diff options
author | Linus Torvalds <torvalds@linux-foundation.org> | 2018-04-02 07:59:23 -0700 |
---|---|---|
committer | Linus Torvalds <torvalds@linux-foundation.org> | 2018-04-02 07:59:23 -0700 |
commit | 320b164abb32db876866a4ff8c2cb710524ac6ea (patch) | |
tree | 1f79119cde6e24c9f1d01fb1e51252bca7c4cdd5 /drivers/gpu/drm/i915/intel_guc_submission.c | |
parent | 0adb32858b0bddf4ada5f364a84ed60b196dbcda (diff) | |
parent | 694f54f680f7fd8e9561928fbfc537d9afbc3d79 (diff) | |
download | linux-320b164abb32db876866a4ff8c2cb710524ac6ea.tar.gz linux-320b164abb32db876866a4ff8c2cb710524ac6ea.tar.bz2 linux-320b164abb32db876866a4ff8c2cb710524ac6ea.zip |
Merge tag 'drm-for-v4.17' of git://people.freedesktop.org/~airlied/linux
Pull drm updates from Dave Airlie:
"Cannonlake and Vega12 support are probably the two major things. This
pull lacks nouveau, Ben had some unforseen leave and a few other
blockers so we'll see how things look or maybe leave it for this merge
window.
core:
- Device links to handle sound/gpu pm dependency
- Color encoding/range properties
- Plane clipping into plane check helper
- Backlight helpers
- DP TP4 + HBR3 helper support
amdgpu:
- Vega12 support
- Enable DC by default on all supported GPUs
- Powerplay restructuring and cleanup
- DC bandwidth calc updates
- DC backlight on pre-DCE11
- TTM backing store dropping support
- SR-IOV fixes
- Adding "wattman" like functionality
- DC crc support
- Improved DC dual-link handling
amdkfd:
- GPUVM support for dGPU
- KFD events for dGPU
- Enable PCIe atomics for dGPUs
- HSA process eviction support
- Live-lock fixes for process eviction
- VM page table allocation fix for large-bar systems
panel:
- Raydium RM68200
- AUO G104SN02 V2
- KEO TX31D200VM0BAA
- ARM Versatile panels
i915:
- Cannonlake support enabled
- AUX-F port support added
- Icelake base enabling until internal milestone of forcewake support
- Query uAPI interface (used for GPU topology information currently)
- Compressed framebuffer support for sprites
- kmem cache shrinking when GPU is idle
- Avoid boosting GPU when waited item is being processed already
- Avoid retraining LSPCON link unnecessarily
- Decrease request signaling latency
- Deprecation of I915_SET_COLORKEY_NONE
- Kerneldoc and compiler warning cleanup for upcoming CI enforcements
- Full range ycbcr toggling
- HDCP support
i915/gvt:
- Big refactor for shadow ppgtt
- KBL context save/restore via LRI cmd (Weinan)
- Properly unmap dma for guest page (Changbin)
vmwgfx:
- Lots of various improvements
etnaviv:
- Use the drm gpu scheduler
- prep work for GC7000L support
vc4:
- fix alpha blending
- Expose perf counters to userspace
pl111:
- Bandwidth checking/limiting
- Versatile panel support
sun4i:
- A83T HDMI support
- A80 support
- YUV plane support
- H3/H5 HDMI support
omapdrm:
- HPD support for DVI connector
- remove lots of static variables
msm:
- DSI updates from 10nm / SDM845
- fix for race condition with a3xx/a4xx fence completion irq
- some refactoring/prep work for eventual a6xx support (ie. when we
have a userspace)
- a5xx debugfs enhancements
- some mdp5 fixes/cleanups to prepare for eventually merging
writeback
- support (ie. when we have a userspace)
tegra:
- mmap() fixes for fbdev devices
- Overlay plane for hw cursor fix
- dma-buf cache maintenance support
mali-dp:
- YUV->RGB conversion support
rockchip:
- rk3399/chromebook fixes and improvements
rcar-du:
- LVDS support move to drm bridge
- DT bindings for R8A77995
- Driver/DT support for R8A77970
tilcdc:
- DRM panel support"
* tag 'drm-for-v4.17' of git://people.freedesktop.org/~airlied/linux: (1646 commits)
drm/i915: Fix hibernation with ACPI S0 target state
drm/i915/execlists: Use a locked clear_bit() for synchronisation with interrupt
drm/i915: Specify which engines to reset following semaphore/event lockups
drm/i915/dp: Write to SET_POWER dpcd to enable MST hub.
drm/amdkfd: Use ordered workqueue to restore processes
drm/amdgpu: Fix acquiring VM on large-BAR systems
drm/amd/pp: clean header file hwmgr.h
drm/amd/pp: use mlck_table.count for array loop index limit
drm: Fix uabi regression by allowing garbage mode->type from userspace
drm/amdgpu: Add an ATPX quirk for hybrid laptop
drm/amdgpu: fix spelling mistake: "asssert" -> "assert"
drm/amd/pp: Add new asic support in pp_psm.c
drm/amd/pp: Clean up powerplay code on Vega12
drm/amd/pp: Add smu irq handlers for legacy asics
drm/amd/pp: Fix set wrong temperature range on smu7
drm/amdgpu: Don't change preferred domian when fallback GTT v5
drm/vmwgfx: Bump version patchlevel and date
drm/vmwgfx: use monotonic event timestamps
drm/vmwgfx: Unpin the screen object backup buffer when not used
drm/vmwgfx: Stricter count of legacy surface device resources
...
Diffstat (limited to 'drivers/gpu/drm/i915/intel_guc_submission.c')
-rw-r--r-- | drivers/gpu/drm/i915/intel_guc_submission.c | 331 |
1 files changed, 127 insertions, 204 deletions
diff --git a/drivers/gpu/drm/i915/intel_guc_submission.c b/drivers/gpu/drm/i915/intel_guc_submission.c index 4d2409466a3a..8a8ad2fe158d 100644 --- a/drivers/gpu/drm/i915/intel_guc_submission.c +++ b/drivers/gpu/drm/i915/intel_guc_submission.c @@ -26,8 +26,14 @@ #include <trace/events/dma_fence.h> #include "intel_guc_submission.h" +#include "intel_lrc_reg.h" #include "i915_drv.h" +#define GUC_PREEMPT_FINISHED 0x1 +#define GUC_PREEMPT_BREADCRUMB_DWORDS 0x8 +#define GUC_PREEMPT_BREADCRUMB_BYTES \ + (sizeof(u32) * GUC_PREEMPT_BREADCRUMB_DWORDS) + /** * DOC: GuC-based command submission * @@ -73,15 +79,13 @@ * ELSP context descriptor dword into Work Item. * See guc_add_request() * - * ADS: - * The Additional Data Struct (ADS) has pointers for different buffers used by - * the GuC. One single gem object contains the ADS struct itself (guc_ads), the - * scheduling policies (guc_policies), a structure describing a collection of - * register sets (guc_mmio_reg_state) and some extra pages for the GuC to save - * its internal state for sleep. - * */ +static inline struct i915_priolist *to_priolist(struct rb_node *rb) +{ + return rb_entry(rb, struct i915_priolist, node); +} + static inline bool is_high_priority(struct intel_guc_client *client) { return (client->priority == GUC_CLIENT_PRIORITY_KMD_HIGH || @@ -503,8 +507,7 @@ static void guc_ring_doorbell(struct intel_guc_client *client) GEM_BUG_ON(db->db_status != GUC_DOORBELL_ENABLED); } -static void guc_add_request(struct intel_guc *guc, - struct drm_i915_gem_request *rq) +static void guc_add_request(struct intel_guc *guc, struct i915_request *rq) { struct intel_guc_client *client = guc->execbuf_client; struct intel_engine_cs *engine = rq->engine; @@ -538,8 +541,6 @@ static void flush_ggtt_writes(struct i915_vma *vma) POSTING_READ_FW(GUC_STATUS); } -#define GUC_PREEMPT_FINISHED 0x1 -#define GUC_PREEMPT_BREADCRUMB_DWORDS 0x8 static void inject_preempt_context(struct work_struct *work) { struct guc_preempt_work *preempt_work = @@ -549,37 +550,17 @@ static void inject_preempt_context(struct work_struct *work) preempt_work[engine->id]); struct intel_guc_client *client = guc->preempt_client; struct guc_stage_desc *stage_desc = __get_stage_desc(client); - struct intel_ring *ring = client->owner->engine[engine->id].ring; u32 ctx_desc = lower_32_bits(intel_lr_context_descriptor(client->owner, engine)); - u32 *cs = ring->vaddr + ring->tail; u32 data[7]; - if (engine->id == RCS) { - cs = gen8_emit_ggtt_write_rcs(cs, GUC_PREEMPT_FINISHED, - intel_hws_preempt_done_address(engine)); - } else { - cs = gen8_emit_ggtt_write(cs, GUC_PREEMPT_FINISHED, - intel_hws_preempt_done_address(engine)); - *cs++ = MI_NOOP; - *cs++ = MI_NOOP; - } - *cs++ = MI_USER_INTERRUPT; - *cs++ = MI_NOOP; - - GEM_BUG_ON(!IS_ALIGNED(ring->size, - GUC_PREEMPT_BREADCRUMB_DWORDS * sizeof(u32))); - GEM_BUG_ON((void *)cs - (ring->vaddr + ring->tail) != - GUC_PREEMPT_BREADCRUMB_DWORDS * sizeof(u32)); - - ring->tail += GUC_PREEMPT_BREADCRUMB_DWORDS * sizeof(u32); - ring->tail &= (ring->size - 1); - - flush_ggtt_writes(ring->vma); - + /* + * The ring contains commands to write GUC_PREEMPT_FINISHED into HWSP. + * See guc_fill_preempt_context(). + */ spin_lock_irq(&client->wq_lock); guc_wq_item_append(client, engine->guc_id, ctx_desc, - ring->tail / sizeof(u64), 0); + GUC_PREEMPT_BREADCRUMB_BYTES / sizeof(u64), 0); spin_unlock_irq(&client->wq_lock); /* @@ -655,7 +636,7 @@ static void guc_submit(struct intel_engine_cs *engine) unsigned int n; for (n = 0; n < execlists_num_ports(execlists); n++) { - struct drm_i915_gem_request *rq; + struct i915_request *rq; unsigned int count; rq = port_unpack(&port[n], &count); @@ -669,19 +650,18 @@ static void guc_submit(struct intel_engine_cs *engine) } } -static void port_assign(struct execlist_port *port, - struct drm_i915_gem_request *rq) +static void port_assign(struct execlist_port *port, struct i915_request *rq) { GEM_BUG_ON(port_isset(port)); - port_set(port, i915_gem_request_get(rq)); + port_set(port, i915_request_get(rq)); } static void guc_dequeue(struct intel_engine_cs *engine) { struct intel_engine_execlists * const execlists = &engine->execlists; struct execlist_port *port = execlists->port; - struct drm_i915_gem_request *last = NULL; + struct i915_request *last = NULL; const struct execlist_port * const last_port = &execlists->port[execlists->port_mask]; bool submit = false; @@ -691,15 +671,12 @@ static void guc_dequeue(struct intel_engine_cs *engine) rb = execlists->first; GEM_BUG_ON(rb_first(&execlists->queue) != rb); - if (!rb) - goto unlock; - if (port_isset(port)) { - if (HAS_LOGICAL_RING_PREEMPTION(engine->i915)) { + if (engine->i915->preempt_context) { struct guc_preempt_work *preempt_work = &engine->i915->guc.preempt_work[engine->id]; - if (rb_entry(rb, struct i915_priolist, node)->priority > + if (execlists->queue_priority > max(port_request(port)->priotree.priority, 0)) { execlists_set_active(execlists, EXECLISTS_ACTIVE_PREEMPT); @@ -715,9 +692,9 @@ static void guc_dequeue(struct intel_engine_cs *engine) } GEM_BUG_ON(port_isset(port)); - do { - struct i915_priolist *p = rb_entry(rb, typeof(*p), node); - struct drm_i915_gem_request *rq, *rn; + while (rb) { + struct i915_priolist *p = to_priolist(rb); + struct i915_request *rq, *rn; list_for_each_entry_safe(rq, rn, &p->requests, priotree.link) { if (last && rq->ctx != last->ctx) { @@ -734,9 +711,8 @@ static void guc_dequeue(struct intel_engine_cs *engine) INIT_LIST_HEAD(&rq->priotree.link); - __i915_gem_request_submit(rq); - trace_i915_gem_request_in(rq, - port_index(port, execlists)); + __i915_request_submit(rq); + trace_i915_request_in(rq, port_index(port, execlists)); last = rq; submit = true; } @@ -746,14 +722,21 @@ static void guc_dequeue(struct intel_engine_cs *engine) INIT_LIST_HEAD(&p->requests); if (p->priority != I915_PRIORITY_NORMAL) kmem_cache_free(engine->i915->priorities, p); - } while (rb); + } done: + execlists->queue_priority = rb ? to_priolist(rb)->priority : INT_MIN; execlists->first = rb; if (submit) { port_assign(port, last); execlists_set_active(execlists, EXECLISTS_ACTIVE_USER); guc_submit(engine); } + + /* We must always keep the beast fed if we have work piled up */ + GEM_BUG_ON(port_isset(execlists->port) && + !execlists_is_active(execlists, EXECLISTS_ACTIVE_USER)); + GEM_BUG_ON(execlists->first && !port_isset(execlists->port)); + unlock: spin_unlock_irq(&engine->timeline->lock); } @@ -763,12 +746,12 @@ static void guc_submission_tasklet(unsigned long data) struct intel_engine_cs * const engine = (struct intel_engine_cs *)data; struct intel_engine_execlists * const execlists = &engine->execlists; struct execlist_port *port = execlists->port; - struct drm_i915_gem_request *rq; + struct i915_request *rq; rq = port_request(&port[0]); - while (rq && i915_gem_request_completed(rq)) { - trace_i915_gem_request_out(rq); - i915_gem_request_put(rq); + while (rq && i915_request_completed(rq)) { + trace_i915_request_out(rq); + i915_request_put(rq); execlists_port_complete(execlists, port); @@ -839,10 +822,12 @@ static int guc_clients_doorbell_init(struct intel_guc *guc) if (ret) return ret; - ret = create_doorbell(guc->preempt_client); - if (ret) { - destroy_doorbell(guc->execbuf_client); - return ret; + if (guc->preempt_client) { + ret = create_doorbell(guc->preempt_client); + if (ret) { + destroy_doorbell(guc->execbuf_client); + return ret; + } } return 0; @@ -855,8 +840,11 @@ static void guc_clients_doorbell_fini(struct intel_guc *guc) * Instead of trying (in vain) to communicate with it, let's just * cleanup the doorbell HW and our internal state. */ - __destroy_doorbell(guc->preempt_client); - __update_doorbell_desc(guc->preempt_client, GUC_DOORBELL_INVALID); + if (guc->preempt_client) { + __destroy_doorbell(guc->preempt_client); + __update_doorbell_desc(guc->preempt_client, + GUC_DOORBELL_INVALID); + } __destroy_doorbell(guc->execbuf_client); __update_doorbell_desc(guc->execbuf_client, GUC_DOORBELL_INVALID); } @@ -968,6 +956,62 @@ static void guc_client_free(struct intel_guc_client *client) kfree(client); } +static inline bool ctx_save_restore_disabled(struct intel_context *ce) +{ + u32 sr = ce->lrc_reg_state[CTX_CONTEXT_CONTROL + 1]; + +#define SR_DISABLED \ + _MASKED_BIT_ENABLE(CTX_CTRL_ENGINE_CTX_RESTORE_INHIBIT | \ + CTX_CTRL_ENGINE_CTX_SAVE_INHIBIT) + + return (sr & SR_DISABLED) == SR_DISABLED; + +#undef SR_DISABLED +} + +static void guc_fill_preempt_context(struct intel_guc *guc) +{ + struct drm_i915_private *dev_priv = guc_to_i915(guc); + struct intel_guc_client *client = guc->preempt_client; + struct intel_engine_cs *engine; + enum intel_engine_id id; + + for_each_engine(engine, dev_priv, id) { + struct intel_context *ce = &client->owner->engine[id]; + u32 addr = intel_hws_preempt_done_address(engine); + u32 *cs; + + GEM_BUG_ON(!ce->pin_count); + + /* + * We rely on this context image *not* being saved after + * preemption. This ensures that the RING_HEAD / RING_TAIL + * remain pointing at initial values forever. + */ + GEM_BUG_ON(!ctx_save_restore_disabled(ce)); + + cs = ce->ring->vaddr; + if (id == RCS) { + cs = gen8_emit_ggtt_write_rcs(cs, + GUC_PREEMPT_FINISHED, + addr); + } else { + cs = gen8_emit_ggtt_write(cs, + GUC_PREEMPT_FINISHED, + addr); + *cs++ = MI_NOOP; + *cs++ = MI_NOOP; + } + *cs++ = MI_USER_INTERRUPT; + *cs++ = MI_NOOP; + + GEM_BUG_ON((void *)cs - ce->ring->vaddr != + GUC_PREEMPT_BREADCRUMB_BYTES); + + flush_ggtt_writes(ce->ring->vma); + } +} + static int guc_clients_create(struct intel_guc *guc) { struct drm_i915_private *dev_priv = guc_to_i915(guc); @@ -986,17 +1030,21 @@ static int guc_clients_create(struct intel_guc *guc) } guc->execbuf_client = client; - client = guc_client_alloc(dev_priv, - INTEL_INFO(dev_priv)->ring_mask, - GUC_CLIENT_PRIORITY_KMD_HIGH, - dev_priv->preempt_context); - if (IS_ERR(client)) { - DRM_ERROR("Failed to create GuC client for preemption!\n"); - guc_client_free(guc->execbuf_client); - guc->execbuf_client = NULL; - return PTR_ERR(client); + if (dev_priv->preempt_context) { + client = guc_client_alloc(dev_priv, + INTEL_INFO(dev_priv)->ring_mask, + GUC_CLIENT_PRIORITY_KMD_HIGH, + dev_priv->preempt_context); + if (IS_ERR(client)) { + DRM_ERROR("Failed to create GuC client for preemption!\n"); + guc_client_free(guc->execbuf_client); + guc->execbuf_client = NULL; + return PTR_ERR(client); + } + guc->preempt_client = client; + + guc_fill_preempt_context(guc); } - guc->preempt_client = client; return 0; } @@ -1005,122 +1053,12 @@ static void guc_clients_destroy(struct intel_guc *guc) { struct intel_guc_client *client; - client = fetch_and_zero(&guc->execbuf_client); - guc_client_free(client); - client = fetch_and_zero(&guc->preempt_client); - guc_client_free(client); -} - -static void guc_policy_init(struct guc_policy *policy) -{ - policy->execution_quantum = POLICY_DEFAULT_EXECUTION_QUANTUM_US; - policy->preemption_time = POLICY_DEFAULT_PREEMPTION_TIME_US; - policy->fault_time = POLICY_DEFAULT_FAULT_TIME_US; - policy->policy_flags = 0; -} - -static void guc_policies_init(struct guc_policies *policies) -{ - struct guc_policy *policy; - u32 p, i; - - policies->dpc_promote_time = POLICY_DEFAULT_DPC_PROMOTE_TIME_US; - policies->max_num_work_items = POLICY_MAX_NUM_WI; - - for (p = 0; p < GUC_CLIENT_PRIORITY_NUM; p++) { - for (i = GUC_RENDER_ENGINE; i < GUC_MAX_ENGINES_NUM; i++) { - policy = &policies->policy[p][i]; - - guc_policy_init(policy); - } - } - - policies->is_valid = 1; -} - -/* - * The first 80 dwords of the register state context, containing the - * execlists and ppgtt registers. - */ -#define LR_HW_CONTEXT_SIZE (80 * sizeof(u32)) - -static int guc_ads_create(struct intel_guc *guc) -{ - struct drm_i915_private *dev_priv = guc_to_i915(guc); - struct i915_vma *vma; - struct page *page; - /* The ads obj includes the struct itself and buffers passed to GuC */ - struct { - struct guc_ads ads; - struct guc_policies policies; - struct guc_mmio_reg_state reg_state; - u8 reg_state_buffer[GUC_S3_SAVE_SPACE_PAGES * PAGE_SIZE]; - } __packed *blob; - struct intel_engine_cs *engine; - enum intel_engine_id id; - const u32 skipped_offset = LRC_HEADER_PAGES * PAGE_SIZE; - const u32 skipped_size = LRC_PPHWSP_SZ * PAGE_SIZE + LR_HW_CONTEXT_SIZE; - u32 base; - - GEM_BUG_ON(guc->ads_vma); - - vma = intel_guc_allocate_vma(guc, PAGE_ALIGN(sizeof(*blob))); - if (IS_ERR(vma)) - return PTR_ERR(vma); - - guc->ads_vma = vma; + if (client) + guc_client_free(client); - page = i915_vma_first_page(vma); - blob = kmap(page); - - /* GuC scheduling policies */ - guc_policies_init(&blob->policies); - - /* MMIO reg state */ - for_each_engine(engine, dev_priv, id) { - blob->reg_state.white_list[engine->guc_id].mmio_start = - engine->mmio_base + GUC_MMIO_WHITE_LIST_START; - - /* Nothing to be saved or restored for now. */ - blob->reg_state.white_list[engine->guc_id].count = 0; - } - - /* - * The GuC requires a "Golden Context" when it reinitialises - * engines after a reset. Here we use the Render ring default - * context, which must already exist and be pinned in the GGTT, - * so its address won't change after we've told the GuC where - * to find it. Note that we have to skip our header (1 page), - * because our GuC shared data is there. - */ - blob->ads.golden_context_lrca = - guc_ggtt_offset(dev_priv->kernel_context->engine[RCS].state) + - skipped_offset; - - /* - * The GuC expects us to exclude the portion of the context image that - * it skips from the size it is to read. It starts reading from after - * the execlist context (so skipping the first page [PPHWSP] and 80 - * dwords). Weird guc is weird. - */ - for_each_engine(engine, dev_priv, id) - blob->ads.eng_state_size[engine->guc_id] = - engine->context_size - skipped_size; - - base = guc_ggtt_offset(vma); - blob->ads.scheduler_policies = base + ptr_offset(blob, policies); - blob->ads.reg_state_buffer = base + ptr_offset(blob, reg_state_buffer); - blob->ads.reg_state_addr = base + ptr_offset(blob, reg_state); - - kunmap(page); - - return 0; -} - -static void guc_ads_destroy(struct intel_guc *guc) -{ - i915_vma_unpin_and_release(&guc->ads_vma); + client = fetch_and_zero(&guc->execbuf_client); + guc_client_free(client); } /* @@ -1146,15 +1084,6 @@ int intel_guc_submission_init(struct intel_guc *guc) */ GEM_BUG_ON(!guc->stage_desc_pool); - ret = intel_guc_log_create(guc); - if (ret < 0) - goto err_stage_desc_pool; - - ret = guc_ads_create(guc); - if (ret < 0) - goto err_log; - GEM_BUG_ON(!guc->ads_vma); - WARN_ON(!guc_verify_doorbells(guc)); ret = guc_clients_create(guc); if (ret) @@ -1167,11 +1096,6 @@ int intel_guc_submission_init(struct intel_guc *guc) return 0; -err_log: - intel_guc_log_destroy(guc); -err_stage_desc_pool: - guc_stage_desc_pool_destroy(guc); - return ret; } void intel_guc_submission_fini(struct intel_guc *guc) @@ -1186,8 +1110,6 @@ void intel_guc_submission_fini(struct intel_guc *guc) guc_clients_destroy(guc); WARN_ON(!guc_verify_doorbells(guc)); - guc_ads_destroy(guc); - intel_guc_log_destroy(guc); guc_stage_desc_pool_destroy(guc); } @@ -1294,7 +1216,8 @@ int intel_guc_submission_enable(struct intel_guc *guc) GEM_BUG_ON(!guc->execbuf_client); guc_reset_wq(guc->execbuf_client); - guc_reset_wq(guc->preempt_client); + if (guc->preempt_client) + guc_reset_wq(guc->preempt_client); err = intel_guc_sample_forcewake(guc); if (err) |