From: Chris Wilson Date: Wed, 20 Jan 2016 13:43:35 +0000 (+0200) Subject: drm/i915: Use ordered seqno write interrupt generation on gen8+ execlists X-Git-Url: https://git.karo-electronics.de/?a=commitdiff_plain;h=7c17d377374dd;p=linux-beck.git drm/i915: Use ordered seqno write interrupt generation on gen8+ execlists Broadwell and later currently use the same unordered command sequence to update the seqno in the HWS status page and then assert the user interrupt. We should apply the w/a from legacy (where we do an mmio read to delay the seqno read after the interrupt), but this is not enough to enforce coherent seqno visibilty on Skylake. Rather than search for the proper post-interrupt seqno barrier, use a strongly ordered command sequence to write the seqno, then assert the user interrupt from the ring. v2: Move around the wa tail dwords to avoid adding duplicate code. v3: Add references, comments on workarounds and bit5 check. References: https://bugs.freedesktop.org/show_bug.cgi?id=93693 Testcase: igt/gem_ring_sync_loop #skl Signed-off-by: Chris Wilson Cc: Mika Kuoppala Reviewed-by: Mika Kuoppala Link: http://patchwork.freedesktop.org/patch/msgid/1453297415-17793-1-git-send-email-mika.kuoppala@intel.com --- diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c index 134379dc4dd9..7f47948d5c40 100644 --- a/drivers/gpu/drm/i915/intel_lrc.c +++ b/drivers/gpu/drm/i915/intel_lrc.c @@ -760,23 +760,34 @@ static int logical_ring_wait_for_space(struct drm_i915_gem_request *req, * on a queue waiting for the ELSP to be ready to accept a new context submission. At that * point, the tail *inside* the context is updated and the ELSP written to. */ -static void +static int intel_logical_ring_advance_and_submit(struct drm_i915_gem_request *request) { - struct intel_engine_cs *ring = request->ring; + struct intel_ringbuffer *ringbuf = request->ringbuf; struct drm_i915_private *dev_priv = request->i915; - intel_logical_ring_advance(request->ringbuf); + intel_logical_ring_advance(ringbuf); + request->tail = ringbuf->tail; - request->tail = request->ringbuf->tail; + /* + * Here we add two extra NOOPs as padding to avoid + * lite restore of a context with HEAD==TAIL. + * + * Caller must reserve WA_TAIL_DWORDS for us! + */ + intel_logical_ring_emit(ringbuf, MI_NOOP); + intel_logical_ring_emit(ringbuf, MI_NOOP); + intel_logical_ring_advance(ringbuf); - if (intel_ring_stopped(ring)) - return; + if (intel_ring_stopped(request->ring)) + return 0; if (dev_priv->guc.execbuf_client) i915_guc_submit(dev_priv->guc.execbuf_client, request); else execlists_context_queue(request); + + return 0; } static void __wrap_ring_buffer(struct intel_ringbuffer *ringbuf) @@ -1845,44 +1856,65 @@ static void bxt_a_set_seqno(struct intel_engine_cs *ring, u32 seqno) intel_flush_status_page(ring, I915_GEM_HWS_INDEX); } +/* + * Reserve space for 2 NOOPs at the end of each request to be + * used as a workaround for not being allowed to do lite + * restore with HEAD==TAIL (WaIdleLiteRestore). + */ +#define WA_TAIL_DWORDS 2 + +static inline u32 hws_seqno_address(struct intel_engine_cs *engine) +{ + return engine->status_page.gfx_addr + I915_GEM_HWS_INDEX_ADDR; +} + static int gen8_emit_request(struct drm_i915_gem_request *request) { struct intel_ringbuffer *ringbuf = request->ringbuf; - struct intel_engine_cs *ring = ringbuf->ring; - u32 cmd; int ret; - /* - * Reserve space for 2 NOOPs at the end of each request to be - * used as a workaround for not being allowed to do lite - * restore with HEAD==TAIL (WaIdleLiteRestore). - */ - ret = intel_logical_ring_begin(request, 8); + ret = intel_logical_ring_begin(request, 6 + WA_TAIL_DWORDS); if (ret) return ret; - cmd = MI_STORE_DWORD_IMM_GEN4; - cmd |= MI_GLOBAL_GTT; + /* w/a: bit 5 needs to be zero for MI_FLUSH_DW address. */ + BUILD_BUG_ON(I915_GEM_HWS_INDEX_ADDR & (1 << 5)); - intel_logical_ring_emit(ringbuf, cmd); intel_logical_ring_emit(ringbuf, - (ring->status_page.gfx_addr + - (I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT))); + (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW); + intel_logical_ring_emit(ringbuf, + hws_seqno_address(request->ring) | + MI_FLUSH_DW_USE_GTT); intel_logical_ring_emit(ringbuf, 0); intel_logical_ring_emit(ringbuf, i915_gem_request_get_seqno(request)); intel_logical_ring_emit(ringbuf, MI_USER_INTERRUPT); intel_logical_ring_emit(ringbuf, MI_NOOP); - intel_logical_ring_advance_and_submit(request); + return intel_logical_ring_advance_and_submit(request); +} - /* - * Here we add two extra NOOPs as padding to avoid - * lite restore of a context with HEAD==TAIL. - */ - intel_logical_ring_emit(ringbuf, MI_NOOP); - intel_logical_ring_emit(ringbuf, MI_NOOP); - intel_logical_ring_advance(ringbuf); +static int gen8_emit_request_render(struct drm_i915_gem_request *request) +{ + struct intel_ringbuffer *ringbuf = request->ringbuf; + int ret; - return 0; + ret = intel_logical_ring_begin(request, 6 + WA_TAIL_DWORDS); + if (ret) + return ret; + + /* w/a for post sync ops following a GPGPU operation we + * need a prior CS_STALL, which is emitted by the flush + * following the batch. + */ + intel_logical_ring_emit(ringbuf, GFX_OP_PIPE_CONTROL(5)); + intel_logical_ring_emit(ringbuf, + (PIPE_CONTROL_GLOBAL_GTT_IVB | + PIPE_CONTROL_CS_STALL | + PIPE_CONTROL_QW_WRITE)); + intel_logical_ring_emit(ringbuf, hws_seqno_address(request->ring)); + intel_logical_ring_emit(ringbuf, 0); + intel_logical_ring_emit(ringbuf, i915_gem_request_get_seqno(request)); + intel_logical_ring_emit(ringbuf, MI_USER_INTERRUPT); + return intel_logical_ring_advance_and_submit(request); } static int intel_lr_context_render_state_init(struct drm_i915_gem_request *req) @@ -2069,6 +2101,7 @@ static int logical_render_ring_init(struct drm_device *dev) ring->init_context = gen8_init_rcs_context; ring->cleanup = intel_fini_pipe_control; ring->emit_flush = gen8_emit_flush_render; + ring->emit_request = gen8_emit_request_render; ring->dev = dev; diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h index 0014fcaa5a0c..b6c7cd2b8a40 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.h +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -426,6 +426,7 @@ intel_write_status_page(struct intel_engine_cs *ring, * The area from dword 0x30 to 0x3ff is available for driver usage. */ #define I915_GEM_HWS_INDEX 0x30 +#define I915_GEM_HWS_INDEX_ADDR (I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT) #define I915_GEM_HWS_SCRATCH_INDEX 0x40 #define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)