* [PATCH 0/2] gpu: nova-core: gsp: cmdq: preparatory patches for r000 and vGPU
@ 2026-11-16 15:06 Alexandre Courbot
2026-11-16 15:06 ` [PATCH 1/2] gpu: nova-core: gsp: cmdq: let GMC commands set their completion timeout Alexandre Courbot
2026-11-16 15:06 ` [PATCH 2/2] gpu: nova-core: gsp: cmdq: support GMC replies without payload Alexandre Courbot
0 siblings, 2 replies; 3+ messages in thread
From: Alexandre Courbot @ 2026-11-16 15:06 UTC (permalink / raw)
To: Danilo Krummrich, Alice Ryhl, David Airlie, Simona Vetter
Cc: John Hubbard, Alistair Popple, Timur Tabi, Eliot Courtney,
Zhi Wang, nova-gpu, dri-devel, linux-kernel, Alexandre Courbot
A couple more command queue patches for the r000 and vGPU work. Patch 1
allows commands to specify a custom timeout, as some commands can take
longer than the default 5 seconds on occasion.
Patch 2 adds support for GMC commands that expect a return status, but
no returned payload.
Based on drm-rust-next.
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
---
Alexandre Courbot (2):
gpu: nova-core: gsp: cmdq: let GMC commands set their completion timeout
gpu: nova-core: gsp: cmdq: support GMC replies without payload
drivers/gpu/nova-core/gsp/cmdq.rs | 5 +++--
drivers/gpu/nova-core/gsp/cmdq/gmc.rs | 31 ++++++++++++++++++++++++-------
drivers/gpu/nova-core/gsp/cmdq/rpc.rs | 2 +-
3 files changed, 28 insertions(+), 10 deletions(-)
---
base-commit: 8045d0e7d208ad389a387f1407ce0db9070805af
change-id: 20261116-cmdq-timeout-6cf40c6afafc
Best regards,
--
Alexandre Courbot <acourbot@nvidia.com>
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH 1/2] gpu: nova-core: gsp: cmdq: let GMC commands set their completion timeout
2026-11-16 15:06 [PATCH 0/2] gpu: nova-core: gsp: cmdq: preparatory patches for r000 and vGPU Alexandre Courbot
@ 2026-11-16 15:06 ` Alexandre Courbot
2026-11-16 15:06 ` [PATCH 2/2] gpu: nova-core: gsp: cmdq: support GMC replies without payload Alexandre Courbot
1 sibling, 0 replies; 3+ messages in thread
From: Alexandre Courbot @ 2026-11-16 15:06 UTC (permalink / raw)
To: Danilo Krummrich, Alice Ryhl, David Airlie, Simona Vetter
Cc: John Hubbard, Alistair Popple, Timur Tabi, Eliot Courtney,
Zhi Wang, nova-gpu, dri-devel, linux-kernel, Alexandre Courbot
Every wait on the message queue uses `Cmdq::RECEIVE_TIMEOUT`, but some
GMC commands (notably r000's GSP_INIT, and a few vGPU commands) might
take longer to complete.
Make the timeout a parameter of `await_msg_with`, and take it from a new
`CommandToGsp::COMPLETION_TIMEOUT` associated constant when waiting for
the reply to a GMC command. The constant defaults to
`Cmdq::RECEIVE_TIMEOUT`, so existing commands are unaffected.
No functional change intended.
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
---
drivers/gpu/nova-core/gsp/cmdq.rs | 5 +++--
drivers/gpu/nova-core/gsp/cmdq/gmc.rs | 14 +++++++++-----
drivers/gpu/nova-core/gsp/cmdq/rpc.rs | 2 +-
3 files changed, 13 insertions(+), 8 deletions(-)
diff --git a/drivers/gpu/nova-core/gsp/cmdq.rs b/drivers/gpu/nova-core/gsp/cmdq.rs
index 8d34acaef9ea..ad37c979d8b7 100644
--- a/drivers/gpu/nova-core/gsp/cmdq.rs
+++ b/drivers/gpu/nova-core/gsp/cmdq.rs
@@ -686,7 +686,7 @@ fn consume_element<R>(
/// Consumes messages from the queue until a message that satisfies `f` is encountered, and
/// returns the result of `f`.
///
- /// The message is waited for up to [`Cmdq::RECEIVE_TIMEOUT`] from the call.
+ /// The message is waited for up to `timeout` from the call.
///
/// `f` is expected to return `Ok(Some(_))` with the result to return to the caller, and
/// `Ok(None)` to signal that the received message was not the one expected (such messages are
@@ -700,9 +700,10 @@ fn consume_element<R>(
/// - Any error returned by `f`.
fn await_msg_with<R>(
&mut self,
+ timeout: Delta,
f: impl Fn(GspParsedMessage<'_>) -> Result<Option<R>>,
) -> Result<R> {
- let deadline = Instant::<Monotonic>::now() + Cmdq::RECEIVE_TIMEOUT;
+ let deadline = Instant::<Monotonic>::now() + timeout;
loop {
let remaining = deadline - Instant::<Monotonic>::now();
if remaining.is_negative() {
diff --git a/drivers/gpu/nova-core/gsp/cmdq/gmc.rs b/drivers/gpu/nova-core/gsp/cmdq/gmc.rs
index 04033c64ca72..512a327a1c89 100644
--- a/drivers/gpu/nova-core/gsp/cmdq/gmc.rs
+++ b/drivers/gpu/nova-core/gsp/cmdq/gmc.rs
@@ -6,6 +6,7 @@
use kernel::{
device,
prelude::*,
+ time::Delta,
transmute::{
AsBytes,
FromBytes, //
@@ -61,6 +62,9 @@ pub(crate) trait CommandToGsp {
/// value).
const MAX_REPLY_SIZE: u32;
+ /// Time given to the GSP to complete this command once it is sent.
+ const COMPLETION_TIMEOUT: Delta = Cmdq::RECEIVE_TIMEOUT;
+
/// Error type returned by [`CommandToGsp::init`].
type InitError;
@@ -283,8 +287,8 @@ impl Cmdq<'_> {
///
/// - `EMSGSIZE` if the command does not fit in a single command queue element.
/// - `ETIMEDOUT` if space does not become available to send the command, or if the reply does
- /// not arrive within [`Self::RECEIVE_TIMEOUT`] of the send, however many events arrive
- /// while waiting.
+ /// not arrive within [`CommandToGsp::COMPLETION_TIMEOUT`] of the send, however many events
+ /// arrive while waiting.
/// - `EIO` if the variable payload requested by the command has not been entirely
/// written to by its [`CommandToGsp::init_variable_payload`] method, or if a message fails
/// framing or checksum validation.
@@ -350,7 +354,7 @@ fn send_gmc_command<C>(&mut self, command: C) -> Result<u32>
}
/// Receives the reply of type `C::Reply` to the command sent with sequence number `seq`,
- /// waiting up to [`Cmdq::RECEIVE_TIMEOUT`] from the call.
+ /// waiting up to [`CommandToGsp::COMPLETION_TIMEOUT`] from the call.
///
/// Any GMC event that arrives first is passed to `on_event`, and any RPC message is logged.
/// Neither extends the deadline.
@@ -376,8 +380,7 @@ fn await_gmc_reply<C: CommandToGsp>(
{
let dev = self.dev;
- // TODO: replace `u32` with `NvStatus` once it lands.
- self.await_msg_with(|message| -> Result<Option<Result<C::Reply, u32>>> {
+ self.await_msg_with(C::COMPLETION_TIMEOUT, |message| {
let gmc_message = match message {
GspParsedMessage::Gmc(gmc_message) => gmc_message,
_ => {
@@ -386,6 +389,7 @@ fn await_gmc_reply<C: CommandToGsp>(
}
};
+ // TODO: use `NvStatus` instead of `u32` for status once it lands.
let status = match gmc_message.reply_status(C::COMMAND, seq) {
Some(status) => status,
// Reply to another command, e.g. one that timed out earlier.
diff --git a/drivers/gpu/nova-core/gsp/cmdq/rpc.rs b/drivers/gpu/nova-core/gsp/cmdq/rpc.rs
index 10d3f68b2f61..012f3c82e7c5 100644
--- a/drivers/gpu/nova-core/gsp/cmdq/rpc.rs
+++ b/drivers/gpu/nova-core/gsp/cmdq/rpc.rs
@@ -401,6 +401,6 @@ fn await_rpc_msg<M: MessageFromGsp>(&mut self) -> Result<M>
}
};
- self.await_msg_with(f)
+ self.await_msg_with(Cmdq::RECEIVE_TIMEOUT, f)
}
}
--
2.56.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH 2/2] gpu: nova-core: gsp: cmdq: support GMC replies without payload
2026-11-16 15:06 [PATCH 0/2] gpu: nova-core: gsp: cmdq: preparatory patches for r000 and vGPU Alexandre Courbot
2026-11-16 15:06 ` [PATCH 1/2] gpu: nova-core: gsp: cmdq: let GMC commands set their completion timeout Alexandre Courbot
@ 2026-11-16 15:06 ` Alexandre Courbot
1 sibling, 0 replies; 3+ messages in thread
From: Alexandre Courbot @ 2026-11-16 15:06 UTC (permalink / raw)
To: Danilo Krummrich, Alice Ryhl, David Airlie, Simona Vetter
Cc: John Hubbard, Alistair Popple, Timur Tabi, Eliot Courtney,
Zhi Wang, nova-gpu, dri-devel, linux-kernel, Alexandre Courbot
Some GMC commands are answered with a status and no payload. Their
reply type cannot be `NoReply`, which marks commands that the GSP does
not answer at all.
Implement `MessageFromGsp` for `()` so these commands can use it as
their reply type.
Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
---
drivers/gpu/nova-core/gsp/cmdq/gmc.rs | 17 +++++++++++++++--
1 file changed, 15 insertions(+), 2 deletions(-)
diff --git a/drivers/gpu/nova-core/gsp/cmdq/gmc.rs b/drivers/gpu/nova-core/gsp/cmdq/gmc.rs
index 512a327a1c89..f4311593e360 100644
--- a/drivers/gpu/nova-core/gsp/cmdq/gmc.rs
+++ b/drivers/gpu/nova-core/gsp/cmdq/gmc.rs
@@ -3,6 +3,8 @@
//! Support for the GMC (GPU Management Controller) API message type on the GSP command queue.
+use core::convert::Infallible;
+
use kernel::{
device,
prelude::*,
@@ -53,8 +55,8 @@ pub(crate) trait CommandToGsp {
/// after the GMC API header.
type Command: FromBytes + AsBytes;
- /// Type of the reply expected from the GSP, or [`NoReply`] for commands that don't have a
- /// reply.
+ /// Type of the reply expected from the GSP. Can be [`NoReply`] for commands that don't have a
+ /// reply, or `()` for commands that only expect a status code without a reply payload.
type Reply;
/// Largest reply this command accepts, in bytes. Zero for commands that the GSP does not
@@ -178,6 +180,17 @@ fn read(
) -> Result<Self, Self::InitError>;
}
+/// Reply type for commands that the GSP answers with a status and no payload.
+impl MessageFromGsp for () {
+ type InitError = Infallible;
+
+ fn read(
+ _payload: &mut SBufferIter<core::array::IntoIter<&[u8], 2>>,
+ ) -> Result<Self, Self::InitError> {
+ Ok(())
+ }
+}
+
/// A GMC message ready to be processed from the message queue.
pub(crate) struct GmcMessage<'a> {
// Reference to the GMC API header of the message.
--
2.56.0
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2026-11-16 15:07 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2026-11-16 15:06 [PATCH 0/2] gpu: nova-core: gsp: cmdq: preparatory patches for r000 and vGPU Alexandre Courbot
2026-11-16 15:06 ` [PATCH 1/2] gpu: nova-core: gsp: cmdq: let GMC commands set their completion timeout Alexandre Courbot
2026-11-16 15:06 ` [PATCH 2/2] gpu: nova-core: gsp: cmdq: support GMC replies without payload Alexandre Courbot
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®