[Intel-gfx] [PATCH 03/23] drm/i915: Remove lrc default desc from GEM context

[Tvrtko Ursulin]

On 23/07/2019 19:38, Chris Wilson wrote:
> We only compute the lrc_descriptor() on pinning the context, i.e.
> infrequently, so we do not benefit from storing the template as the
> addressing mode is also fixed for the lifetime of the intel_context.
> 
> Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
> ---
>   drivers/gpu/drm/i915/gem/i915_gem_context.c   | 28 ++-----------------
>   .../gpu/drm/i915/gem/i915_gem_context_types.h |  2 --
>   drivers/gpu/drm/i915/gt/intel_lrc.c           | 12 +++++---
>   drivers/gpu/drm/i915/gvt/scheduler.c          |  3 --
>   4 files changed, 10 insertions(+), 35 deletions(-)
> 
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context.c b/drivers/gpu/drm/i915/gem/i915_gem_context.c
> index b28c7ca681a8..1b3dc7258ef2 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_context.c
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_context.c
> @@ -397,30 +397,6 @@ static void context_close(struct i915_gem_context *ctx)
>   	i915_gem_context_put(ctx);
>   }
>   
> -static u32 default_desc_template(const struct drm_i915_private *i915,
> -				 const struct i915_address_space *vm)
> -{
> -	u32 address_mode;
> -	u32 desc;
> -
> -	desc = GEN8_CTX_VALID | GEN8_CTX_PRIVILEGE;
> -
> -	address_mode = INTEL_LEGACY_32B_CONTEXT;
> -	if (vm && i915_vm_is_4lvl(vm))
> -		address_mode = INTEL_LEGACY_64B_CONTEXT;
> -	desc |= address_mode << GEN8_CTX_ADDRESSING_MODE_SHIFT;
> -
> -	if (IS_GEN(i915, 8))
> -		desc |= GEN8_CTX_L3LLC_COHERENT;
> -
> -	/* TODO: WaDisableLiteRestore when we start using semaphore
> -	 * signalling between Command Streamers
> -	 * ring->ctx_desc_template |= GEN8_CTX_FORCE_RESTORE;
> -	 */
> -
> -	return desc;
> -}
> -
>   static struct i915_gem_context *
>   __create_context(struct drm_i915_private *i915)
>   {
> @@ -459,7 +435,6 @@ __create_context(struct drm_i915_private *i915)
>   	i915_gem_context_set_recoverable(ctx);
>   
>   	ctx->ring_size = 4 * PAGE_SIZE;
> -	ctx->desc_template = default_desc_template(i915, NULL);
>   
>   	for (i = 0; i < ARRAY_SIZE(ctx->hang_timestamp); i++)
>   		ctx->hang_timestamp[i] = jiffies - CONTEXT_FAST_HANG_JIFFIES;
> @@ -478,8 +453,9 @@ __set_ppgtt(struct i915_gem_context *ctx, struct i915_address_space *vm)
>   	struct i915_gem_engines_iter it;
>   	struct intel_context *ce;
>   
> +	GEM_BUG_ON(old && i915_vm_is_4lvl(vm) != i915_vm_is_4lvl(old));
> +
>   	ctx->vm = i915_vm_get(vm);
> -	ctx->desc_template = default_desc_template(ctx->i915, vm);
>   
>   	for_each_gem_engine(ce, i915_gem_context_lock_engines(ctx), it) {
>   		i915_vm_put(ce->vm);
> diff --git a/drivers/gpu/drm/i915/gem/i915_gem_context_types.h b/drivers/gpu/drm/i915/gem/i915_gem_context_types.h
> index 0ee61482ef94..a02d98494078 100644
> --- a/drivers/gpu/drm/i915/gem/i915_gem_context_types.h
> +++ b/drivers/gpu/drm/i915/gem/i915_gem_context_types.h
> @@ -171,8 +171,6 @@ struct i915_gem_context {
>   
>   	/** ring_size: size for allocating the per-engine ring buffer */
>   	u32 ring_size;
> -	/** desc_template: invariant fields for the HW context descriptor */
> -	u32 desc_template;
>   
>   	/** guilty_count: How many times this context has caused a GPU hang. */
>   	atomic_t guilty_count;
> diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c
> index 632344c163a8..5fdac40015cf 100644
> --- a/drivers/gpu/drm/i915/gt/intel_lrc.c
> +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c
> @@ -418,13 +418,17 @@ lrc_descriptor(struct intel_context *ce, struct intel_engine_cs *engine)
>   	BUILD_BUG_ON(MAX_CONTEXT_HW_ID > (BIT(GEN8_CTX_ID_WIDTH)));
>   	BUILD_BUG_ON(GEN11_MAX_CONTEXT_HW_ID > (BIT(GEN11_SW_CTX_ID_WIDTH)));
>   
> -	desc = ctx->desc_template;				/* bits  0-11 */
> -	GEM_BUG_ON(desc & GENMASK_ULL(63, 12));
> +	desc = INTEL_LEGACY_32B_CONTEXT;
> +	if (i915_vm_is_4lvl(ce->vm))
> +		desc = INTEL_LEGACY_64B_CONTEXT;

if-else now that the vm null check is gone.

> +	desc <<= GEN8_CTX_ADDRESSING_MODE_SHIFT;
> +
> +	desc |= GEN8_CTX_VALID | GEN8_CTX_PRIVILEGE;
> +	if (IS_GEN(engine->i915, 8))
> +		desc |= GEN8_CTX_L3LLC_COHERENT;

Don't know.. it's nicer to keep it stored it both for Gen and context 
state. What's the problem with it?

Regards,

Tvrtko

>   
>   	desc |= i915_ggtt_offset(ce->state) + LRC_HEADER_PAGES * PAGE_SIZE;
>   								/* bits 12-31 */
> -	GEM_BUG_ON(desc & GENMASK_ULL(63, 32));
> -
>   	/*
>   	 * The following 32bits are copied into the OA reports (dword 2).
>   	 * Consider updating oa_get_render_ctx_id in i915_perf.c when changing
> diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c
> index f68798ab1e7c..4c018fb1359c 100644
> --- a/drivers/gpu/drm/i915/gvt/scheduler.c
> +++ b/drivers/gpu/drm/i915/gvt/scheduler.c
> @@ -291,9 +291,6 @@ shadow_context_descriptor_update(struct intel_context *ce,
>   	 * Update bits 0-11 of the context descriptor which includes flags
>   	 * like GEN8_CTX_* cached in desc_template
>   	 */
> -	desc &= U64_MAX << 12;
> -	desc |= ce->gem_context->desc_template & ((1ULL << 12) - 1);
> -
>   	desc &= ~(0x3 << GEN8_CTX_ADDRESSING_MODE_SHIFT);
>   	desc |= workload->ctx_desc.addressing_mode <<
>   		GEN8_CTX_ADDRESSING_MODE_SHIFT;
>