Sized {
> + /// Tries to read `Self` from `message`. `dev` is the queue's device, to
> be used for logging.
> + fn read(dev: &device::Device, message: GspMessage<'_>) -> Result<Self>;
> +}
> +
> /// Trait representing messages received from the GSP.
> ///
> /// A reply that [`Cmdq::send_command`] waits for, or an event that
> [`Cmdq::await_msg`] waits for.
> @@ -224,6 +233,45 @@ fn read(
> ) -> Result<Self, Self::InitError>;
> }
>
> +/// Wrapper type for receiving a RPC message from a command queue element.
> +///
> +/// [`MessageElement`] cannot be directly implemented for all
> [`MessageFromGsp`] with a blanket
> +/// implementation as it would conflict with other future message types.
> +struct RpcMessageElement<M>(M);
> +
> +impl<M> MessageElement for RpcMessageElement<M>
> +where
> + M: MessageFromGsp,
> + Error: From<M::InitError>,
> +{
> + fn read(dev: &device::Device, message: GspMessage<'_>) -> Result<Self> {
> + let rpc_message = RpcMessage::parse(dev, message)?;
> + let function = rpc_message.header.function();
> +
> + // An early return here would leave the read pointer on this message.
I think this comment is no longer accurate.
> + let result = if matches!(function, Ok(f) if f == M::FUNCTION) {
> + let (cmd, contents_1) =
> +
> M::Message::from_bytes_prefix(rpc_message.contents.0).ok_or(EIO)?;
> + let mut sbuffer = SBufferIter::new_reader([contents_1,
> rpc_message.contents.1]);
> +
> + M::read(cmd, &mut sbuffer)
> + .map(Self)
> + .map_err(|e| e.into())
> + .inspect(|_| {
> + if !sbuffer.is_empty() {
> + dev_warn!(dev, "GSP message {:?} has unprocessed
> data\n", M::FUNCTION);
> + }
> + })
> + } else {
> + rpc_message.log(dev);
> +
> + Err(ENOMSG)
> + };
> +
> + result
> + }
> +}
nit: temporary `result` here is not necessary since you can just have
the if/else as the last expression. Clippy (only at 1.85.0) complains
about this, so might be worth fixing.
> +
> /// Number of GSP pages making the [`Msgq`].
> pub(crate) const MSGQ_NUM_PAGES: u32 = 0x3f;
>
> @@ -886,7 +934,7 @@ fn send_command<M>(&mut self, command: M) -> Result
> }
> }
>
> - /// Wait for a message to become available on the message queue.
> + /// Wait for the next element to become available on the message queue.
> ///
> /// This works purely at the transport layer and does not interpret or
> validate the message
> /// beyond the advertised length in its [`GspMsgElement`].
> @@ -933,53 +981,19 @@ fn wait_for_msg(&self, timeout: Delta) ->
> Result<GspMessage<'_>> {
> })
> }
>
> - /// Receive a message from the GSP.
> + /// Wait for the next element on the message queue, pass it to
> [`MessageElement::read`], and
> + /// advances the read pointer past it.
> ///
> - /// A message whose function code is `M::FUNCTION` is decoded and
> returned. Any other message
> - /// is logged as an event.
> - ///
> - /// The read pointer is always advanced past the message, regardless of
> whether it matched.
> + /// The read pointer advances regardless of whether
> [`MessageElement::read`] succeeds or not.
> ///
> /// # Errors
> ///
> - /// - `ETIMEDOUT` if `timeout` has elapsed before any message becomes
> available.
> - /// - `EIO` if there was some inconsistency (e.g. message shorter than
> advertised) on the
> - /// message queue.
> - /// - `ENOMSG` if the message was not the awaited reply.
> - ///
> - /// Error codes returned by [`MessageFromGsp::read`] are propagated
> as-is.
> - fn receive_msg<M: MessageFromGsp>(&mut self, timeout: Delta) -> Result<M>
> - where
> - // This allows all error types, including `Infallible`, to be used
> for `M::InitError`.
> - Error: From<M::InitError>,
> - {
> + /// Errors from [`Self::wait_for_msg`] and from [`MessageElement::read`]
> are propagated
> + /// as-is.
> + fn consume_element<M: MessageElement>(&mut self, timeout: Delta) ->
> Result<M> {
Ok, so we are changing the semantics to this always consumes the element
in a few more cases. I think this is ok and it's documented so sgtm. And
maybe we do something for poisoning later?
Reviewed-by: Eliot Courtney <[email protected]>
