[Intel-gfx] [PATCH 04/17] drm/i915: Wrap around the tail offset before setting ring->tail

[Mika Kuoppala]

Chris Wilson <chris at chris-wilson.co.uk> writes:

> The HW only accepts offsets within ring->size, and fails peculiarly if
> the RING_HEAD or RING_TAIL is set to ring->size. Therefore whenever we
> set ring->head/ring->tail we want to make sure it is within value (using
> intel_ring_wrap()).
>
> v2: Double check execlists as well
>
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Joonas Lahtinen <joonas.lahtinen at linux.intel.com>
> Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
> Cc: Matthew Auld <matthew.william.auld at gmail.com>
> Cc: Tvrtko Ursulin <tvrtko.ursulin at intel.com>
> ---
>  drivers/gpu/drm/i915/intel_lrc.c        |  6 ++++--
>  drivers/gpu/drm/i915/intel_ringbuffer.c |  5 +++++
>  drivers/gpu/drm/i915/intel_ringbuffer.h | 12 ++++++++++++
>  3 files changed, 21 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
> index 091e28f0e024..3e008adf5a01 100644
> --- a/drivers/gpu/drm/i915/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/intel_lrc.c
> @@ -1413,6 +1413,7 @@ __execlists_context_pin(struct intel_engine_cs *engine,
>  	ce->lrc_reg_state = vaddr + LRC_STATE_PN * PAGE_SIZE;
>  	ce->lrc_reg_state[CTX_RING_BUFFER_START+1] =
>  		i915_ggtt_offset(ce->ring->vma);
> +	GEM_BUG_ON(!intel_ring_offset_valid(ce->ring, ce->ring->head));
>  	ce->lrc_reg_state[CTX_RING_HEAD+1] = ce->ring->head;
>  
>  	ce->state->obj->pin_global++;
> @@ -2001,9 +2002,10 @@ static void execlists_reset(struct intel_engine_cs *engine,
>  
>  	/* Move the RING_HEAD onto the breadcrumb, past the hanging batch */
>  	regs[CTX_RING_BUFFER_START + 1] = i915_ggtt_offset(request->ring->vma);
> -	regs[CTX_RING_HEAD + 1] = request->postfix;
>  
> -	request->ring->head = request->postfix;
> +	request->ring->head = intel_ring_wrap(request->ring, request->postfix);
> +	regs[CTX_RING_HEAD + 1] = request->ring->head;
> +
>  	intel_ring_update_space(request->ring);
>  
>  	/* Reset WaIdleLiteRestore:bdw,skl as well */
> diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
> index 409f499c0a45..7970ecb199e2 100644
> --- a/drivers/gpu/drm/i915/intel_ringbuffer.c
> +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
> @@ -496,6 +496,10 @@ static int init_ring_common(struct intel_engine_cs *engine)
>  		DRM_DEBUG_DRIVER("%s initialization failed [head=%08x], fudging\n",
>  				 engine->name, I915_READ_HEAD(engine));
>  
> +	/* Check that the ring offsets point within the ring! */
> +	GEM_BUG_ON(!intel_ring_offset_valid(ring, ring->head));
> +	GEM_BUG_ON(!intel_ring_offset_valid(ring, ring->tail));
> +
>  	intel_ring_update_space(ring);
>  	I915_WRITE_HEAD(engine, ring->head);
>  	I915_WRITE_TAIL(engine, ring->tail);
> @@ -1064,6 +1068,7 @@ int intel_ring_pin(struct intel_ring *ring,
>  
>  void intel_ring_reset(struct intel_ring *ring, u32 tail)
>  {
> +	tail = intel_ring_wrap(ring, tail);

I am pondering this wrap here and it's usefulness. Where
could we ever get a tail which is not valid? From corrupted
context?

-Mika


>  	ring->tail = tail;
>  	ring->head = tail;
>  	ring->emit = tail;
> diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
> index b44c67849749..1d8140ac2016 100644
> --- a/drivers/gpu/drm/i915/intel_ringbuffer.h
> +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
> @@ -805,6 +805,18 @@ static inline u32 intel_ring_wrap(const struct intel_ring *ring, u32 pos)
>  	return pos & (ring->size - 1);
>  }
>  
> +static inline bool
> +intel_ring_offset_valid(const struct intel_ring *ring, u32 pos)
> +{
> +	if (pos & -ring->size) /* must be strictly within the ring */
> +		return false;
> +
> +	if (!IS_ALIGNED(pos, 8)) /* must be qword aligned */
> +		return false;
> +
> +	return true;
> +}
> +
>  static inline u32 intel_ring_offset(const struct i915_request *rq, void *addr)
>  {
>  	/* Don't write ring->size (equivalent to 0) as that hangs some GPUs. */
> -- 
> 2.17.1