From 79e6770cb1f5e32eb49590edbb794a97d0113aed Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Mon, 20 Nov 2017 20:55:01 +0000 Subject: [PATCH] drm/i915: Remove obsolete ringbuffer emission for gen8+ Since removing the module parameter to force selection of ringbuffer emission for gen8, the code is defunct. Remove it. To put the difference into perspective, a couple of microbenchmarks (bdw i7-5557u, 20170324): ring execlists exec continuous nops on all rings: 1.491us 2.223us exec sequential nops on each ring: 12.508us 53.682us single nop + sync: 9.272us 30.291us vblank_mode=0 glxgears: ~11000fps ~9000fps Since the earlier submission, gen8 ringbuffer submission has fallen further and further behind in features. So while ringbuffer may hold the throughput crown, in terms of interactive latency, execlists is much better. Alas, we have no convenient metrics for such, other than demonstrating things we can do with execlists but can not using legacy ringbuffer submission. We have made a few improvements to lowlevel execlists throughput, and ringbuffer currently panics on boot! (bdw i7-5557u, 20171026): ring execlists exec continuous nops on all rings: n/a 1.921us exec sequential nops on each ring: n/a 44.621us single nop + sync: n/a 21.953us vblank_mode=0 glxgears: n/a ~18500fps References: https://bugs.freedesktop.org/show_bug.cgi?id=87725 Signed-off-by: Chris Wilson Once-upon-a-time-Reviewed-by: Joonas Lahtinen Reviewed-by: Mika Kuoppala Link: https://patchwork.freedesktop.org/patch/msgid/20171120205504.21892-2-chris@chris-wilson.co.uk --- drivers/gpu/drm/i915/i915_debugfs.c | 44 +-- drivers/gpu/drm/i915/i915_drv.h | 2 - drivers/gpu/drm/i915/i915_gem.c | 2 +- drivers/gpu/drm/i915/i915_gem_context.c | 47 +-- drivers/gpu/drm/i915/i915_gpu_error.c | 36 -- drivers/gpu/drm/i915/intel_engine_cs.c | 14 - drivers/gpu/drm/i915/intel_hangcheck.c | 44 +-- drivers/gpu/drm/i915/intel_ringbuffer.c | 431 ++++-------------------- drivers/gpu/drm/i915/intel_ringbuffer.h | 25 +- 9 files changed, 94 insertions(+), 551 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c index 5e2a6e18771f..9cef1463d411 100644 --- a/drivers/gpu/drm/i915/i915_debugfs.c +++ b/drivers/gpu/drm/i915/i915_debugfs.c @@ -3254,44 +3254,12 @@ static int i915_semaphore_status(struct seq_file *m, void *unused) return ret; intel_runtime_pm_get(dev_priv); - if (IS_BROADWELL(dev_priv)) { - struct page *page; - uint64_t *seqno; - - page = i915_gem_object_get_page(dev_priv->semaphore->obj, 0); - - seqno = (uint64_t *)kmap_atomic(page); - for_each_engine(engine, dev_priv, id) { - uint64_t offset; - - seq_printf(m, "%s\n", engine->name); - - seq_puts(m, " Last signal:"); - for (j = 0; j < num_rings; j++) { - offset = id * I915_NUM_ENGINES + j; - seq_printf(m, "0x%08llx (0x%02llx) ", - seqno[offset], offset * 8); - } - seq_putc(m, '\n'); - - seq_puts(m, " Last wait: "); - for (j = 0; j < num_rings; j++) { - offset = id + (j * I915_NUM_ENGINES); - seq_printf(m, "0x%08llx (0x%02llx) ", - seqno[offset], offset * 8); - } - seq_putc(m, '\n'); - - } - kunmap_atomic(seqno); - } else { - seq_puts(m, " Last signal:"); - for_each_engine(engine, dev_priv, id) - for (j = 0; j < num_rings; j++) - seq_printf(m, "0x%08x\n", - I915_READ(engine->semaphore.mbox.signal[j])); - seq_putc(m, '\n'); - } + seq_puts(m, " Last signal:"); + for_each_engine(engine, dev_priv, id) + for (j = 0; j < num_rings; j++) + seq_printf(m, "0x%08x\n", + I915_READ(engine->semaphore.mbox.signal[j])); + seq_putc(m, '\n'); intel_runtime_pm_put(dev_priv); mutex_unlock(&dev->struct_mutex); diff --git a/drivers/gpu/drm/i915/i915_drv.h b/drivers/gpu/drm/i915/i915_drv.h index d2c6ba7f8acd..5ce778c26bb8 100644 --- a/drivers/gpu/drm/i915/i915_drv.h +++ b/drivers/gpu/drm/i915/i915_drv.h @@ -942,7 +942,6 @@ struct i915_gpu_state { u64 fence[I915_MAX_NUM_FENCES]; struct intel_overlay_error_state *overlay; struct intel_display_error_state *display; - struct drm_i915_error_object *semaphore; struct drm_i915_error_engine { int engine_id; @@ -2291,7 +2290,6 @@ struct drm_i915_private { struct i915_gem_context *kernel_context; /* Context only to be used for injecting preemption commands */ struct i915_gem_context *preempt_context; - struct i915_vma *semaphore; struct drm_dma_handle *status_page_dmah; struct resource mch_res; diff --git a/drivers/gpu/drm/i915/i915_gem.c b/drivers/gpu/drm/i915/i915_gem.c index 494c6be035b2..d470e10b2edd 100644 --- a/drivers/gpu/drm/i915/i915_gem.c +++ b/drivers/gpu/drm/i915/i915_gem.c @@ -4999,7 +4999,7 @@ int i915_gem_init_hw(struct drm_i915_private *dev_priv) bool intel_sanitize_semaphores(struct drm_i915_private *dev_priv, int value) { - if (INTEL_INFO(dev_priv)->gen < 6) + if (INTEL_GEN(dev_priv) < 6) return false; /* TODO: make semaphores and Execlists play nicely together */ diff --git a/drivers/gpu/drm/i915/i915_gem_context.c b/drivers/gpu/drm/i915/i915_gem_context.c index c75941d3d7e7..0704d9af261b 100644 --- a/drivers/gpu/drm/i915/i915_gem_context.c +++ b/drivers/gpu/drm/i915/i915_gem_context.c @@ -574,21 +574,21 @@ mi_set_context(struct drm_i915_gem_request *req, u32 flags) enum intel_engine_id id; const int num_rings = /* Use an extended w/a on gen7 if signalling from other rings */ - (i915_modparams.semaphores && INTEL_GEN(dev_priv) == 7) ? + (i915_modparams.semaphores && IS_GEN7(dev_priv)) ? INTEL_INFO(dev_priv)->num_rings - 1 : 0; int len; u32 *cs; flags |= MI_MM_SPACE_GTT; - if (IS_HASWELL(dev_priv) || INTEL_GEN(dev_priv) >= 8) + if (IS_HASWELL(dev_priv)) /* These flags are for resource streamer on HSW+ */ flags |= HSW_MI_RS_SAVE_STATE_EN | HSW_MI_RS_RESTORE_STATE_EN; else flags |= MI_SAVE_EXT_STATE_EN | MI_RESTORE_EXT_STATE_EN; len = 4; - if (INTEL_GEN(dev_priv) >= 7) + if (IS_GEN7(dev_priv)) len += 2 + (num_rings ? 4*num_rings + 6 : 0); cs = intel_ring_begin(req, len); @@ -596,7 +596,7 @@ mi_set_context(struct drm_i915_gem_request *req, u32 flags) return PTR_ERR(cs); /* WaProgramMiArbOnOffAroundMiSetContext:ivb,vlv,hsw,bdw,chv */ - if (INTEL_GEN(dev_priv) >= 7) { + if (IS_GEN7(dev_priv)) { *cs++ = MI_ARB_ON_OFF | MI_ARB_DISABLE; if (num_rings) { struct intel_engine_cs *signaller; @@ -623,7 +623,7 @@ mi_set_context(struct drm_i915_gem_request *req, u32 flags) */ *cs++ = MI_NOOP; - if (INTEL_GEN(dev_priv) >= 7) { + if (IS_GEN7(dev_priv)) { if (num_rings) { struct intel_engine_cs *signaller; i915_reg_t last_reg = {}; /* keep gcc quiet */ @@ -714,27 +714,7 @@ needs_pd_load_pre(struct i915_hw_ppgtt *ppgtt, struct intel_engine_cs *engine) if (engine->id != RCS) return true; - if (INTEL_GEN(engine->i915) < 8) - return true; - - return false; -} - -static bool -needs_pd_load_post(struct i915_hw_ppgtt *ppgtt, - struct i915_gem_context *to, - u32 hw_flags) -{ - if (!ppgtt) - return false; - - if (!IS_GEN8(to->i915)) - return false; - - if (hw_flags & MI_RESTORE_INHIBIT) - return true; - - return false; + return true; } static int do_rcs_switch(struct drm_i915_gem_request *req) @@ -784,21 +764,6 @@ static int do_rcs_switch(struct drm_i915_gem_request *req) engine->legacy_active_context = to; } - /* GEN8 does *not* require an explicit reload if the PDPs have been - * setup, and we do not wish to move them. - */ - if (needs_pd_load_post(ppgtt, to, hw_flags)) { - trace_switch_mm(engine, to); - ret = ppgtt->switch_mm(ppgtt, req); - /* The hardware context switch is emitted, but we haven't - * actually changed the state - so it's probably safe to bail - * here. Still, let the user know something dangerous has - * happened. - */ - if (ret) - return ret; - } - if (ppgtt) ppgtt->pd_dirty_rings &= ~intel_engine_flag(engine); diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c index 7481c8e1b5a8..3d18d67d065f 100644 --- a/drivers/gpu/drm/i915/i915_gpu_error.c +++ b/drivers/gpu/drm/i915/i915_gpu_error.c @@ -793,8 +793,6 @@ int i915_error_state_to_str(struct drm_i915_error_state_buf *m, "WA batchbuffer", ee->wa_batchbuffer); } - print_error_obj(m, NULL, "Semaphores", error->semaphore); - if (error->overlay) intel_overlay_print_error_state(m, error->overlay); @@ -903,8 +901,6 @@ void __i915_gpu_state_free(struct kref *error_ref) kfree(ee->waiters); } - i915_error_object_free(error->semaphore); - for (i = 0; i < ARRAY_SIZE(error->active_bo); i++) kfree(error->active_bo[i]); kfree(error->pinned_bo); @@ -1116,34 +1112,6 @@ gen8_engine_sync_index(struct intel_engine_cs *engine, return idx; } -static void gen8_record_semaphore_state(struct i915_gpu_state *error, - struct intel_engine_cs *engine, - struct drm_i915_error_engine *ee) -{ - struct drm_i915_private *dev_priv = engine->i915; - struct intel_engine_cs *to; - enum intel_engine_id id; - - if (!error->semaphore) - return; - - for_each_engine(to, dev_priv, id) { - int idx; - u16 signal_offset; - u32 *tmp; - - if (engine == to) - continue; - - signal_offset = - (GEN8_SIGNAL_OFFSET(engine, id) & (PAGE_SIZE - 1)) / 4; - tmp = error->semaphore->pages[0]; - idx = gen8_engine_sync_index(engine, to); - - ee->semaphore_mboxes[idx] = tmp[signal_offset]; - } -} - static void gen6_record_semaphore_state(struct intel_engine_cs *engine, struct drm_i915_error_engine *ee) { @@ -1218,7 +1186,6 @@ static void error_record_engine_registers(struct i915_gpu_state *error, if (INTEL_GEN(dev_priv) >= 6) { ee->rc_psmi = I915_READ(RING_PSMI_CTL(engine->mmio_base)); if (INTEL_GEN(dev_priv) >= 8) { - gen8_record_semaphore_state(error, engine, ee); ee->fault_reg = I915_READ(GEN8_RING_FAULT_REG); } else { gen6_record_semaphore_state(engine, ee); @@ -1453,9 +1420,6 @@ static void i915_gem_record_rings(struct drm_i915_private *dev_priv, struct i915_ggtt *ggtt = &dev_priv->ggtt; int i; - error->semaphore = - i915_error_object_create(dev_priv, dev_priv->semaphore); - for (i = 0; i < I915_NUM_ENGINES; i++) { struct intel_engine_cs *engine = dev_priv->engine[i]; struct drm_i915_error_engine *ee = &error->engine[i]; diff --git a/drivers/gpu/drm/i915/intel_engine_cs.c b/drivers/gpu/drm/i915/intel_engine_cs.c index cae76f47fec2..1fca7ac3b059 100644 --- a/drivers/gpu/drm/i915/intel_engine_cs.c +++ b/drivers/gpu/drm/i915/intel_engine_cs.c @@ -37,8 +37,6 @@ * Resource Streamer, is 66944 bytes, which rounds to 17 pages. */ #define HSW_CXT_TOTAL_SIZE (17 * PAGE_SIZE) -/* Same as Haswell, but 72064 bytes now. */ -#define GEN8_CXT_TOTAL_SIZE (18 * PAGE_SIZE) #define GEN8_LR_CONTEXT_RENDER_SIZE (20 * PAGE_SIZE) #define GEN9_LR_CONTEXT_RENDER_SIZE (22 * PAGE_SIZE) @@ -364,18 +362,6 @@ void intel_engine_init_global_seqno(struct intel_engine_cs *engine, u32 seqno) if (HAS_VEBOX(dev_priv)) I915_WRITE(RING_SYNC_2(engine->mmio_base), 0); } - if (dev_priv->semaphore) { - struct page *page = i915_vma_first_page(dev_priv->semaphore); - void *semaphores; - - /* Semaphores are in noncoherent memory, flush to be safe */ - semaphores = kmap_atomic(page); - memset(semaphores + GEN8_SEMAPHORE_OFFSET(engine->id, 0), - 0, I915_NUM_ENGINES * gen8_semaphore_seqno_size); - drm_clflush_virt_range(semaphores + GEN8_SEMAPHORE_OFFSET(engine->id, 0), - I915_NUM_ENGINES * gen8_semaphore_seqno_size); - kunmap_atomic(semaphores); - } intel_write_status_page(engine, I915_GEM_HWS_INDEX, seqno); clear_bit(ENGINE_IRQ_BREADCRUMB, &engine->irq_posted); diff --git a/drivers/gpu/drm/i915/intel_hangcheck.c b/drivers/gpu/drm/i915/intel_hangcheck.c index 12ac270a5f93..95bbb5a79c4f 100644 --- a/drivers/gpu/drm/i915/intel_hangcheck.c +++ b/drivers/gpu/drm/i915/intel_hangcheck.c @@ -27,13 +27,9 @@ static bool ipehr_is_semaphore_wait(struct intel_engine_cs *engine, u32 ipehr) { - if (INTEL_GEN(engine->i915) >= 8) { - return (ipehr >> 23) == 0x1c; - } else { - ipehr &= ~MI_SEMAPHORE_SYNC_MASK; - return ipehr == (MI_SEMAPHORE_MBOX | MI_SEMAPHORE_COMPARE | - MI_SEMAPHORE_REGISTER); - } + ipehr &= ~MI_SEMAPHORE_SYNC_MASK; + return ipehr == (MI_SEMAPHORE_MBOX | MI_SEMAPHORE_COMPARE | + MI_SEMAPHORE_REGISTER); } static struct intel_engine_cs * @@ -41,31 +37,20 @@ semaphore_wait_to_signaller_ring(struct intel_engine_cs *engine, u32 ipehr, u64 offset) { struct drm_i915_private *dev_priv = engine->i915; + u32 sync_bits = ipehr & MI_SEMAPHORE_SYNC_MASK; struct intel_engine_cs *signaller; enum intel_engine_id id; - if (INTEL_GEN(dev_priv) >= 8) { - for_each_engine(signaller, dev_priv, id) { - if (engine == signaller) - continue; - - if (offset == signaller->semaphore.signal_ggtt[engine->hw_id]) - return signaller; - } - } else { - u32 sync_bits = ipehr & MI_SEMAPHORE_SYNC_MASK; - - for_each_engine(signaller, dev_priv, id) { - if(engine == signaller) - continue; + for_each_engine(signaller, dev_priv, id) { + if (engine == signaller) + continue; - if (sync_bits == signaller->semaphore.mbox.wait[engine->hw_id]) - return signaller; - } + if (sync_bits == signaller->semaphore.mbox.wait[engine->hw_id]) + return signaller; } - DRM_DEBUG_DRIVER("No signaller ring found for %s, ipehr 0x%08x, offset 0x%016llx\n", - engine->name, ipehr, offset); + DRM_DEBUG_DRIVER("No signaller ring found for %s, ipehr 0x%08x\n", + engine->name, ipehr); return ERR_PTR(-ENODEV); } @@ -135,11 +120,6 @@ semaphore_waits_for(struct intel_engine_cs *engine, u32 *seqno) return NULL; *seqno = ioread32(vaddr + head + 4) + 1; - if (INTEL_GEN(dev_priv) >= 8) { - offset = ioread32(vaddr + head + 12); - offset <<= 32; - offset |= ioread32(vaddr + head + 8); - } return semaphore_wait_to_signaller_ring(engine, ipehr, offset); } @@ -273,7 +253,7 @@ engine_stuck(struct intel_engine_cs *engine, u64 acthd) return ENGINE_WAIT_KICK; } - if (INTEL_GEN(dev_priv) >= 6 && tmp & RING_WAIT_SEMAPHORE) { + if (IS_GEN(dev_priv, 6, 7) && tmp & RING_WAIT_SEMAPHORE) { switch (semaphore_passed(engine)) { default: return ENGINE_DEAD; diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c index be98868115bf..5465e708545f 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.c +++ b/drivers/gpu/drm/i915/intel_ringbuffer.c @@ -340,50 +340,6 @@ gen7_render_ring_flush(struct drm_i915_gem_request *req, u32 mode) return 0; } -static int -gen8_render_ring_flush(struct drm_i915_gem_request *req, u32 mode) -{ - u32 flags; - u32 *cs; - - cs = intel_ring_begin(req, mode & EMIT_INVALIDATE ? 12 : 6); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - flags = PIPE_CONTROL_CS_STALL; - - if (mode & EMIT_FLUSH) { - flags |= PIPE_CONTROL_RENDER_TARGET_CACHE_FLUSH; - flags |= PIPE_CONTROL_DEPTH_CACHE_FLUSH; - flags |= PIPE_CONTROL_DC_FLUSH_ENABLE; - flags |= PIPE_CONTROL_FLUSH_ENABLE; - } - if (mode & EMIT_INVALIDATE) { - flags |= PIPE_CONTROL_TLB_INVALIDATE; - flags |= PIPE_CONTROL_INSTRUCTION_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_VF_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_CONST_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_STATE_CACHE_INVALIDATE; - flags |= PIPE_CONTROL_QW_WRITE; - flags |= PIPE_CONTROL_GLOBAL_GTT_IVB; - - /* WaCsStallBeforeStateCacheInvalidate:bdw,chv */ - cs = gen8_emit_pipe_control(cs, - PIPE_CONTROL_CS_STALL | - PIPE_CONTROL_STALL_AT_SCOREBOARD, - 0); - } - - cs = gen8_emit_pipe_control(cs, flags, - i915_ggtt_offset(req->engine->scratch) + - 2 * CACHELINE_BYTES); - - intel_ring_advance(req, cs); - - return 0; -} - static void ring_setup_phys_status_page(struct intel_engine_cs *engine) { struct drm_i915_private *dev_priv = engine->i915; @@ -427,7 +383,6 @@ static void intel_ring_setup_status_page(struct intel_engine_cs *engine) } else if (IS_GEN6(dev_priv)) { mmio = RING_HWS_PGA_GEN6(engine->mmio_base); } else { - /* XXX: gen8 returns to sanity */ mmio = RING_HWS_PGA(engine->mmio_base); } @@ -437,13 +392,7 @@ static void intel_ring_setup_status_page(struct intel_engine_cs *engine) I915_WRITE(mmio, engine->status_page.ggtt_offset); POSTING_READ(mmio); - /* - * Flush the TLB for this page - * - * FIXME: These two bits have disappeared on gen8, so a question - * arises: do we still need this and if so how should we go about - * invalidating the TLB? - */ + /* Flush the TLB for this page */ if (IS_GEN(dev_priv, 6, 7)) { i915_reg_t reg = RING_INSTPM(engine->mmio_base); @@ -611,8 +560,6 @@ static void reset_ring_common(struct intel_engine_cs *engine, struct intel_context *ce = &request->ctx->engine[engine->id]; struct i915_hw_ppgtt *ppgtt; - /* FIXME consider gen8 reset */ - if (ce->state) { I915_WRITE(CCID, i915_ggtt_offset(ce->state) | @@ -713,62 +660,6 @@ static int init_render_ring(struct intel_engine_cs *engine) return init_workarounds_ring(engine); } -static void render_ring_cleanup(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - - i915_vma_unpin_and_release(&dev_priv->semaphore); -} - -static u32 *gen8_rcs_signal(struct drm_i915_gem_request *req, u32 *cs) -{ - struct drm_i915_private *dev_priv = req->i915; - struct intel_engine_cs *waiter; - enum intel_engine_id id; - - for_each_engine(waiter, dev_priv, id) { - u64 gtt_offset = req->engine->semaphore.signal_ggtt[id]; - if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID) - continue; - - *cs++ = GFX_OP_PIPE_CONTROL(6); - *cs++ = PIPE_CONTROL_GLOBAL_GTT_IVB | PIPE_CONTROL_QW_WRITE | - PIPE_CONTROL_CS_STALL; - *cs++ = lower_32_bits(gtt_offset); - *cs++ = upper_32_bits(gtt_offset); - *cs++ = req->global_seqno; - *cs++ = 0; - *cs++ = MI_SEMAPHORE_SIGNAL | - MI_SEMAPHORE_TARGET(waiter->hw_id); - *cs++ = 0; - } - - return cs; -} - -static u32 *gen8_xcs_signal(struct drm_i915_gem_request *req, u32 *cs) -{ - struct drm_i915_private *dev_priv = req->i915; - struct intel_engine_cs *waiter; - enum intel_engine_id id; - - for_each_engine(waiter, dev_priv, id) { - u64 gtt_offset = req->engine->semaphore.signal_ggtt[id]; - if (gtt_offset == MI_SEMAPHORE_SYNC_INVALID) - continue; - - *cs++ = (MI_FLUSH_DW + 1) | MI_FLUSH_DW_OP_STOREDW; - *cs++ = lower_32_bits(gtt_offset) | MI_FLUSH_DW_USE_GTT; - *cs++ = upper_32_bits(gtt_offset); - *cs++ = req->global_seqno; - *cs++ = MI_SEMAPHORE_SIGNAL | - MI_SEMAPHORE_TARGET(waiter->hw_id); - *cs++ = 0; - } - - return cs; -} - static u32 *gen6_signal(struct drm_i915_gem_request *req, u32 *cs) { struct drm_i915_private *dev_priv = req->i915; @@ -851,70 +742,6 @@ static void gen6_sema_emit_breadcrumb(struct drm_i915_gem_request *req, u32 *cs) req->engine->semaphore.signal(req, cs)); } -static void gen8_render_emit_breadcrumb(struct drm_i915_gem_request *req, - u32 *cs) -{ - struct intel_engine_cs *engine = req->engine; - - if (engine->semaphore.signal) - cs = engine->semaphore.signal(req, cs); - - *cs++ = GFX_OP_PIPE_CONTROL(6); - *cs++ = PIPE_CONTROL_GLOBAL_GTT_IVB | PIPE_CONTROL_CS_STALL | - PIPE_CONTROL_QW_WRITE; - *cs++ = intel_hws_seqno_address(engine); - *cs++ = 0; - *cs++ = req->global_seqno; - /* We're thrashing one dword of HWS. */ - *cs++ = 0; - *cs++ = MI_USER_INTERRUPT; - *cs++ = MI_NOOP; - - req->tail = intel_ring_offset(req, cs); - assert_ring_tail_valid(req->ring, req->tail); -} - -static const int gen8_render_emit_breadcrumb_sz = 8; - -/** - * intel_ring_sync - sync the waiter to the signaller on seqno - * - * @waiter - ring that is waiting - * @signaller - ring which has, or will signal - * @seqno - seqno which the waiter will block on - */ - -static int -gen8_ring_sync_to(struct drm_i915_gem_request *req, - struct drm_i915_gem_request *signal) -{ - struct drm_i915_private *dev_priv = req->i915; - u64 offset = GEN8_WAIT_OFFSET(req->engine, signal->engine->id); - struct i915_hw_ppgtt *ppgtt; - u32 *cs; - - cs = intel_ring_begin(req, 4); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - *cs++ = MI_SEMAPHORE_WAIT | MI_SEMAPHORE_GLOBAL_GTT | - MI_SEMAPHORE_SAD_GTE_SDD; - *cs++ = signal->global_seqno; - *cs++ = lower_32_bits(offset); - *cs++ = upper_32_bits(offset); - intel_ring_advance(req, cs); - - /* When the !RCS engines idle waiting upon a semaphore, they lose their - * pagetables and we must reload them before executing the batch. - * We do this on the i915_switch_context() following the wait and - * before the dispatch. - */ - ppgtt = req->ctx->ppgtt; - if (ppgtt && req->engine->id != RCS) - ppgtt->pd_dirty_rings |= intel_engine_flag(req->engine); - return 0; -} - static int gen6_ring_sync_to(struct drm_i915_gem_request *req, struct drm_i915_gem_request *signal) @@ -1090,25 +917,6 @@ hsw_vebox_irq_disable(struct intel_engine_cs *engine) gen6_mask_pm_irq(dev_priv, engine->irq_enable_mask); } -static void -gen8_irq_enable(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - - I915_WRITE_IMR(engine, - ~(engine->irq_enable_mask | - engine->irq_keep_mask)); - POSTING_READ_FW(RING_IMR(engine->mmio_base)); -} - -static void -gen8_irq_disable(struct intel_engine_cs *engine) -{ - struct drm_i915_private *dev_priv = engine->i915; - - I915_WRITE_IMR(engine, ~engine->irq_keep_mask); -} - static int i965_emit_bb_start(struct drm_i915_gem_request *req, u64 offset, u32 length, @@ -1796,8 +1604,6 @@ static int gen6_bsd_ring_flush(struct drm_i915_gem_request *req, u32 mode) return PTR_ERR(cs); cmd = MI_FLUSH_DW; - if (INTEL_GEN(req->i915) >= 8) - cmd += 1; /* We always require a command barrier so that subsequent * commands, such as breadcrumb interrupts, are strictly ordered @@ -1817,38 +1623,9 @@ static int gen6_bsd_ring_flush(struct drm_i915_gem_request *req, u32 mode) *cs++ = cmd; *cs++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT; - if (INTEL_GEN(req->i915) >= 8) { - *cs++ = 0; /* upper addr */ - *cs++ = 0; /* value */ - } else { - *cs++ = 0; - *cs++ = MI_NOOP; - } - intel_ring_advance(req, cs); - return 0; -} - -static int -gen8_emit_bb_start(struct drm_i915_gem_request *req, - u64 offset, u32 len, - unsigned int dispatch_flags) -{ - bool ppgtt = USES_PPGTT(req->i915) && - !(dispatch_flags & I915_DISPATCH_SECURE); - u32 *cs; - - cs = intel_ring_begin(req, 4); - if (IS_ERR(cs)) - return PTR_ERR(cs); - - /* FIXME(BDW): Address space and security selectors. */ - *cs++ = MI_BATCH_BUFFER_START_GEN8 | (ppgtt << 8) | (dispatch_flags & - I915_DISPATCH_RS ? MI_BATCH_RESOURCE_STREAMER : 0); - *cs++ = lower_32_bits(offset); - *cs++ = upper_32_bits(offset); + *cs++ = 0; *cs++ = MI_NOOP; intel_ring_advance(req, cs); - return 0; } @@ -1905,8 +1682,6 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req, u32 mode) return PTR_ERR(cs); cmd = MI_FLUSH_DW; - if (INTEL_GEN(req->i915) >= 8) - cmd += 1; /* We always require a command barrier so that subsequent * commands, such as breadcrumb interrupts, are strictly ordered @@ -1925,13 +1700,8 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req, u32 mode) cmd |= MI_INVALIDATE_TLB; *cs++ = cmd; *cs++ = I915_GEM_HWS_SCRATCH_ADDR | MI_FLUSH_DW_USE_GTT; - if (INTEL_GEN(req->i915) >= 8) { - *cs++ = 0; /* upper addr */ - *cs++ = 0; /* value */ - } else { - *cs++ = 0; - *cs++ = MI_NOOP; - } + *cs++ = 0; + *cs++ = MI_NOOP; intel_ring_advance(req, cs); return 0; @@ -1940,110 +1710,61 @@ static int gen6_ring_flush(struct drm_i915_gem_request *req, u32 mode) static void intel_ring_init_semaphores(struct drm_i915_private *dev_priv, struct intel_engine_cs *engine) { - struct drm_i915_gem_object *obj; - int ret, i; + int i; if (!i915_modparams.semaphores) return; - if (INTEL_GEN(dev_priv) >= 8 && !dev_priv->semaphore) { - struct i915_vma *vma; - - obj = i915_gem_object_create(dev_priv, PAGE_SIZE); - if (IS_ERR(obj)) - goto err; - - vma = i915_vma_instance(obj, &dev_priv->ggtt.base, NULL); - if (IS_ERR(vma)) - goto err_obj; - - ret = i915_gem_object_set_to_gtt_domain(obj, false); - if (ret) - goto err_obj; - - ret = i915_vma_pin(vma, 0, 0, PIN_GLOBAL | PIN_HIGH); - if (ret) - goto err_obj; + GEM_BUG_ON(INTEL_GEN(dev_priv) < 6); + engine->semaphore.sync_to = gen6_ring_sync_to; + engine->semaphore.signal = gen6_signal; - dev_priv->semaphore = vma; - } - - if (INTEL_GEN(dev_priv) >= 8) { - u32 offset = i915_ggtt_offset(dev_priv->semaphore); - - engine->semaphore.sync_to = gen8_ring_sync_to; - engine->semaphore.signal = gen8_xcs_signal; - - for (i = 0; i < I915_NUM_ENGINES; i++) { - u32 ring_offset; - - if (i != engine->id) - ring_offset = offset + GEN8_SEMAPHORE_OFFSET(engine->id, i); - else - ring_offset = MI_SEMAPHORE_SYNC_INVALID; - - engine->semaphore.signal_ggtt[i] = ring_offset; - } - } else if (INTEL_GEN(dev_priv) >= 6) { - engine->semaphore.sync_to = gen6_ring_sync_to; - engine->semaphore.signal = gen6_signal; - - /* - * The current semaphore is only applied on pre-gen8 - * platform. And there is no VCS2 ring on the pre-gen8 - * platform. So the semaphore between RCS and VCS2 is - * initialized as INVALID. Gen8 will initialize the - * sema between VCS2 and RCS later. - */ - for (i = 0; i < GEN6_NUM_SEMAPHORES; i++) { - static const struct { - u32 wait_mbox; - i915_reg_t mbox_reg; - } sem_data[GEN6_NUM_SEMAPHORES][GEN6_NUM_SEMAPHORES] = { - [RCS_HW] = { - [VCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_RV, .mbox_reg = GEN6_VRSYNC }, - [BCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_RB, .mbox_reg = GEN6_BRSYNC }, - [VECS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_RVE, .mbox_reg = GEN6_VERSYNC }, - }, - [VCS_HW] = { - [RCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VR, .mbox_reg = GEN6_RVSYNC }, - [BCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VB, .mbox_reg = GEN6_BVSYNC }, - [VECS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VVE, .mbox_reg = GEN6_VEVSYNC }, - }, - [BCS_HW] = { - [RCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_BR, .mbox_reg = GEN6_RBSYNC }, - [VCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_BV, .mbox_reg = GEN6_VBSYNC }, - [VECS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_BVE, .mbox_reg = GEN6_VEBSYNC }, - }, - [VECS_HW] = { - [RCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VER, .mbox_reg = GEN6_RVESYNC }, - [VCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VEV, .mbox_reg = GEN6_VVESYNC }, - [BCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VEB, .mbox_reg = GEN6_BVESYNC }, - }, - }; + /* + * The current semaphore is only applied on pre-gen8 + * platform. And there is no VCS2 ring on the pre-gen8 + * platform. So the semaphore between RCS and VCS2 is + * initialized as INVALID. + */ + for (i = 0; i < GEN6_NUM_SEMAPHORES; i++) { + static const struct { u32 wait_mbox; i915_reg_t mbox_reg; + } sem_data[GEN6_NUM_SEMAPHORES][GEN6_NUM_SEMAPHORES] = { + [RCS_HW] = { + [VCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_RV, .mbox_reg = GEN6_VRSYNC }, + [BCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_RB, .mbox_reg = GEN6_BRSYNC }, + [VECS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_RVE, .mbox_reg = GEN6_VERSYNC }, + }, + [VCS_HW] = { + [RCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VR, .mbox_reg = GEN6_RVSYNC }, + [BCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VB, .mbox_reg = GEN6_BVSYNC }, + [VECS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VVE, .mbox_reg = GEN6_VEVSYNC }, + }, + [BCS_HW] = { + [RCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_BR, .mbox_reg = GEN6_RBSYNC }, + [VCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_BV, .mbox_reg = GEN6_VBSYNC }, + [VECS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_BVE, .mbox_reg = GEN6_VEBSYNC }, + }, + [VECS_HW] = { + [RCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VER, .mbox_reg = GEN6_RVESYNC }, + [VCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VEV, .mbox_reg = GEN6_VVESYNC }, + [BCS_HW] = { .wait_mbox = MI_SEMAPHORE_SYNC_VEB, .mbox_reg = GEN6_BVESYNC }, + }, + }; + u32 wait_mbox; + i915_reg_t mbox_reg; - if (i == engine->hw_id) { - wait_mbox = MI_SEMAPHORE_SYNC_INVALID; - mbox_reg = GEN6_NOSYNC; - } else { - wait_mbox = sem_data[engine->hw_id][i].wait_mbox; - mbox_reg = sem_data[engine->hw_id][i].mbox_reg; - } - - engine->semaphore.mbox.wait[i] = wait_mbox; - engine->semaphore.mbox.signal[i] = mbox_reg; + if (i == engine->hw_id) { + wait_mbox = MI_SEMAPHORE_SYNC_INVALID; + mbox_reg = GEN6_NOSYNC; + } else { + wait_mbox = sem_data[engine->hw_id][i].wait_mbox; + mbox_reg = sem_data[engine->hw_id][i].mbox_reg; } - } - - return; -err_obj: - i915_gem_object_put(obj); -err: - DRM_DEBUG_DRIVER("Failed to allocate space for semaphores, disabling\n"); - i915_modparams.semaphores = 0; + engine->semaphore.mbox.wait[i] = wait_mbox; + engine->semaphore.mbox.signal[i] = mbox_reg; + } } static void intel_ring_init_irq(struct drm_i915_private *dev_priv, @@ -2051,11 +1772,7 @@ static void intel_ring_init_irq(struct drm_i915_private *dev_priv, { engine->irq_enable_mask = GT_RENDER_USER_INTERRUPT << engine->irq_shift; - if (INTEL_GEN(dev_priv) >= 8) { - engine->irq_enable = gen8_irq_enable; - engine->irq_disable = gen8_irq_disable; - engine->irq_seqno_barrier = gen6_seqno_barrier; - } else if (INTEL_GEN(dev_priv) >= 6) { + if (INTEL_GEN(dev_priv) >= 6) { engine->irq_enable = gen6_irq_enable; engine->irq_disable = gen6_irq_disable; engine->irq_seqno_barrier = gen6_seqno_barrier; @@ -2090,6 +1807,9 @@ static void gen6_bsd_set_default_submission(struct intel_engine_cs *engine) static void intel_ring_default_vfuncs(struct drm_i915_private *dev_priv, struct intel_engine_cs *engine) { + /* gen8+ are only supported with execlists */ + GEM_BUG_ON(INTEL_GEN(dev_priv) >= 8); + intel_ring_init_irq(dev_priv, engine); intel_ring_init_semaphores(dev_priv, engine); @@ -2109,20 +1829,14 @@ static void intel_ring_default_vfuncs(struct drm_i915_private *dev_priv, engine->emit_breadcrumb = gen6_sema_emit_breadcrumb; num_rings = INTEL_INFO(dev_priv)->num_rings - 1; - if (INTEL_GEN(dev_priv) >= 8) { - engine->emit_breadcrumb_sz += num_rings * 6; - } else { - engine->emit_breadcrumb_sz += num_rings * 3; - if (num_rings & 1) - engine->emit_breadcrumb_sz++; - } + engine->emit_breadcrumb_sz += num_rings * 3; + if (num_rings & 1) + engine->emit_breadcrumb_sz++; } engine->set_default_submission = i9xx_set_default_submission; - if (INTEL_GEN(dev_priv) >= 8) - engine->emit_bb_start = gen8_emit_bb_start; - else if (INTEL_GEN(dev_priv) >= 6) + if (INTEL_GEN(dev_priv) >= 6) engine->emit_bb_start = gen6_emit_bb_start; else if (INTEL_GEN(dev_priv) >= 4) engine->emit_bb_start = i965_emit_bb_start; @@ -2142,20 +1856,7 @@ int intel_init_render_ring_buffer(struct intel_engine_cs *engine) if (HAS_L3_DPF(dev_priv)) engine->irq_keep_mask = GT_RENDER_L3_PARITY_ERROR_INTERRUPT; - if (INTEL_GEN(dev_priv) >= 8) { - engine->init_context = intel_rcs_ctx_init; - engine->emit_breadcrumb = gen8_render_emit_breadcrumb; - engine->emit_breadcrumb_sz = gen8_render_emit_breadcrumb_sz; - engine->emit_flush = gen8_render_ring_flush; - if (i915_modparams.semaphores) { - int num_rings; - - engine->semaphore.signal = gen8_rcs_signal; - - num_rings = INTEL_INFO(dev_priv)->num_rings - 1; - engine->emit_breadcrumb_sz += num_rings * 8; - } - } else if (INTEL_GEN(dev_priv) >= 6) { + if (INTEL_GEN(dev_priv) >= 6) { engine->init_context = intel_rcs_ctx_init; engine->emit_flush = gen7_render_ring_flush; if (IS_GEN6(dev_priv)) @@ -2174,7 +1875,6 @@ int intel_init_render_ring_buffer(struct intel_engine_cs *engine) engine->emit_bb_start = hsw_emit_bb_start; engine->init_hw = init_render_ring; - engine->cleanup = render_ring_cleanup; ret = intel_init_ring_buffer(engine); if (ret) @@ -2204,8 +1904,7 @@ int intel_init_bsd_ring_buffer(struct intel_engine_cs *engine) if (IS_GEN6(dev_priv)) engine->set_default_submission = gen6_bsd_set_default_submission; engine->emit_flush = gen6_bsd_ring_flush; - if (INTEL_GEN(dev_priv) < 8) - engine->irq_enable_mask = GT_BSD_USER_INTERRUPT; + engine->irq_enable_mask = GT_BSD_USER_INTERRUPT; } else { engine->mmio_base = BSD_RING_BASE; engine->emit_flush = bsd_ring_flush; @@ -2225,8 +1924,7 @@ int intel_init_blt_ring_buffer(struct intel_engine_cs *engine) intel_ring_default_vfuncs(dev_priv, engine); engine->emit_flush = gen6_ring_flush; - if (INTEL_GEN(dev_priv) < 8) - engine->irq_enable_mask = GT_BLT_USER_INTERRUPT; + engine->irq_enable_mask = GT_BLT_USER_INTERRUPT; return intel_init_ring_buffer(engine); } @@ -2238,12 +1936,9 @@ int intel_init_vebox_ring_buffer(struct intel_engine_cs *engine) intel_ring_default_vfuncs(dev_priv, engine); engine->emit_flush = gen6_ring_flush; - - if (INTEL_GEN(dev_priv) < 8) { - engine->irq_enable_mask = PM_VEBOX_USER_INTERRUPT; - engine->irq_enable = hsw_vebox_irq_enable; - engine->irq_disable = hsw_vebox_irq_disable; - } + engine->irq_enable_mask = PM_VEBOX_USER_INTERRUPT; + engine->irq_enable = hsw_vebox_irq_enable; + engine->irq_disable = hsw_vebox_irq_disable; return intel_init_ring_buffer(engine); } diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h index e437a9556c08..80cd7812ce02 100644 --- a/drivers/gpu/drm/i915/intel_ringbuffer.h +++ b/drivers/gpu/drm/i915/intel_ringbuffer.h @@ -46,16 +46,6 @@ struct intel_hw_status_page { /* seqno size is actually only a uint32, but since we plan to use MI_FLUSH_DW to * do the writes, and that must have qw aligned offsets, simply pretend it's 8b. */ -#define gen8_semaphore_seqno_size sizeof(uint64_t) -#define GEN8_SEMAPHORE_OFFSET(__from, __to) \ - (((__from) * I915_NUM_ENGINES + (__to)) * gen8_semaphore_seqno_size) -#define GEN8_SIGNAL_OFFSET(__ring, to) \ - (dev_priv->semaphore->node.start + \ - GEN8_SEMAPHORE_OFFSET((__ring)->id, (to))) -#define GEN8_WAIT_OFFSET(__ring, from) \ - (dev_priv->semaphore->node.start + \ - GEN8_SEMAPHORE_OFFSET(from, (__ring)->id)) - enum intel_engine_hangcheck_action { ENGINE_IDLE = 0, ENGINE_WAIT, @@ -467,18 +457,15 @@ struct intel_engine_cs { * ie. transpose of f(x, y) */ struct { - union { #define GEN6_SEMAPHORE_LAST VECS_HW #define GEN6_NUM_SEMAPHORES (GEN6_SEMAPHORE_LAST + 1) #define GEN6_SEMAPHORES_MASK GENMASK(GEN6_SEMAPHORE_LAST, 0) - struct { - /* our mbox written by others */ - u32 wait[GEN6_NUM_SEMAPHORES]; - /* mboxes this ring signals to */ - i915_reg_t signal[GEN6_NUM_SEMAPHORES]; - } mbox; - u64 signal_ggtt[I915_NUM_ENGINES]; - }; + struct { + /* our mbox written by others */ + u32 wait[GEN6_NUM_SEMAPHORES]; + /* mboxes this ring signals to */ + i915_reg_t signal[GEN6_NUM_SEMAPHORES]; + } mbox; /* AKA wait() */ int (*sync_to)(struct drm_i915_gem_request *req, -- GitLab