From 04d6067f1f19e70a418f92fa3170cf7fe53b7fdf Mon Sep 17 00:00:00 2001 From: Zhenyu Wang Date: Tue, 3 Mar 2020 13:54:12 +0800 Subject: drm/i915/gvt: Fix unnecessary schedule timer when no vGPU exits From commit f25a49ab8ab9 ("drm/i915/gvt: Use vgpu_lock to protect per vgpu access") the vgpu idr destroy is moved later than vgpu resource destroy, then it would fail to stop timer for schedule policy clean which to check vgpu idr for any left vGPU. So this trys to destroy vgpu idr earlier. Cc: Colin Xu Fixes: f25a49ab8ab9 ("drm/i915/gvt: Use vgpu_lock to protect per vgpu access") Acked-by: Colin Xu Signed-off-by: Zhenyu Wang Link: http://patchwork.freedesktop.org/patch/msgid/20200229055445.31481-1-zhenyuw@linux.intel.com --- drivers/gpu/drm/i915/gvt/vgpu.c | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/i915/gvt/vgpu.c b/drivers/gpu/drm/i915/gvt/vgpu.c index 487af6ea9972..345c2aa3b491 100644 --- a/drivers/gpu/drm/i915/gvt/vgpu.c +++ b/drivers/gpu/drm/i915/gvt/vgpu.c @@ -272,10 +272,17 @@ void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu) { struct intel_gvt *gvt = vgpu->gvt; - mutex_lock(&vgpu->vgpu_lock); - WARN(vgpu->active, "vGPU is still active!\n"); + /* + * remove idr first so later clean can judge if need to stop + * service if no active vgpu. + */ + mutex_lock(&gvt->lock); + idr_remove(&gvt->vgpu_idr, vgpu->id); + mutex_unlock(&gvt->lock); + + mutex_lock(&vgpu->vgpu_lock); intel_gvt_debugfs_remove_vgpu(vgpu); intel_vgpu_clean_sched_policy(vgpu); intel_vgpu_clean_submission(vgpu); @@ -290,7 +297,6 @@ void intel_gvt_destroy_vgpu(struct intel_vgpu *vgpu) mutex_unlock(&vgpu->vgpu_lock); mutex_lock(&gvt->lock); - idr_remove(&gvt->vgpu_idr, vgpu->id); if (idr_is_empty(&gvt->vgpu_idr)) intel_gvt_clean_irq(gvt); intel_gvt_update_vgpu_types(gvt); -- cgit v1.2.3-58-ga151 From 2fa7e15c5f466fdd0c0b196b1dc4a65d191efd96 Mon Sep 17 00:00:00 2001 From: Tina Zhang Date: Thu, 5 Mar 2020 21:15:59 +0800 Subject: drm/i915/gvt: Fix emulated vbt size issue The emulated vbt doesn't tell its size correctly. According to the intel_vbt_defs.h, vbt_header.vbt_size should the size of VBT (VBT Header, BDB Header and data blocks), and bdb_header.bdb_size should be the size of BDB (BDB Header and data blocks). This patch fixes the issue and lets vbt provided by GVT-g pass the guest i915's sanity test. v2: refine the commit message. (Zhenyu) Signed-off-by: Tina Zhang Reviewed-by: Zhenyu Wang Signed-off-by: Zhenyu Wang Link: http://patchwork.freedesktop.org/patch/msgid/20200305131600.29640-1-tina.zhang@intel.com --- drivers/gpu/drm/i915/gvt/opregion.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) diff --git a/drivers/gpu/drm/i915/gvt/opregion.c b/drivers/gpu/drm/i915/gvt/opregion.c index 867e7629025b..33569b910ed5 100644 --- a/drivers/gpu/drm/i915/gvt/opregion.c +++ b/drivers/gpu/drm/i915/gvt/opregion.c @@ -147,15 +147,14 @@ static void virt_vbt_generation(struct vbt *v) /* there's features depending on version! */ v->header.version = 155; v->header.header_size = sizeof(v->header); - v->header.vbt_size = sizeof(struct vbt) - sizeof(v->header); + v->header.vbt_size = sizeof(struct vbt); v->header.bdb_offset = offsetof(struct vbt, bdb_header); strcpy(&v->bdb_header.signature[0], "BIOS_DATA_BLOCK"); v->bdb_header.version = 186; /* child_dev_size = 33 */ v->bdb_header.header_size = sizeof(v->bdb_header); - v->bdb_header.bdb_size = sizeof(struct vbt) - sizeof(struct vbt_header) - - sizeof(struct bdb_header); + v->bdb_header.bdb_size = sizeof(struct vbt) - sizeof(struct vbt_header); /* general features */ v->general_features_header.id = BDB_GENERAL_FEATURES; -- cgit v1.2.3-58-ga151 From c67b35d970ed3391069c21f3071a26f687399ab2 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Thu, 5 Mar 2020 10:42:10 +0000 Subject: drm/i915: Actually emit the await_start Fix the inverted test to emit the wait on the end of the previous request if we /haven't/ already. Fixes: 6a79d848403d ("drm/i915: Lock signaler timeline while navigating") Signed-off-by: Chris Wilson Cc: Tvrtko Ursulin Cc: # v5.5+ Reviewed-by: Tvrtko Ursulin Link: https://patchwork.freedesktop.org/patch/msgid/20200305104210.2619967-1-chris@chris-wilson.co.uk (cherry picked from commit 07e9c59d63df6a1c44c1975c01827ba18b69270a) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/i915_request.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c index dcaa85a91090..468ca1052bcb 100644 --- a/drivers/gpu/drm/i915/i915_request.c +++ b/drivers/gpu/drm/i915/i915_request.c @@ -821,7 +821,7 @@ i915_request_await_start(struct i915_request *rq, struct i915_request *signal) return 0; err = 0; - if (intel_timeline_sync_is_later(i915_request_timeline(rq), fence)) + if (!intel_timeline_sync_is_later(i915_request_timeline(rq), fence)) err = i915_sw_fence_await_dma_fence(&rq->submit, fence, 0, I915_FENCE_GFP); -- cgit v1.2.3-58-ga151 From c951b0af2dddbb1f34be103029eb9030392d5554 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Thu, 5 Mar 2020 13:48:22 +0000 Subject: drm/i915: Return early for await_start on same timeline Requests within a timeline are ordered by that timeline, so awaiting for the start of a request within the timeline is a no-op. This used to work by falling out of the mutex_trylock() as the signaler and waiter had the same timeline and not returning an error. Fixes: 6a79d848403d ("drm/i915: Lock signaler timeline while navigating") Signed-off-by: Chris Wilson Cc: Tvrtko Ursulin Cc: # v5.5+ Reviewed-by: Tvrtko Ursulin Link: https://patchwork.freedesktop.org/patch/msgid/20200305134822.2750496-1-chris@chris-wilson.co.uk (cherry picked from commit ab7a69020fb5d5c7ba19fba60f62fd6f9ca9f779) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/i915_request.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c index 468ca1052bcb..7cd60b4a329b 100644 --- a/drivers/gpu/drm/i915/i915_request.c +++ b/drivers/gpu/drm/i915/i915_request.c @@ -776,8 +776,8 @@ i915_request_await_start(struct i915_request *rq, struct i915_request *signal) struct dma_fence *fence; int err; - GEM_BUG_ON(i915_request_timeline(rq) == - rcu_access_pointer(signal->timeline)); + if (i915_request_timeline(rq) == rcu_access_pointer(signal->timeline)) + return 0; if (i915_request_started(signal)) return 0; -- cgit v1.2.3-58-ga151 From 259170cb4c84f4165a36c0b05811eb74c495412c Mon Sep 17 00:00:00 2001 From: Tina Zhang Date: Thu, 27 Feb 2020 09:00:41 +0800 Subject: drm/i915/gvt: Fix dma-buf display blur issue on CFL Commit c3b5a8430daad ("drm/i915/gvt: Enable gfx virtualiztion for CFL") added the support on CFL. The vgpu emulation hotplug support on CFL was supposed to be included in that patch. Without the vgpu emulation hotplug support, the dma-buf based display gives us a blur face. So fix this issue by adding the vgpu emulation hotplug support on CFL. Fixes: c3b5a8430daad ("drm/i915/gvt: Enable gfx virtualiztion for CFL") Signed-off-by: Tina Zhang Acked-by: Zhenyu Wang Signed-off-by: Zhenyu Wang Link: http://patchwork.freedesktop.org/patch/msgid/20200227010041.32248-1-tina.zhang@intel.com (cherry picked from commit 135dde8853c7e00f6002e710f7e4787ed8585c0e) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/gvt/display.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/gvt/display.c b/drivers/gpu/drm/i915/gvt/display.c index e1c313da6c00..a62bdf9be682 100644 --- a/drivers/gpu/drm/i915/gvt/display.c +++ b/drivers/gpu/drm/i915/gvt/display.c @@ -457,7 +457,8 @@ void intel_vgpu_emulate_hotplug(struct intel_vgpu *vgpu, bool connected) struct drm_i915_private *dev_priv = vgpu->gvt->dev_priv; /* TODO: add more platforms support */ - if (IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv)) { + if (IS_SKYLAKE(dev_priv) || IS_KABYLAKE(dev_priv) || + IS_COFFEELAKE(dev_priv)) { if (connected) { vgpu_vreg_t(vgpu, SFUSE_STRAP) |= SFUSE_STRAP_DDID_DETECTED; -- cgit v1.2.3-58-ga151 From 1d61c5d711a2dc0b978ae905535edee9601f9449 Mon Sep 17 00:00:00 2001 From: Matthew Auld Date: Thu, 5 Mar 2020 20:35:34 +0000 Subject: drm/i915: be more solid in checking the alignment The alignment is u64, and yet is_power_of_2() assumes unsigned long, which might give different results between 32b and 64b kernel. Signed-off-by: Matthew Auld Cc: Chris Wilson Reviewed-by: Chris Wilson Signed-off-by: Chris Wilson Link: https://patchwork.freedesktop.org/patch/msgid/20200305203534.210466-1-matthew.auld@intel.com Cc: stable@vger.kernel.org (cherry picked from commit 2920516b2f719546f55079bc39a7fe409d9e80ab) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c | 3 ++- drivers/gpu/drm/i915/i915_utils.h | 5 +++++ 2 files changed, 7 insertions(+), 1 deletion(-) diff --git a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c index 60c984e10c4a..7643a30ba4cd 100644 --- a/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c +++ b/drivers/gpu/drm/i915/gem/i915_gem_execbuffer.c @@ -423,7 +423,8 @@ eb_validate_vma(struct i915_execbuffer *eb, if (unlikely(entry->flags & eb->invalid_flags)) return -EINVAL; - if (unlikely(entry->alignment && !is_power_of_2(entry->alignment))) + if (unlikely(entry->alignment && + !is_power_of_2_u64(entry->alignment))) return -EINVAL; /* diff --git a/drivers/gpu/drm/i915/i915_utils.h b/drivers/gpu/drm/i915/i915_utils.h index b0ade76bec90..d34141f7dcd8 100644 --- a/drivers/gpu/drm/i915/i915_utils.h +++ b/drivers/gpu/drm/i915/i915_utils.h @@ -234,6 +234,11 @@ static inline u64 ptr_to_u64(const void *ptr) __idx; \ }) +static inline bool is_power_of_2_u64(u64 n) +{ + return (n != 0 && ((n & (n - 1)) == 0)); +} + static inline void __list_del_many(struct list_head *head, struct list_head *first) { -- cgit v1.2.3-58-ga151 From eafc2aa20fba319b6e791a1b0c45a91511eccb6b Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Fri, 6 Mar 2020 11:30:10 +0000 Subject: drm/i915/execlists: Enable timeslice on partial virtual engine dequeue If we stop filling the ELSP due to an incompatible virtual engine request, check if we should enable the timeslice on behalf of the queue. This fixes the case where we are inspecting the last->next element when we know that the last element is the last request in the execution queue, and so decided we did not need to enable timeslicing despite the intent to do so! Fixes: 8ee36e048c98 ("drm/i915/execlists: Minimalistic timeslicing") Signed-off-by: Chris Wilson Cc: Mika Kuoppala Cc: Tvrtko Ursulin Cc: # v5.4+ Reviewed-by: Mika Kuoppala Link: https://patchwork.freedesktop.org/patch/msgid/20200306113012.3184606-1-chris@chris-wilson.co.uk (cherry picked from commit 3df2deed411e0f1b7312baf0139aab8bba4c0410) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/gt/intel_lrc.c | 29 ++++++++++++++++++----------- 1 file changed, 18 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_lrc.c b/drivers/gpu/drm/i915/gt/intel_lrc.c index fe8a59aaa629..940e7f7df69a 100644 --- a/drivers/gpu/drm/i915/gt/intel_lrc.c +++ b/drivers/gpu/drm/i915/gt/intel_lrc.c @@ -1679,11 +1679,9 @@ need_timeslice(struct intel_engine_cs *engine, const struct i915_request *rq) if (!intel_engine_has_timeslices(engine)) return false; - if (list_is_last(&rq->sched.link, &engine->active.requests)) - return false; - - hint = max(rq_prio(list_next_entry(rq, sched.link)), - engine->execlists.queue_priority_hint); + hint = engine->execlists.queue_priority_hint; + if (!list_is_last(&rq->sched.link, &engine->active.requests)) + hint = max(hint, rq_prio(list_next_entry(rq, sched.link))); return hint >= effective_prio(rq); } @@ -1725,6 +1723,18 @@ static void set_timeslice(struct intel_engine_cs *engine) set_timer_ms(&engine->execlists.timer, active_timeslice(engine)); } +static void start_timeslice(struct intel_engine_cs *engine) +{ + struct intel_engine_execlists *execlists = &engine->execlists; + + execlists->switch_priority_hint = execlists->queue_priority_hint; + + if (timer_pending(&execlists->timer)) + return; + + set_timer_ms(&execlists->timer, timeslice(engine)); +} + static void record_preemption(struct intel_engine_execlists *execlists) { (void)I915_SELFTEST_ONLY(execlists->preempt_hang.count++); @@ -1888,11 +1898,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine) * Even if ELSP[1] is occupied and not worthy * of timeslices, our queue might be. */ - if (!execlists->timer.expires && - need_timeslice(engine, last)) - set_timer_ms(&execlists->timer, - timeslice(engine)); - + start_timeslice(engine); return; } } @@ -1927,7 +1933,8 @@ static void execlists_dequeue(struct intel_engine_cs *engine) if (last && !can_merge_rq(last, rq)) { spin_unlock(&ve->base.active.lock); - return; /* leave this for another */ + start_timeslice(engine); + return; /* leave this for another sibling */ } ENGINE_TRACE(engine, -- cgit v1.2.3-58-ga151 From 8ea6bb8e4d47e07518e5dba4f5cb77e210f0df82 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Fri, 6 Mar 2020 15:46:47 +0000 Subject: drm/i915/gt: Close race between cacheline_retire and free If the cacheline may still be busy, atomically mark it for future release, and only if we can determine that it will never be used again, immediately free it. Closes: https://gitlab.freedesktop.org/drm/intel/issues/1392 Fixes: ebece7539242 ("drm/i915: Keep timeline HWSP allocated until idle across the system") Signed-off-by: Chris Wilson Cc: Tvrtko Ursulin Cc: Mika Kuoppala Cc: Matthew Auld Reviewed-by: Mika Kuoppala Cc: # v5.2+ Link: https://patchwork.freedesktop.org/patch/msgid/20200306154647.3528345-1-chris@chris-wilson.co.uk (cherry picked from commit 2d4bd971f5baa51418625f379a69f5d58b5a0450) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/gt/intel_timeline.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/drivers/gpu/drm/i915/gt/intel_timeline.c b/drivers/gpu/drm/i915/gt/intel_timeline.c index 87716529cd2f..d8d9f1179c2b 100644 --- a/drivers/gpu/drm/i915/gt/intel_timeline.c +++ b/drivers/gpu/drm/i915/gt/intel_timeline.c @@ -192,11 +192,15 @@ static void cacheline_release(struct intel_timeline_cacheline *cl) static void cacheline_free(struct intel_timeline_cacheline *cl) { + if (!i915_active_acquire_if_busy(&cl->active)) { + __idle_cacheline_free(cl); + return; + } + GEM_BUG_ON(ptr_test_bit(cl->vaddr, CACHELINE_FREE)); cl->vaddr = ptr_set_bit(cl->vaddr, CACHELINE_FREE); - if (i915_active_is_idle(&cl->active)) - __idle_cacheline_free(cl); + i915_active_release(&cl->active); } int intel_timeline_init(struct intel_timeline *timeline, -- cgit v1.2.3-58-ga151 From 14a0d527a479eb2cb6067f9e5e163e1bf35db2a9 Mon Sep 17 00:00:00 2001 From: Chris Wilson Date: Tue, 10 Mar 2020 10:17:20 +0000 Subject: drm/i915: Defer semaphore priority bumping to a workqueue Since the semaphore fence may be signaled from inside an interrupt handler from inside a request holding its request->lock, we cannot then enter into the engine->active.lock for processing the semaphore priority bump as we may traverse our call tree and end up on another held request. CPU 0: [ 2243.218864] _raw_spin_lock_irqsave+0x9a/0xb0 [ 2243.218867] i915_schedule_bump_priority+0x49/0x80 [i915] [ 2243.218869] semaphore_notify+0x6d/0x98 [i915] [ 2243.218871] __i915_sw_fence_complete+0x61/0x420 [i915] [ 2243.218874] ? kmem_cache_free+0x211/0x290 [ 2243.218876] i915_sw_fence_complete+0x58/0x80 [i915] [ 2243.218879] dma_i915_sw_fence_wake+0x3e/0x80 [i915] [ 2243.218881] signal_irq_work+0x571/0x690 [i915] [ 2243.218883] irq_work_run_list+0xd7/0x120 [ 2243.218885] irq_work_run+0x1d/0x50 [ 2243.218887] smp_irq_work_interrupt+0x21/0x30 [ 2243.218889] irq_work_interrupt+0xf/0x20 CPU 1: [ 2242.173107] _raw_spin_lock+0x8f/0xa0 [ 2242.173110] __i915_request_submit+0x64/0x4a0 [i915] [ 2242.173112] __execlists_submission_tasklet+0x8ee/0x2120 [i915] [ 2242.173114] ? i915_sched_lookup_priolist+0x1e3/0x2b0 [i915] [ 2242.173117] execlists_submit_request+0x2e8/0x2f0 [i915] [ 2242.173119] submit_notify+0x8f/0xc0 [i915] [ 2242.173121] __i915_sw_fence_complete+0x61/0x420 [i915] [ 2242.173124] ? _raw_spin_unlock_irqrestore+0x39/0x40 [ 2242.173137] i915_sw_fence_complete+0x58/0x80 [i915] [ 2242.173140] i915_sw_fence_commit+0x16/0x20 [i915] Closes: https://gitlab.freedesktop.org/drm/intel/issues/1318 Fixes: b7404c7ecb38 ("drm/i915: Bump ready tasks ahead of busywaits") Signed-off-by: Chris Wilson Cc: Tvrtko Ursulin Cc: # v5.2+ Reviewed-by: Tvrtko Ursulin Link: https://patchwork.freedesktop.org/patch/msgid/20200310101720.9944-1-chris@chris-wilson.co.uk (cherry picked from commit 209df10bb4536c81c2540df96c02cd079435357f) Signed-off-by: Jani Nikula --- drivers/gpu/drm/i915/i915_request.c | 22 +++++++++++++++++----- drivers/gpu/drm/i915/i915_request.h | 2 ++ 2 files changed, 19 insertions(+), 5 deletions(-) diff --git a/drivers/gpu/drm/i915/i915_request.c b/drivers/gpu/drm/i915/i915_request.c index 7cd60b4a329b..a18b2a244706 100644 --- a/drivers/gpu/drm/i915/i915_request.c +++ b/drivers/gpu/drm/i915/i915_request.c @@ -527,19 +527,31 @@ submit_notify(struct i915_sw_fence *fence, enum i915_sw_fence_notify state) return NOTIFY_DONE; } +static void irq_semaphore_cb(struct irq_work *wrk) +{ + struct i915_request *rq = + container_of(wrk, typeof(*rq), semaphore_work); + + i915_schedule_bump_priority(rq, I915_PRIORITY_NOSEMAPHORE); + i915_request_put(rq); +} + static int __i915_sw_fence_call semaphore_notify(struct i915_sw_fence *fence, enum i915_sw_fence_notify state) { - struct i915_request *request = - container_of(fence, typeof(*request), semaphore); + struct i915_request *rq = container_of(fence, typeof(*rq), semaphore); switch (state) { case FENCE_COMPLETE: - i915_schedule_bump_priority(request, I915_PRIORITY_NOSEMAPHORE); + if (!(READ_ONCE(rq->sched.attr.priority) & I915_PRIORITY_NOSEMAPHORE)) { + i915_request_get(rq); + init_irq_work(&rq->semaphore_work, irq_semaphore_cb); + irq_work_queue(&rq->semaphore_work); + } break; case FENCE_FREE: - i915_request_put(request); + i915_request_put(rq); break; } @@ -1318,9 +1330,9 @@ void __i915_request_queue(struct i915_request *rq, * decide whether to preempt the entire chain so that it is ready to * run at the earliest possible convenience. */ - i915_sw_fence_commit(&rq->semaphore); if (attr && rq->engine->schedule) rq->engine->schedule(rq, attr); + i915_sw_fence_commit(&rq->semaphore); i915_sw_fence_commit(&rq->submit); } diff --git a/drivers/gpu/drm/i915/i915_request.h b/drivers/gpu/drm/i915/i915_request.h index f57eadcf3583..fccc339949ec 100644 --- a/drivers/gpu/drm/i915/i915_request.h +++ b/drivers/gpu/drm/i915/i915_request.h @@ -26,6 +26,7 @@ #define I915_REQUEST_H #include +#include #include #include "gem/i915_gem_context_types.h" @@ -208,6 +209,7 @@ struct i915_request { }; struct list_head execute_cb; struct i915_sw_fence semaphore; + struct irq_work semaphore_work; /* * A list of everyone we wait upon, and everyone who waits upon us. -- cgit v1.2.3-58-ga151