[PATCH 1/2] drm/amdgpu: drop volatile from ring buffer

[Tvrtko Ursulin]

On 08/10/2024 19:11, Christian König wrote:
> Volatile only prevents the compiler from re-ordering reads and writes.
> Since we always only modify the ring buffer from one CPU thread and have
> an explicit barrier before signaling the HW this should have no effect at
> all and just prevents compiler optimisations.
> 
> Signed-off-by: Christian König <christian.koenig at amd.com>
> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c | 10 +++-------
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h | 11 ++++-------
>   2 files changed, 7 insertions(+), 14 deletions(-)
> 
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> index 910293664902..a6e28fe3f8d6 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
> @@ -109,21 +109,17 @@ int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned int ndw)
>   void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count)
>   {
>   	uint32_t occupied, chunk1, chunk2;
> -	uint32_t *dst;
>   
>   	occupied = ring->wptr & ring->buf_mask;
> -	dst = &ring->ring[occupied];
>   	chunk1 = ring->buf_mask + 1 - occupied;
>   	chunk1 = (chunk1 >= count) ? count : chunk1;
>   	chunk2 = count - chunk1;
>   
>   	if (chunk1)
> -		memset32(dst, ring->funcs->nop, chunk1);
> +		memset32(&ring->ring[occupied], ring->funcs->nop, chunk1);
>   
> -	if (chunk2) {
> -		dst = ring->ring;
> -		memset32(dst, ring->funcs->nop, chunk2);
> -	}
> +	if (chunk2)
> +		memset32(ring->ring, ring->funcs->nop, chunk2);
>   
>   	ring->wptr += count;
>   	ring->wptr &= ring->ptr_mask;
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> index 574336d6714a..36fc9578c53c 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.h
> @@ -246,7 +246,7 @@ struct amdgpu_ring {
>   	struct drm_gpu_scheduler	sched;
>   
>   	struct amdgpu_bo	*ring_obj;
> -	volatile uint32_t	*ring;
> +	uint32_t		*ring;
>   	unsigned		rptr_offs;
>   	u64			rptr_gpu_addr;
>   	volatile u32		*rptr_cpu_addr;
> @@ -288,7 +288,7 @@ struct amdgpu_ring {
>   	u64			cond_exe_gpu_addr;
>   	volatile u32		*cond_exe_cpu_addr;
>   	unsigned int		set_q_mode_offs;
> -	volatile u32		*set_q_mode_ptr;
> +	u32			*set_q_mode_ptr;
>   	u64			set_q_mode_token;
>   	unsigned		vm_hub;
>   	unsigned		vm_inv_eng;
> @@ -386,10 +386,8 @@ static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
>   					      void *src, int count_dw)
>   {
>   	unsigned occupied, chunk1, chunk2;
> -	void *dst;
>   
>   	occupied = ring->wptr & ring->buf_mask;
> -	dst = (void *)&ring->ring[occupied];
>   	chunk1 = ring->buf_mask + 1 - occupied;
>   	chunk1 = (chunk1 >= count_dw) ? count_dw : chunk1;
>   	chunk2 = count_dw - chunk1;
> @@ -397,12 +395,11 @@ static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
>   	chunk2 <<= 2;
>   
>   	if (chunk1)
> -		memcpy(dst, src, chunk1);
> +		memcpy(&ring->ring[occupied], src, chunk1);
>   
>   	if (chunk2) {
>   		src += chunk1;
> -		dst = (void *)ring->ring;
> -		memcpy(dst, src, chunk2);
> +		memcpy(ring->ring, src, chunk2);
>   	}
>   
>   	ring->wptr += count_dw;

Dropping volatile and removal of local variables in 
amdgpu_ring_insert_nop() and amdgpu_ring_write_multiple() look unrelated 
to me. Best practice would be if you had split those two changes. Or at 
least add to the commit message the usual "while at it" blurb.

Regards,

Tvrtko