Re: [PATCH v8 2/3] drm/tyr: add GPU reset infrastructure
From: Alice Ryhl
Date: Wed Nov 18 2026 - 07:10:45 EST
On Fri, Nov 13, 2026 at 05:32:27PM +0300, Onur Özkan wrote:
> + /// Processes one scheduled reset request.
> + ///
> + /// If the pending reset cannot be claimed, the worker returns immediately.
> + ///
> + /// It first claims [`ResetState::Pending`], then waits for earlier hardware
> + /// accesses to complete before issuing the reset and returning the worker
> + /// state to [`ResetState::Idle`].
> + ///
> + /// Panthor reference:
> + /// - drivers/gpu/drm/panthor/panthor_device.c::panthor_device_reset_work()
> + fn reset_work(&self) {
> + if !self.try_transition(ResetState::Pending, ResetState::InProgress) {
> + return;
> + }
> +
> + dev_dbg!(self.pdev, "Starting GPU reset.\n");
> +
> + // Wait for current hardware accesses to finish before resetting.
> + let reset_guard = self.hw.close();
> + let reset_result = run_reset(self.pdev.as_ref(), &self.iomem);
> + drop(reset_guard);
> +
> + if let Err(e) = reset_result {
> + dev_err!(self.pdev, "GPU reset failed: {:?}\n", e);
> +
> + // TODO: Unplug the GPU.
> + // There is no API for unplugging the GPU and this is unreachable
> + // for now since there are no hardware users for reset API.
> + } else {
> + dev_dbg!(self.pdev, "GPU reset completed.\n");
> + }
> +
> + let _ = self.try_transition(ResetState::InProgress, ResetState::Idle);
I'm pretty sure this needs to happen before drop(reset_guard).
> + /// Creates [`ResetHandle`].
> + ///
> + /// # Safety
> + ///
> + /// The returned handle must not be leaked or otherwise prevented from
> + /// running [`Drop`], since it owns work that may borrow from `'reset`.
> + pub(crate) unsafe fn new(
> + pdev: &'reset platform::Device<Bound>,
> + iomem: ArcBorrow<'_, IoMem<'reset>>,
> + ) -> Result<impl PinInit<Self, Error>> {
> + let iomem = iomem.into();
> +
> + Ok(try_pin_init!(Self {
> + // SAFETY: The caller guarantees the handle is dropped.
> + controller <- unsafe {
> + kernel::new_scoped_work!("tyr::reset", Controller::new(pdev, iomem))
> + },
> + wq: Queue::new_ordered().build(c"tyr-reset-wq")?,
> + }))
The Result is not needed. This can simplify to:
) -> impl PinInit<Self, Error> {
let iomem = iomem.into();
try_pin_init!(Self {
// SAFETY: The caller guarantees the handle is dropped.
controller <- unsafe {
kernel::new_scoped_work!("tyr::reset", Controller::new(pdev, iomem))
},
wq: Queue::new_ordered().build(c"tyr-reset-wq")?,
})
> + }
> +
> + /// Schedules a GPU reset on the dedicated workqueue.
> + ///
> + /// If a reset is already pending or in progress the call is a no-op.
> + #[expect(dead_code)]
> + pub(crate) fn schedule(&'reset self) {
> + // TODO: Similar to `panthor_device_schedule_reset()` in Panthor, add a
> + // power management check once Tyr supports it.
> +
> + if self
> + .controller
> + .try_transition(ResetState::Idle, ResetState::Pending)
> + {
> + let _ = self.wq.enqueue_scoped(&self.controller);
> + }
Because this type is invariant over 'reset, it's impossible to call a
method that takes &'reset self.
I think you can just change this to &self.
Alice