[Intel-gfx] [PATCH 2/3] drm/i915/gt: Stall around xcs invalidations on tgl

[Mika Kuoppala]

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

> Whether this is an arbitrary stall or a vital ingredient, neverthess the
> impact is noticeable. If we do not have the stall around the xcs
> invalidation before a request, writes within that request sometimes go
> astray.
>
> Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/2169
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> Cc: Mika Kuoppala <mika.kuoppala at linux.intel.com>
> ---
>  drivers/gpu/drm/i915/gt/intel_lrc.c | 18 ++++++++++++------
>  1 file changed, 12 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> index 353b1717fe84..104bef04498d 100644
> --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> @@ -4761,10 +4761,12 @@ static int gen12_emit_flush_render(struct i915_request *request,
>  
>  static int gen12_emit_flush(struct i915_request *request, u32 mode)
>  {
> +#define WA_CNT 32 /* Magic delay */

Utterly nasty. We need to hunt for an explanation.

Acked-by: Mika Kuoppala <mika.kuoppala at linux.intel.com>

>  	intel_engine_mask_t aux_inv = 0;
>  	u32 cmd, *cs;
> +	int n;
>  
> -	cmd = 4;
> +	cmd = 4 * WA_CNT;
>  	if (mode & EMIT_INVALIDATE)
>  		cmd += 2;
>  	if (mode & EMIT_INVALIDATE)
> @@ -4781,7 +4783,8 @@ static int gen12_emit_flush(struct i915_request *request, u32 mode)
>  
>  	cmd = MI_FLUSH_DW + 1;
>  
> -	/* We always require a command barrier so that subsequent
> +	/*
> +	 * We always require a command barrier so that subsequent
>  	 * commands, such as breadcrumb interrupts, are strictly ordered
>  	 * wrt the contents of the write cache being flushed to memory
>  	 * (and thus being coherent from the CPU).
> @@ -4794,10 +4797,12 @@ static int gen12_emit_flush(struct i915_request *request, u32 mode)
>  			cmd |= MI_INVALIDATE_BSD;
>  	}
>  
> -	*cs++ = cmd;
> -	*cs++ = LRC_PPHWSP_SCRATCH_ADDR;
> -	*cs++ = 0; /* upper addr */
> -	*cs++ = 0; /* value */
> +	for (n = 0; n < WA_CNT; n++) {
> +		*cs++ = cmd;
> +		*cs++ = LRC_PPHWSP_SCRATCH_ADDR;
> +		*cs++ = 0; /* upper addr */
> +		*cs++ = 0; /* value */
> +	}
>  
>  	if (aux_inv) { /* hsdes: 1809175790 */
>  		struct intel_engine_cs *engine;
> @@ -4818,6 +4823,7 @@ static int gen12_emit_flush(struct i915_request *request, u32 mode)
>  	intel_ring_advance(request, cs);
>  
>  	return 0;
> +#undef WA_CNT
>  }
>  
>  static void assert_request_valid(struct i915_request *rq)
> -- 
> 2.20.1