drm/i915: Untangle execlist tracking
Subsystems (2)
cover drm/i915: Untangle execlist tracking
This patchset merges execlist queue items in to gem requests. It does this by using the reference count added by the "Replace seqno values with request structures" patchset to ensure that the gem request is available for the whole execlist submission lifespan.
v2: merge intel_ctx_submit_request and drm_i915_gem_request, rebase changes &
add cover letter
v3: Rebase over upstreamed "Replace seqno values with request structures" and add overzealous freeing fix.
v4: Removed re-addition of cleanup work queue (found by Daniel Vetter)
v5: Fixed non-building individual patch (0002). Separated out the tail pointer from the postfix pointer (found by Thomas Daniel)
v6: Actual removal of intel_ctx_submit_request. Update both tail and postfix pointer in __i915_add_request (found by Thomas Daniel)
v7: Removed unrelated changes
Nick Hoath (4):
drm/i915: execlist request keeps ptr/ref to gem_request drm/i915: Removed duplicate members from submit_request drm/i915: Remove FIXME_lrc_ctx backpointer drm/i915: Subsume intel_ctx_submit_request in to
drm_i915_gem_request
drivers/gpu/drm/i915/i915_debugfs.c | 4 +- drivers/gpu/drm/i915/i915_drv.h | 30 +++++++- drivers/gpu/drm/i915/i915_gem.c | 19 +++-- drivers/gpu/drm/i915/i915_gpu_error.c | 2 +- drivers/gpu/drm/i915/i915_reg.h | 32 +++++++++ drivers/gpu/drm/i915/intel_display.c | 6 +- drivers/gpu/drm/i915/intel_lrc.c | 123 +++++++++++++++++++------------- drivers/gpu/drm/i915/intel_lrc.h | 41 +++-------- drivers/gpu/drm/i915/intel_ringbuffer.c | 2 +- drivers/gpu/drm/i915/intel_ringbuffer.h | 14 ++-- 10 files changed, 167 insertions(+), 106 deletions(-)
Signature · 6 lines
2.1.1 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 1/4 drm/i915: execlist request keeps ptr/ref to gem_request
Add a reference and pointer from the execlist queue item to the associated gem request. For execlist requests that don't have a request, create one as a placeholder.
Issue: VIZ-4274
v1: Rebase after upstream of "Replace seqno values with request structures" patchset.
--- drivers/gpu/drm/i915/intel_lrc.c | 31 +++++++++++++++++++++++++------ drivers/gpu/drm/i915/intel_lrc.h | 6 +++++- 2 files changed, 30 insertions(+), 7 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +25 -6
@@ -534,7 +534,8 @@ void intel_lrc_irq_handler(struct intel_engine_cs *ring)static int execlists_context_queue (struct intel_engine_cs *ring,struct intel_context *to, -u32 tail) +u32 tail, +struct drm_i915_gem_request *request) {struct intel_ctx_submit_request *req =NULL , *cursor;struct drm_i915_private *dev_priv = ring->dev->dev_private;
@@ -553,6 +554,21 @@ static int execlists_context_queue(struct intel_engine_cs *ring, req->ring = ring; req->tail = tail; +if (!request) { +/* +* If there isn't a request associated with this submission, +* create one as a temporary holder. +*/ +WARN (1 ,"execlist context submission without request" ); + request =kzalloc (sizeof (*request),GFP_KERNEL ); +if (request ==NULL ) +return -ENOMEM ; + request->ctx = to; + request->ring = ring; + } + req->request = request; +i915_gem_request_reference (request); +intel_runtime_pm_get (dev_priv);spin_lock_irqsave (&ring->execlist_lock, flags);
@@ -766,6 +782,7 @@ void intel_execlists_retire_requests(struct intel_engine_cs *ring)intel_lr_context_unpin (ring, ctx);intel_runtime_pm_put (dev_priv);i915_gem_context_unreference (req->ctx); +i915_gem_request_unreference (req->request);list_del (&req->execlist_link);kfree (req); }
@@ -818,7 +835,8 @@ int logical_ring_flush_all_caches(struct intel_ringbuffer *ringbuf)* 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. */ -void intel_logical_ring_advance_and_submit (struct intel_ringbuffer *ringbuf) +void intel_logical_ring_advance_and_submit (struct intel_ringbuffer *ringbuf, +struct drm_i915_gem_request *request) {struct intel_engine_cs *ring = ringbuf->ring;struct intel_context *ctx = ringbuf->FIXME_lrc_ctx;
@@ -828,7 +846,7 @@ void intel_logical_ring_advance_and_submit(struct intel_ringbuffer *ringbuf)if (intel_ring_stopped (ring))return ; -execlists_context_queue (ring, ctx, ringbuf->tail); +execlists_context_queue (ring, ctx, ringbuf->tail, request); } static int intel_lr_context_pin(struct intel_engine_cs *ring,
@@ -972,7 +990,7 @@ static int logical_ring_wait_for_space(struct intel_ringbuffer *ringbuf,return ret;/* Force the context submission in case we have been skipping it */ -intel_logical_ring_advance_and_submit (ringbuf); +intel_logical_ring_advance_and_submit (ringbuf,NULL );/* With GEM the hangcheck timer should kick us out of the loop, * leaving it early runs the risk of corrupting GEM state (due
@@ -1310,7 +1328,8 @@ static void gen8_set_seqno(struct intel_engine_cs *ring, u32 seqno)intel_write_status_page (ring,I915_GEM_HWS_INDEX , seqno); } -static int gen8_emit_request (struct intel_ringbuffer *ringbuf) +static int gen8_emit_request (struct intel_ringbuffer *ringbuf, +struct drm_i915_gem_request *request) {struct intel_engine_cs *ring = ringbuf->ring;u32 cmd;
@@ -1332,7 +1351,7 @@ static int gen8_emit_request(struct intel_ringbuffer *ringbuf) i915_gem_request_get_seqno(ring->outstanding_lazy_request));intel_logical_ring_emit (ringbuf,MI_USER_INTERRUPT );intel_logical_ring_emit (ringbuf,MI_NOOP ); -intel_logical_ring_advance_and_submit (ringbuf); +intel_logical_ring_advance_and_submit (ringbuf, request);return 0 ; }
drivers/gpu/drm/i915/intel_lrc.h +5 -1
@@ -39,7 +39,9 @@ void intel_logical_ring_cleanup(struct intel_engine_cs *ring);int intel_logical_rings_init (struct drm_device *dev);int logical_ring_flush_all_caches (struct intel_ringbuffer *ringbuf); -void intel_logical_ring_advance_and_submit (struct intel_ringbuffer *ringbuf); +void intel_logical_ring_advance_and_submit ( +struct intel_ringbuffer *ringbuf, +struct drm_i915_gem_request *request);/** * intel_logical_ring_advance() - advance the ringbuffer tail * @ringbuf: Ringbuffer to advance.
@@ -110,6 +112,8 @@ struct intel_ctx_submit_request {struct list_head execlist_link;int elsp_submitted; + +struct drm_i915_gem_request *request; };void intel_lrc_irq_handler (struct intel_engine_cs *ring);
Signature · 6 lines
2.1.1 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/4 drm/i915: Removed duplicate members from submit_request
In reply to Nick HoathWhere there were duplicate variables for the tail, context and ring (engine) in the gem request and the execlist queue item, use the one from the request and remove the duplicate from the execlist queue item.
Issue: VIZ-4274
v1: Rebase
v2: Fixed build issues. Keep separate postfix & tail pointers as these are used in different ways. Reinserted missing full tail pointer update.
--- drivers/gpu/drm/i915/i915_debugfs.c | 4 ++-- drivers/gpu/drm/i915/i915_drv.h | 9 ++++++++- drivers/gpu/drm/i915/i915_gem.c | 8 ++++---- drivers/gpu/drm/i915/i915_gpu_error.c | 2 +- drivers/gpu/drm/i915/intel_lrc.c | 25 +++++++++++-------------- drivers/gpu/drm/i915/intel_lrc.h | 4 ---- drivers/gpu/drm/i915/intel_ringbuffer.c | 2 +- drivers/gpu/drm/i915/intel_ringbuffer.h | 3 ++- 8 files changed, 29 insertions(+), 28 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +2 -2
@@ -1968,11 +1968,11 @@ static int i915_execlists(struct seq_file *m, void *data)if (head_req) {struct drm_i915_gem_object *ctx_obj; - ctx_obj = head_req->ctx->engine[ring_id].state; + ctx_obj = head_req->request->ctx->engine[ring_id].state;seq_printf (m,"\tHead request id: %u\n" ,intel_execlists_ctx_id (ctx_obj));seq_printf (m,"\tHead request tail: %u\n" , - head_req->tail); + head_req->request->tail); }seq_putc (m,'\n' );
drivers/gpu/drm/i915/i915_drv.h +8 -1
@@ -2062,7 +2062,14 @@ struct drm_i915_gem_request {/** Position in the ringbuffer of the start of the request */ u32 head; -/** Position in the ringbuffer of the end of the request */ +/** +* Position in the ringbuffer of the start of the postfix. +* This is required to calculate the maximum available ringbuffer +* space without overwriting the postfix. +*/ +u32 postfix; + +/** Position in the ringbuffer of the end of the whole request */ u32 tail;/** Context related to this request */
drivers/gpu/drm/i915/i915_gem.c +4 -4
@@ -2453,7 +2453,7 @@ int __i915_add_request(struct intel_engine_cs *ring, request_ring_position =intel_ring_get_tail (ringbuf);if (i915.enable_execlists) { - ret = ring->emit_request (ringbuf); + ret = ring->emit_request (ringbuf, request);if (ret)return ret; }else {
@@ -2463,7 +2463,7 @@ int __i915_add_request(struct intel_engine_cs *ring, } request->head = request_start; - request->tail = request_ring_position; + request->postfix = request_ring_position;/* Whilst this request exists, batch_obj will be on the * active_list, and so will hold the active reference. Only when this
@@ -2657,7 +2657,7 @@ static void i915_gem_reset_ring_cleanup(struct drm_i915_private *dev_priv, execlist_link);list_del (&submit_req->execlist_link);intel_runtime_pm_put (dev_priv); -i915_gem_context_unreference (submit_req->ctx); +i915_gem_context_unreference (submit_req->request->ctx);kfree (submit_req); }
@@ -2783,7 +2783,7 @@ i915_gem_retire_requests_ring(struct intel_engine_cs *ring)* of tail of the request to update the last known position * of the GPU head. */ - ringbuf->last_retired_head = request->tail; + ringbuf->last_retired_head = request->postfix;i915_gem_free_request (request); }
drivers/gpu/drm/i915/i915_gpu_error.c +1 -1
@@ -1052,7 +1052,7 @@ static void i915_gem_record_rings(struct drm_device *dev, erq = &error->ring[i].requests[count++]; erq->seqno = request->seqno; erq->jiffies = request->emitted_jiffies; - erq->tail = request->tail; + erq->tail = request->postfix; } } }
drivers/gpu/drm/i915/intel_lrc.c +11 -14
@@ -417,7 +417,7 @@ static void execlists_context_unqueue(struct intel_engine_cs *ring) execlist_link) {if (!req0) { req0 = cursor; - }else if (req0->ctx == cursor->ctx) { + }else if (req0->request->ctx == cursor->request->ctx) {/* Same ctx: ignore first request, as second request * will update tail past first request's workload */ cursor->elsp_submitted = req0->elsp_submitted;
@@ -433,9 +433,9 @@ static void execlists_context_unqueue(struct intel_engine_cs *ring)WARN_ON (req1 && req1->elsp_submitted); -execlists_submit_contexts (ring, req0->ctx, req0->tail, - req1 ? req1->ctx :NULL , - req1 ? req1->tail :0 ); +execlists_submit_contexts (ring, req0->request->ctx, req0->request->tail, + req1 ? req1->request->ctx :NULL , + req1 ? req1->request->tail :0 ); req0->elsp_submitted++;if (req1)
@@ -455,7 +455,7 @@ static bool execlists_check_remove_request(struct intel_engine_cs *ring,if (head_req !=NULL ) {struct drm_i915_gem_object *ctx_obj = - head_req->ctx->engine[ring->id].state; + head_req->request->ctx->engine[ring->id].state;if (intel_execlists_ctx_id (ctx_obj) == request_id) {WARN (head_req->elsp_submitted ==0 ,"Never submitted head request\n" );
@@ -545,15 +545,10 @@ static int execlists_context_queue(struct intel_engine_cs *ring, req =kzalloc (sizeof (*req),GFP_KERNEL );if (req ==NULL )return -ENOMEM ; - req->ctx = to; -i915_gem_context_reference (req->ctx);if (to != ring->default_context)intel_lr_context_pin (ring, to); - req->ring = ring; - req->tail = tail; -if (!request) {/* * If there isn't a request associated with this submission,
@@ -563,11 +558,13 @@ static int execlists_context_queue(struct intel_engine_cs *ring, request =kzalloc (sizeof (*request),GFP_KERNEL );if (request ==NULL )return -ENOMEM ; - request->ctx = to; request->ring = ring; } + request->ctx = to; + request->tail = tail; req->request = request;i915_gem_request_reference (request); +i915_gem_context_reference (req->request->ctx);intel_runtime_pm_get (dev_priv);
@@ -584,7 +581,7 @@ static int execlists_context_queue(struct intel_engine_cs *ring, struct intel_ctx_submit_request, execlist_link); -if (to == tail_req->ctx) { +if (to == tail_req->request->ctx) {WARN (tail_req->elsp_submitted !=0 ,"More than 2 already-submitted reqs queued\n" );list_del (&tail_req->execlist_link);
@@ -774,14 +771,14 @@ void intel_execlists_retire_requests(struct intel_engine_cs *ring)spin_unlock_irqrestore (&ring->execlist_lock, flags);list_for_each_entry_safe (req, tmp, &retired_list, execlist_link) { -struct intel_context *ctx = req->ctx; +struct intel_context *ctx = req->request->ctx;struct drm_i915_gem_object *ctx_obj = ctx->engine[ring->id].state;if (ctx_obj && (ctx != ring->default_context))intel_lr_context_unpin (ring, ctx);intel_runtime_pm_put (dev_priv); -i915_gem_context_unreference (req->ctx); +i915_gem_context_unreference (ctx);i915_gem_request_unreference (req->request);list_del (&req->execlist_link);kfree (req);
drivers/gpu/drm/i915/intel_lrc.h +0 -4
@@ -105,10 +105,6 @@ u32 intel_execlists_ctx_id(struct drm_i915_gem_object *ctx_obj);* All accesses to the queue are mediated by a spinlock (ring->execlist_lock). */ struct intel_ctx_submit_request { -struct intel_context *ctx; -struct intel_engine_cs *ring; -u32 tail; -struct list_head execlist_link;int elsp_submitted;
drivers/gpu/drm/i915/intel_ringbuffer.c +1 -1
@@ -1931,7 +1931,7 @@ static int intel_ring_wait_request(struct intel_engine_cs *ring, int n)return 0 ;list_for_each_entry (request, &ring->request_list, list) { -if (__intel_ring_space (request->tail, ringbuf->tail, +if (__intel_ring_space (request->postfix, ringbuf->tail, ringbuf->size) >= n) {break ; }
drivers/gpu/drm/i915/intel_ringbuffer.h +2 -1
@@ -239,7 +239,8 @@ struct intel_engine_cs {struct list_head execlist_retired_req_list;u8 next_context_status_buffer;u32 irq_keep_mask;/* bitmask for interrupts that should not be masked */ -int (*emit_request)(struct intel_ringbuffer *ringbuf); +int (*emit_request)(struct intel_ringbuffer *ringbuf, +struct drm_i915_gem_request *request);int (*emit_flush)(struct intel_ringbuffer *ringbuf,u32 invalidate_domains,u32 flush_domains);
Signature · 6 lines
2.1.1 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 3/4 drm/i915: Remove FIXME_lrc_ctx backpointer
In reply to Nick HoathThe first pass implementation of execlists required a backpointer to the context to be held in the intel_ringbuffer. However the context pointer is available higher in the call stack. Remove the backpointer from the ring buffer structure and instead pass it down through the call stack.
v2: Integrate this changeset with the removal of duplicate request/execlist queue item members.
v3: Rebase
v4: Rebase. Remove passing of context when the request is passed.
--- drivers/gpu/drm/i915/i915_gem.c | 5 +-- drivers/gpu/drm/i915/intel_lrc.c | 67 ++++++++++++++++++++------------- drivers/gpu/drm/i915/intel_lrc.h | 8 +++- drivers/gpu/drm/i915/intel_ringbuffer.h | 11 ++---- 4 files changed, 52 insertions(+), 39 deletions(-)
drivers/gpu/drm/i915/i915_gem.c +2 -3
@@ -2422,8 +2422,7 @@ int __i915_add_request(struct intel_engine_cs *ring,return -ENOMEM ;if (i915.enable_execlists) { -struct intel_context *ctx = request->ctx; - ringbuf = ctx->engine[ring->id].ringbuf; + ringbuf = request->ctx->engine[ring->id].ringbuf; }else ringbuf = ring->buffer;
@@ -2436,7 +2435,7 @@ int __i915_add_request(struct intel_engine_cs *ring,* what. */ if (i915.enable_execlists) { - ret =logical_ring_flush_all_caches (ringbuf); + ret =logical_ring_flush_all_caches (ringbuf, request->ctx);if (ret)return ret; }else {
drivers/gpu/drm/i915/intel_lrc.c +40 -27
263 lines not on this page · raw message
drivers/gpu/drm/i915/intel_lrc.h +6 -2
@@ -38,9 +38,11 @@ void intel_logical_ring_stop(struct intel_engine_cs *ring);void intel_logical_ring_cleanup (struct intel_engine_cs *ring);int intel_logical_rings_init (struct drm_device *dev); -int logical_ring_flush_all_caches (struct intel_ringbuffer *ringbuf); +int logical_ring_flush_all_caches (struct intel_ringbuffer *ringbuf, +struct intel_context *ctx);void intel_logical_ring_advance_and_submit (struct intel_ringbuffer *ringbuf, +struct intel_context *ctx,struct drm_i915_gem_request *request);/** * intel_logical_ring_advance() - advance the ringbuffer tail
@@ -63,7 +65,9 @@ static inline void intel_logical_ring_emit(struct intel_ringbuffer *ringbuf,iowrite32 (data, ringbuf->virtual_start + ringbuf->tail); ringbuf->tail +=4 ; } -int intel_logical_ring_begin (struct intel_ringbuffer *ringbuf,int num_dwords); +int intel_logical_ring_begin (struct intel_ringbuffer *ringbuf, +struct intel_context *ctx, +int num_dwords);/* Logical Ring Contexts */ int intel_lr_context_render_state_init(struct intel_engine_cs *ring,
drivers/gpu/drm/i915/intel_ringbuffer.h +4 -7
@@ -99,13 +99,6 @@ struct intel_ringbuffer {struct intel_engine_cs *ring; -/* -* FIXME: This backpointer is an artifact of the history of how the -* execlist patches came into being. It will get removed once the basic -* code has landed. -*/ -struct intel_context *FIXME_lrc_ctx; -u32 head;u32 tail;int space;
@@ -123,6 +116,8 @@ struct intel_ringbuffer {u32 last_retired_head; }; +struct intel_context ; + struct intel_engine_cs { const char *name; enum intel_ring_id {
@@ -242,9 +237,11 @@ struct intel_engine_cs {int (*emit_request)(struct intel_ringbuffer *ringbuf,struct drm_i915_gem_request *request);int (*emit_flush)(struct intel_ringbuffer *ringbuf, +struct intel_context *ctx,u32 invalidate_domains,u32 flush_domains);int (*emit_bb_start)(struct intel_ringbuffer *ringbuf, +struct intel_context *ctx,u64 offset,unsigned flags);/**
Signature · 6 lines
2.1.1 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 4/4 drm/i915: Subsume intel_ctx_submit_request in to drm_i915_gem_request
In reply to Nick HoathMove all remaining elements that were unique to execlists queue items in to the associated request.
Issue: VIZ-4274
v2: Rebase. Fixed issue of overzealous freeing of request.
v3: Removed re-addition of cleanup work queue (found by Daniel Vetter)
v4: Rebase.
v5: Actual removal of intel_ctx_submit_request. Update both tail and postfix pointer in __i915_add_request (found by Thomas Daniel)
v6: Removed unrelated changes
--- drivers/gpu/drm/i915/i915_debugfs.c | 8 +++---- drivers/gpu/drm/i915/i915_drv.h | 21 +++++++++++++++++ drivers/gpu/drm/i915/i915_gem.c | 12 +++++----- drivers/gpu/drm/i915/i915_reg.h | 32 ++++++++++++++++++++++++++ drivers/gpu/drm/i915/intel_display.c | 6 ++++- drivers/gpu/drm/i915/intel_lrc.c | 44 ++++++++++++++++-------------------- drivers/gpu/drm/i915/intel_lrc.h | 27 ---------------------- 7 files changed, 87 insertions(+), 63 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +4 -4
@@ -1928,7 +1928,7 @@ static int i915_execlists(struct seq_file *m, void *data)intel_runtime_pm_get (dev_priv);for_each_ring (ring, dev_priv, ring_id) { -struct intel_ctx_submit_request *head_req =NULL ; +struct drm_i915_gem_request *head_req =NULL ;int count =0 ;unsigned long flags;
@@ -1961,18 +1961,18 @@ static int i915_execlists(struct seq_file *m, void *data)list_for_each (cursor, &ring->execlist_queue) count++; head_req =list_first_entry_or_null (&ring->execlist_queue, - struct intel_ctx_submit_request, execlist_link); + struct drm_i915_gem_request, execlist_link);spin_unlock_irqrestore (&ring->execlist_lock, flags);seq_printf (m,"\t%d requests in queue\n" , count);if (head_req) {struct drm_i915_gem_object *ctx_obj; - ctx_obj = head_req->request->ctx->engine[ring_id].state; + ctx_obj = head_req->ctx->engine[ring_id].state;seq_printf (m,"\tHead request id: %u\n" ,intel_execlists_ctx_id (ctx_obj));seq_printf (m,"\tHead request tail: %u\n" , - head_req->request->tail); + head_req->tail); }seq_putc (m,'\n' );
drivers/gpu/drm/i915/i915_drv.h +21 -0
@@ -2089,6 +2089,27 @@ struct drm_i915_gem_request {struct list_head client_list;uint32_t uniq; + +/** +* The ELSP only accepts two elements at a time, so we queue +* context/tail pairs on a given queue (ring->execlist_queue) until the +* hardware is available. The queue serves a double purpose: we also use +* it to keep track of the up to 2 contexts currently in the hardware +* (usually one in execution and the other queued up by the GPU): We +* only remove elements from the head +* of the queue when the hardware informs us that an element has been +* completed. +* +* All accesses to the queue are mediated by a spinlock +* (ring->execlist_lock). +*/ + +/** Execlist link in the submission queue.*/ +struct list_head execlist_link; + +/** Execlists no. of times this request has been sent to the ELSP */ +int elsp_submitted; + };void i915_gem_request_free (struct kref *req_ref);
drivers/gpu/drm/i915/i915_gem.c +6 -6
@@ -2414,7 +2414,7 @@ int __i915_add_request(struct intel_engine_cs *ring,struct drm_i915_private *dev_priv = ring->dev->dev_private;struct drm_i915_gem_request *request;struct intel_ringbuffer *ringbuf; -u32 request_ring_position, request_start; +u32 request_start;int ret; request = ring->outstanding_lazy_request;
@@ -2449,7 +2449,7 @@ int __i915_add_request(struct intel_engine_cs *ring,* GPU processing the request, we never over-estimate the * position of the head. */ - request_ring_position =intel_ring_get_tail (ringbuf); + request->postfix =intel_ring_get_tail (ringbuf);if (i915.enable_execlists) { ret = ring->emit_request (ringbuf, request);
@@ -2462,7 +2462,7 @@ int __i915_add_request(struct intel_engine_cs *ring, } request->head = request_start; - request->postfix = request_ring_position; + request->tail =intel_ring_get_tail (ringbuf);/* Whilst this request exists, batch_obj will be on the * active_list, and so will hold the active reference. Only when this
@@ -2649,14 +2649,14 @@ static void i915_gem_reset_ring_cleanup(struct drm_i915_private *dev_priv,* pinned in place. */ while (!list_empty (&ring->execlist_queue)) { -struct intel_ctx_submit_request *submit_req; +struct drm_i915_gem_request *submit_req; submit_req =list_first_entry (&ring->execlist_queue, - struct intel_ctx_submit_request, + struct drm_i915_gem_request, execlist_link);list_del (&submit_req->execlist_link);intel_runtime_pm_put (dev_priv); -i915_gem_context_unreference (submit_req->request->ctx); +i915_gem_context_unreference (submit_req->ctx);kfree (submit_req); }
drivers/gpu/drm/i915/intel_lrc.c +19 -25
132 lines not on this page · raw message
drivers/gpu/drm/i915/intel_lrc.h +0 -27
@@ -89,33 +89,6 @@ int intel_execlists_submission(struct drm_device *dev, struct drm_file *file, u64 exec_start, u32 flags);u32 intel_execlists_ctx_id (struct drm_i915_gem_object *ctx_obj); -/** -* struct intel_ctx_submit_request - queued context submission request -* @ctx: Context to submit to the ELSP. -* @ring: Engine to submit it to. -* @tail: how far in the context's ringbuffer this request goes to. -* @execlist_link: link in the submission queue. -* @work: workqueue for processing this request in a bottom half. -* @elsp_submitted: no. of times this request has been sent to the ELSP. -* -* The ELSP only accepts two elements at a time, so we queue context/tail -* pairs on a given queue (ring->execlist_queue) until the hardware is -* available. The queue serves a double purpose: we also use it to keep track -* of the up to 2 contexts currently in the hardware (usually one in execution -* and the other queued up by the GPU): We only remove elements from the head -* of the queue when the hardware informs us that an element has been -* completed. -* -* All accesses to the queue are mediated by a spinlock (ring->execlist_lock). -*/ -struct intel_ctx_submit_request { -struct list_head execlist_link; - -int elsp_submitted; - -struct drm_i915_gem_request *request; -}; -void intel_lrc_irq_handler (struct intel_engine_cs *ring);void intel_execlists_retire_requests (struct intel_engine_cs *ring);
Signature · 6 lines
2.1.1 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/4] drm/i915: Subsume intel_ctx_submit_request in to drm_i915_gem_request
Task id: 5587 -------------------------------------Summary-------------------------------------
Platform Delta drm-intel-nightly Series Applied
PNV 353/353 353/353 ILK 200/200 200/200 SNB 400/422 400/422 IVB 487/487 487/487 BYT 296/296 296/296
HSW -1 487/508 486/508
BDW 401/402 401/402
-------------------------------------Detailed-------------------------------------
Platform Test drm-intel-nightly Series Applied
HSW igt_kms_flip_event_leak NSPT(2, M19)PASS(1, M19) NSPT(1, M19)
Note: You need to pay more attention to line start with '*'
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH 4/4] drm/i915: Subsume intel_ctx_submit_request in to drm_i915_gem_request
In reply to Nick HoathQuoted 17 lines · Nick Hoath
void intel_execlists_retire_requests(struct intel_engine_cs *ring) {drivers/gpu/drm/i915/intel_lrc.c +2
-2-struct intel_ctx_submit_request *req, *tmp; +struct drm_i915_gem_request *req, *tmp;struct drm_i915_private *dev_priv = ring->dev->dev_private;unsigned long flags;struct list_head retired_list;@@ -776,7 +771,7 @@ void intel_execlists_retire_requests(struct intel_engine_cs *ring)spin_unlock_irqrestore (&ring->execlist_lock, flags);list_for_each_entry_safe (req, tmp, &retired_list, execlist_link) { -struct intel_context *ctx = req->request->ctx; +struct intel_context *ctx = req->ctx;struct drm_i915_gem_object *ctx_obj = ctx->engine[ring->id].state;
drivers/gpu/drm/i915/intel_lrc.c void intel_execlists_retire_requests(struct intel_engine_cs *ring) · line 783
Quoted 6 lines
drivers/gpu/drm/i915/intel_lrc.c +1
-1@@ -784,9 +779,8 @@ void intel_execlists_retire_requests(struct intel_engine_cs *ring)intel_lr_context_unpin (ring, ctx);intel_runtime_pm_put (dev_priv);i915_gem_context_unreference (ctx); -i915_gem_request_unreference (req->request); +i915_gem_request_unreference (req);list_del (&req->execlist_link);
It looks like the this req unreference can be the last one in which case list_del explodes. I don't know if it was intended that it cannot be the last unreference, but I have a log which proves it can be. :)
Regards,
Tvrtko
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 20 lines
-----Original Message-----
From: Intel-gfx [mailto:[email protected]] On Behalf Of Nick Hoath
Sent: Thursday, January 15, 2015 1:11 PM
To: [email protected]
Cc: [email protected]
Subject: [Intel-gfx] [PATCH 0/4] drm/i915: Untangle execlist trackingThis patchset merges execlist queue items in to gem requests. It does this by using the reference count added by the "Replace seqno values with request structures" patchset to ensure that the gem request is available for the whole execlist submission lifespan.
v2: merge
intel_ctx_submit_requestanddrm_i915_gem_request, rebase changes &
add cover letterv3: Rebase over upstreamed "Replace seqno values with request structures" and add overzealous freeing fix.
Quoted 6 lines
v4: Removed re-addition of cleanup work queue (found by Daniel Vetter)
v5: Fixed non-building individual patch (0002). Separated out the tail pointer from the postfix pointer (found by Thomas Daniel)
v6: Actual removal ofintel_ctx_submit_request. Update both tail and postfix pointer in__i915_add_request(found by Thomas Daniel)
v7: Removed unrelated changes
Well, I'm happy with v7 of this patchset now.
_______________________________________________
Intel-gfx mailing list
[email protected]
http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 20 lines · Thomas
Quoted 20 lines
-----Original Message-----
From: Intel-gfx [mailto:[email protected]] On Behalf Of Nick Hoath
Sent: Thursday, January 15, 2015 1:11 PM
To: [email protected]
Cc: [email protected]
Subject: [Intel-gfx] [PATCH 0/4] drm/i915: Untangle execlist trackingThis patchset merges execlist queue items in to gem requests. It does this by using the reference count added by the "Replace seqno values with request structures" patchset to ensure that the gem request is available for the whole execlist submission lifespan.
v2: merge
intel_ctx_submit_requestanddrm_i915_gem_request, rebase changes &
add cover letterv3: Rebase over upstreamed "Replace seqno values with request structures" and add overzealous freeing fix.
Quoted 8 lines
Quoted 6 lines
v4: Removed re-addition of cleanup work queue (found by Daniel Vetter)
v5: Fixed non-building individual patch (0002). Separated out the tail pointer from the postfix pointer (found by Thomas Daniel)
v6: Actual removal ofintel_ctx_submit_request. Update both tail and postfix pointer in__i915_add_request(found by Thomas Daniel)
v7: Removed unrelated changesWell, I'm happy with v7 of this patchset now.
Reviewed-by Thomas Daniel <[email protected]>
Entire series merged, thanks a lot for patches&review. I've reformatted one comment in patch 4 while applying, linebreaks got screwed up a bit somehow.
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx