Preemption with GuC, fourth try
Subsystems (2)
cover Preemption with GuC, fourth try
No major changes from previous iteration.
Dropped the workaround for missing interrupt (which turned out to be self-inflicted, now properly fixed by Chris), and applied the review comments.
Dave Gordon (1):
drm/i915/guc: Add a second client, to be used for preemption
Michał Winiarski (11):
drm/i915/guc: Do not use 0 for GuC doorbell cookie drm/i915/guc: Extract GuC stage desc pool creation into a helper drm/i915/guc: Allocate separate shared data object for GuC
communication
drm/i915/guc: Add preemption action to GuC firmware interface drm/i915/guc: Split guc_wq_item_append
drm/i915: Extract "emit write" part of emit breadcrumb functions drm/i915: Add information needed to track engine preempt state drm/i915/guc: Keep request->priority for its lifetime drm/i915: Rename helpers used for unwinding, use macro for can_preempt drm/i915/guc: Preemption! With GuC
HAX Enable GuC Submission for CI
drivers/gpu/drm/i915/i915_debugfs.c | 2 + drivers/gpu/drm/i915/i915_drv.c | 5 +- drivers/gpu/drm/i915/i915_drv.h | 2 + drivers/gpu/drm/i915/i915_gem.c | 10 + drivers/gpu/drm/i915/i915_gem_gtt.c | 8 +- drivers/gpu/drm/i915/i915_guc_submission.c | 480 +++++++++++++++++++++++------ drivers/gpu/drm/i915/i915_params.h | 4 +- drivers/gpu/drm/i915/intel_engine_cs.c | 6 +- drivers/gpu/drm/i915/intel_guc.c | 14 +- drivers/gpu/drm/i915/intel_guc.h | 11 + drivers/gpu/drm/i915/intel_guc_fwif.h | 39 +++ drivers/gpu/drm/i915/intel_lrc.c | 65 ++-- drivers/gpu/drm/i915/intel_lrc.h | 1 - drivers/gpu/drm/i915/intel_ringbuffer.h | 51 +++ 14 files changed, 534 insertions(+), 164 deletions(-)
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 1/12 drm/i915/guc: Do not use 0 for GuC doorbell cookie
Apparently, this value is reserved and may be interpreted as changing doorbell ownership. Even though we're not observing any side effects now, let's skip over it to be consistent with the spec.
v2: Apply checkpatch (Sagar)
--- drivers/gpu/drm/i915/i915_guc_submission.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)
drivers/gpu/drm/i915/i915_guc_submission.c +5 -2
@@ -475,9 +475,12 @@ static void guc_ring_doorbell(struct i915_guc_client *client)/* pointer of current doorbell cacheline */ db =__get_doorbell (client); -/* we're not expecting the doorbell cookie to change behind our back */ +/* +* We're not expecting the doorbell cookie to change behind our back, +* we also need to treat 0 as a reserved value. +*/ cookie =READ_ONCE (db->cookie); -WARN_ON_ONCE (xchg (&db->cookie, cookie +1 ) != cookie); +WARN_ON_ONCE (xchg (&db->cookie, cookie +1 ?: cookie +2 ) != cookie);/* XXX: doorbell was lost and need to acquire it again */ GEM_BUG_ON (db->db_status !=GUC_DOORBELL_ENABLED );
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 2/12 drm/i915/guc: Extract GuC stage desc pool creation into a helper
In reply to Michał WiniarskiSince it's a two-step process, we can have a cleaner error handling in the caller if we do the allocations in a helper.
--- drivers/gpu/drm/i915/i915_guc_submission.c | 65 +++++++++++++++++------------- 1 file changed, 38 insertions(+), 27 deletions(-)
drivers/gpu/drm/i915/i915_guc_submission.c +38 -27
@@ -311,6 +311,37 @@ static void guc_proc_desc_init(struct intel_guc *guc, desc->priority = client->priority; } +static int guc_stage_desc_pool_create (struct intel_guc *guc) +{ +struct i915_vma *vma; +void *vaddr; + + vma =intel_guc_allocate_vma (guc, +PAGE_ALIGN (sizeof (struct guc_stage_desc ) * +GUC_MAX_STAGE_DESCRIPTORS )); +if (IS_ERR (vma)) +return PTR_ERR (vma); + + vaddr =i915_gem_object_pin_map (vma->obj,I915_MAP_WB ); +if (IS_ERR (vaddr)) { +i915_vma_unpin_and_release (&vma); +return PTR_ERR (vaddr); + } + + guc->stage_desc_pool = vma; + guc->stage_desc_pool_vaddr = vaddr; +ida_init (&guc->stage_ids); + +return 0 ; +} + +static void guc_stage_desc_pool_destroy (struct intel_guc *guc) +{ +ida_destroy (&guc->stage_ids); +i915_gem_object_unpin_map (guc->stage_desc_pool->obj); +i915_vma_unpin_and_release (&guc->stage_desc_pool); +} +/* * Initialise/clear the stage descriptor shared with the GuC firmware. *
@@ -953,47 +984,29 @@ static void guc_ads_destroy(struct intel_guc *guc)int i915_guc_submission_init (struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -struct i915_vma *vma; -void *vaddr;int ret;if (guc->stage_desc_pool)return 0 ; - vma =intel_guc_allocate_vma (guc, -PAGE_ALIGN (sizeof (struct guc_stage_desc ) * -GUC_MAX_STAGE_DESCRIPTORS )); -if (IS_ERR (vma)) -return PTR_ERR (vma); - - guc->stage_desc_pool = vma; - - vaddr =i915_gem_object_pin_map (guc->stage_desc_pool->obj,I915_MAP_WB ); -if (IS_ERR (vaddr)) { - ret =PTR_ERR (vaddr); - goto err_vma; - } - - guc->stage_desc_pool_vaddr = vaddr; + ret =guc_stage_desc_pool_create (guc); +if (ret) +return ret; ret =intel_guc_log_create (guc);if (ret <0 ) - goto err_vaddr; + goto err_stage_desc_pool; ret =guc_ads_create (guc);if (ret <0 ) goto err_log; -ida_init (&guc->stage_ids); -return 0 ; err_log:intel_guc_log_destroy (guc); -err_vaddr: -i915_gem_object_unpin_map (guc->stage_desc_pool->obj); -err_vma: -i915_vma_unpin_and_release (&guc->stage_desc_pool); +err_stage_desc_pool: +guc_stage_desc_pool_destroy (guc);return ret; }
@@ -1001,11 +1014,9 @@ void i915_guc_submission_fini(struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -ida_destroy (&guc->stage_ids);guc_ads_destroy (guc);intel_guc_log_destroy (guc); -i915_gem_object_unpin_map (guc->stage_desc_pool->obj); -i915_vma_unpin_and_release (&guc->stage_desc_pool); +guc_stage_desc_pool_destroy (guc); }static void guc_interrupts_capture (struct drm_i915_private *dev_priv)
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 3/12 drm/i915/guc: Allocate separate shared data object for GuC communication
In reply to Michał WiniarskiWe were using first page of kernel context render state for sharing data with GuC. While it's justified by the fact that those pages are not used (note, GuC still enforces this layout and refuses to work if we remove the extra page in front), it's also confusing (why are we using this particular page?). Let's allocate a separate object instead.
v2: Drop kernel_context from GuC suspend/resume action handlers (Michel)
--- drivers/gpu/drm/i915/i915_guc_submission.c | 36 +++++++++++++++++++++++++++++- drivers/gpu/drm/i915/intel_guc.c | 14 ++---------- drivers/gpu/drm/i915/intel_guc.h | 2 ++ 3 files changed, 39 insertions(+), 13 deletions(-)
drivers/gpu/drm/i915/i915_guc_submission.c +35 -1
@@ -437,6 +437,33 @@ static void guc_stage_desc_fini(struct intel_guc *guc,memset (desc,0 ,sizeof (*desc)); } +static int guc_shared_data_create (struct intel_guc *guc) +{ +struct i915_vma *vma; +void *vaddr; + + vma =intel_guc_allocate_vma (guc,PAGE_SIZE ); +if (IS_ERR (vma)) +return PTR_ERR (vma); + + vaddr =i915_gem_object_pin_map (vma->obj,I915_MAP_WB ); +if (IS_ERR (vaddr)) { +i915_vma_unpin_and_release (&vma); +return PTR_ERR (vaddr); + } + + guc->shared_data = vma; + guc->shared_data_vaddr = vaddr; + +return 0 ; +} + +static void guc_shared_data_destroy (struct intel_guc *guc) +{ +i915_gem_object_unpin_map (guc->shared_data->obj); +i915_vma_unpin_and_release (&guc->shared_data); +} +/* Construct a Work Item and append it to the GuC's Work Queue */ static void guc_wq_item_append (struct i915_guc_client *client,struct drm_i915_gem_request *rq)
@@ -993,9 +1020,13 @@ int i915_guc_submission_init(struct drm_i915_private *dev_priv)if (ret)return ret; + ret =guc_shared_data_create (guc); +if (ret) + goto err_stage_desc_pool; + ret =intel_guc_log_create (guc);if (ret <0 ) - goto err_stage_desc_pool; + goto err_shared_data; ret =guc_ads_create (guc);if (ret <0 )
@@ -1005,6 +1036,8 @@ int i915_guc_submission_init(struct drm_i915_private *dev_priv) err_log:intel_guc_log_destroy (guc); +err_shared_data: +guc_shared_data_destroy (guc); err_stage_desc_pool:guc_stage_desc_pool_destroy (guc);return ret;
@@ -1016,6 +1049,7 @@ void i915_guc_submission_fini(struct drm_i915_private *dev_priv)guc_ads_destroy (guc);intel_guc_log_destroy (guc); +guc_shared_data_destroy (guc);guc_stage_desc_pool_destroy (guc); }
drivers/gpu/drm/i915/intel_guc.c +2 -12
@@ -268,7 +268,6 @@ int intel_guc_auth_huc(struct intel_guc *guc, u32 rsa_offset)int intel_guc_suspend (struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -struct i915_gem_context *ctx;u32 data[3 ];if (guc->fw.load_status !=INTEL_UC_FIRMWARE_SUCCESS )
@@ -276,14 +275,10 @@ int intel_guc_suspend(struct drm_i915_private *dev_priv)gen9_disable_guc_interrupts (dev_priv); - ctx = dev_priv->kernel_context; - data[0 ] =INTEL_GUC_ACTION_ENTER_S_STATE ;/* any value greater than GUC_POWER_D0 */ data[1 ] =GUC_POWER_D1 ; -/* first page is shared data with GuC */ - data[2 ] =guc_ggtt_offset (ctx->engine[RCS ].state) + -LRC_GUCSHR_PN *PAGE_SIZE ; + data[2 ] =guc_ggtt_offset (guc->shared_data);return intel_guc_send (guc, data,ARRAY_SIZE (data)); }
@@ -295,7 +290,6 @@ int intel_guc_suspend(struct drm_i915_private *dev_priv)int intel_guc_resume (struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -struct i915_gem_context *ctx;u32 data[3 ];if (guc->fw.load_status !=INTEL_UC_FIRMWARE_SUCCESS )
@@ -304,13 +298,9 @@ int intel_guc_resume(struct drm_i915_private *dev_priv)if (i915_modparams.guc_log_level >=0 )gen9_enable_guc_interrupts (dev_priv); - ctx = dev_priv->kernel_context; - data[0 ] =INTEL_GUC_ACTION_EXIT_S_STATE ; data[1 ] =GUC_POWER_D0 ; -/* first page is shared data with GuC */ - data[2 ] =guc_ggtt_offset (ctx->engine[RCS ].state) + -LRC_GUCSHR_PN *PAGE_SIZE ; + data[2 ] =guc_ggtt_offset (guc->shared_data);return intel_guc_send (guc, data,ARRAY_SIZE (data)); }
drivers/gpu/drm/i915/intel_guc.h +2 -0
@@ -54,6 +54,8 @@ struct intel_guc {struct i915_vma *stage_desc_pool;void *stage_desc_pool_vaddr;struct ida stage_ids; +struct i915_vma *shared_data; +void *shared_data_vaddr;struct i915_guc_client *execbuf_client;
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v2 03/12] drm/i915/guc: Allocate separate shared data object for GuC communication
Quoted 7 lines · Michał Winiarski
We were using first page of kernel context render state for sharing data with GuC. While it's justified by the fact that those pages are not used (note, GuC still enforces this layout and refuses to work if we remove the extra page in front), it's also confusing (why are we using this particular page?). Let's allocate a separate object instead.
v2: Drop
kernel_contextfrom GuC suspend/resume action handlers (Michel)
v2 Reviewed-by: Michel Thierry <[email protected]>
Quoted 152 lines
Suggested-by Daniele Ceraolo Spurio <[email protected]>Signed-off-by Michał Winiarski <[email protected]>Cc Chris Wilson <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Jeff McGee <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Oscar Mateo <[email protected]>Reviewed-by Daniele Ceraolo Spurio <[email protected]>--- drivers/gpu/drm/i915/i915_guc_submission.c | 36 +++++++++++++++++++++++++++++- drivers/gpu/drm/i915/intel_guc.c | 14 ++---------- drivers/gpu/drm/i915/intel_guc.h | 2 ++ 3 files changed, 39 insertions(+), 13 deletions(-)drivers/gpu/drm/i915/i915_guc_submission.c +35
-1@@ -437,6 +437,33 @@ static void guc_stage_desc_fini(struct intel_guc *guc,memset (desc,0 ,sizeof (*desc)); } +static int guc_shared_data_create (struct intel_guc *guc) +{ +struct i915_vma *vma; +void *vaddr; + + vma =intel_guc_allocate_vma (guc,PAGE_SIZE ); +if (IS_ERR (vma)) +return PTR_ERR (vma); + + vaddr =i915_gem_object_pin_map (vma->obj,I915_MAP_WB ); +if (IS_ERR (vaddr)) { +i915_vma_unpin_and_release (&vma); +return PTR_ERR (vaddr); + } + + guc->shared_data = vma; + guc->shared_data_vaddr = vaddr; + +return 0 ; +} + +static void guc_shared_data_destroy (struct intel_guc *guc) +{ +i915_gem_object_unpin_map (guc->shared_data->obj); +i915_vma_unpin_and_release (&guc->shared_data); +} +/* Construct a Work Item and append it to the GuC's Work Queue */ static void guc_wq_item_append (struct i915_guc_client *client,struct drm_i915_gem_request *rq)@@ -993,9 +1020,13 @@ int i915_guc_submission_init(struct drm_i915_private *dev_priv)if (ret)return ret; + ret =guc_shared_data_create (guc); +if (ret) + goto err_stage_desc_pool; + ret =intel_guc_log_create (guc);if (ret <0 ) - goto err_stage_desc_pool; + goto err_shared_data; ret =guc_ads_create (guc);if (ret <0 )@@ -1005,6 +1036,8 @@ int i915_guc_submission_init(struct drm_i915_private *dev_priv) err_log:intel_guc_log_destroy (guc); +err_shared_data: +guc_shared_data_destroy (guc); err_stage_desc_pool:guc_stage_desc_pool_destroy (guc);return ret;@@ -1016,6 +1049,7 @@ void i915_guc_submission_fini(struct drm_i915_private *dev_priv)guc_ads_destroy (guc);intel_guc_log_destroy (guc); +guc_shared_data_destroy (guc);guc_stage_desc_pool_destroy (guc); }drivers/gpu/drm/i915/intel_guc.c +2
-12@@ -268,7 +268,6 @@ int intel_guc_auth_huc(struct intel_guc *guc, u32 rsa_offset)int intel_guc_suspend (struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -struct i915_gem_context *ctx;u32 data[3 ];if (guc->fw.load_status !=INTEL_UC_FIRMWARE_SUCCESS )@@ -276,14 +275,10 @@ int intel_guc_suspend(struct drm_i915_private *dev_priv)gen9_disable_guc_interrupts (dev_priv); - ctx = dev_priv->kernel_context; - data[0 ] =INTEL_GUC_ACTION_ENTER_S_STATE ;/* any value greater than GUC_POWER_D0 */ data[1 ] =GUC_POWER_D1 ; -/* first page is shared data with GuC */ - data[2 ] =guc_ggtt_offset (ctx->engine[RCS ].state) + -LRC_GUCSHR_PN *PAGE_SIZE ; + data[2 ] =guc_ggtt_offset (guc->shared_data);return intel_guc_send (guc, data,ARRAY_SIZE (data)); }@@ -295,7 +290,6 @@ int intel_guc_suspend(struct drm_i915_private *dev_priv)int intel_guc_resume (struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -struct i915_gem_context *ctx;u32 data[3 ];if (guc->fw.load_status !=INTEL_UC_FIRMWARE_SUCCESS )@@ -304,13 +298,9 @@ int intel_guc_resume(struct drm_i915_private *dev_priv)if (i915_modparams.guc_log_level >=0 )gen9_enable_guc_interrupts (dev_priv); - ctx = dev_priv->kernel_context; - data[0 ] =INTEL_GUC_ACTION_EXIT_S_STATE ; data[1 ] =GUC_POWER_D0 ; -/* first page is shared data with GuC */ - data[2 ] =guc_ggtt_offset (ctx->engine[RCS ].state) + -LRC_GUCSHR_PN *PAGE_SIZE ; + data[2 ] =guc_ggtt_offset (guc->shared_data);return intel_guc_send (guc, data,ARRAY_SIZE (data)); }drivers/gpu/drm/i915/intel_guc.h +2
-1@@ -54,6 +54,8 @@ struct intel_guc {struct i915_vma *stage_desc_pool;void *stage_desc_pool_vaddr;struct ida stage_ids; +struct i915_vma *shared_data; +void *shared_data_vaddr;struct i915_guc_client *execbuf_client; --2.13.6
Quoted 5 lines
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 4/12 drm/i915/guc: Add preemption action to GuC firmware interface
In reply to Michał WiniarskiWe're using GuC action to request preemption. However, after requesting preemption we need to wait for GuC to finish its own post-processing before we start submitting our requests. Firmware is using shared context to report its status.
Let's update GuC firmware interface with those new definitions.
v2: Drop unused INTEL_GUC_PREEMPT_OPTION_IMMEDIATE
--- drivers/gpu/drm/i915/intel_guc_fwif.h | 39 +++++++++++++++++++++++++++++++++++ 1 file changed, 39 insertions(+)
drivers/gpu/drm/i915/intel_guc_fwif.h +39 -0
@@ -544,9 +544,36 @@ union guc_log_control {u32 value; } __packed; +struct guc_ctx_report { +u32 report_return_status; +u32 reserved1[64 ]; +u32 affected_count; +u32 reserved2[2 ]; +} __packed; + +/* GuC Shared Context Data Struct */ +struct guc_shared_ctx_data { +u32 addr_of_last_preempted_data_low; +u32 addr_of_last_preempted_data_high; +u32 addr_of_last_preempted_data_high_tmp; +u32 padding; +u32 is_mapped_to_proxy; +u32 proxy_ctx_id; +u32 engine_reset_ctx_id; +u32 media_reset_count; +u32 reserved1[8 ]; +u32 uk_last_ctx_switch_reason; +u32 was_reset; +u32 lrca_gpu_addr; +u64 execlist_ctx; +u32 reserved2[66 ]; +struct guc_ctx_report preempt_ctx_report[GUC_MAX_ENGINES_NUM ]; +} __packed; +/* This Action will be programmed in C180 - SOFT_SCRATCH_O_REG */ enum intel_guc_action {INTEL_GUC_ACTION_DEFAULT =0x0 , +INTEL_GUC_ACTION_REQUEST_PREEMPTION =0x2 ,INTEL_GUC_ACTION_SAMPLE_FORCEWAKE =0x6 ,INTEL_GUC_ACTION_ALLOCATE_DOORBELL =0x10 ,INTEL_GUC_ACTION_DEALLOCATE_DOORBELL =0x20 ,
@@ -562,6 +589,18 @@ enum intel_guc_action { INTEL_GUC_ACTION_LIMIT }; +enum intel_guc_preempt_options { +INTEL_GUC_PREEMPT_OPTION_DROP_WORK_Q =0x4 , +INTEL_GUC_PREEMPT_OPTION_DROP_SUBMIT_Q =0x8 , +}; + +enum intel_guc_report_status { +INTEL_GUC_REPORT_STATUS_UNKNOWN =0x0 , +INTEL_GUC_REPORT_STATUS_ACKED =0x1 , +INTEL_GUC_REPORT_STATUS_ERROR =0x2 , +INTEL_GUC_REPORT_STATUS_COMPLETE =0x4 , +}; +/* * The GuC sends its response to a command by overwriting the * command in SS0. The response is distinguishable from a command
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 5/12 drm/i915/guc: Add a second client, to be used for preemption
In reply to Michał WiniarskiFrom: Dave Gordon <[email protected]>
This second client is created with priority KMD_HIGH, and marked as preemptive. This will allow us to request preemption using GuC actions.
v2: Extract clients creation into a helper, debugfs fixups. (Michał) Recreate doorbell on init. (Daniele)
Move clients into an array.
v3: And move clients back from an array, to get rid of the enum (Michał)
--- drivers/gpu/drm/i915/i915_debugfs.c | 2 + drivers/gpu/drm/i915/i915_guc_submission.c | 112 ++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc.h | 1 + 3 files changed, 82 insertions(+), 33 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +2 -0
@@ -2484,6 +2484,8 @@ static int i915_guc_info(struct seq_file *m, void *data)seq_printf (m,"\nGuC execbuf client @ %p:\n" , guc->execbuf_client);i915_guc_client_info (m, dev_priv, guc->execbuf_client); +seq_printf (m,"\nGuC preempt client @ %p:\n" , guc->preempt_client); +i915_guc_client_info (m, dev_priv, guc->preempt_client);i915_guc_log_info (m, dev_priv);
drivers/gpu/drm/i915/i915_guc_submission.c +79 -33
165 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +1 -0
@@ -58,6 +58,7 @@ struct intel_guc {void *shared_data_vaddr;struct i915_guc_client *execbuf_client; +struct i915_guc_client *preempt_client;DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 05/12] drm/i915/guc: Add a second client, to be used for preemption
Quoted 65 lines · Michał Winiarski
From: Dave Gordon <[email protected]>
This second client is created with priority
KMD_HIGH, and marked as preemptive. This will allow us to request preemption using GuC actions.v2: Extract clients creation into a helper, debugfs fixups. (Michał) Recreate doorbell on init. (Daniele)
Move clients into an array.v3: And move clients back from an array, to get rid of the enum (Michał)
Signed-off-by Dave Gordon <[email protected]>Signed-off-by Michał Winiarski <[email protected]>Cc Chris Wilson <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Jeff McGee <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Oscar Mateo <[email protected]>--- drivers/gpu/drm/i915/i915_debugfs.c | 2 + drivers/gpu/drm/i915/i915_guc_submission.c | 112 ++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc.h | 1 + 3 files changed, 82 insertions(+), 33 deletions(-)drivers/gpu/drm/i915/i915_debugfs.c +2
-0@@ -2484,6 +2484,8 @@ static int i915_guc_info(struct seq_file *m, void *data)seq_printf (m,"\nGuC execbuf client @ %p:\n" , guc->execbuf_client);i915_guc_client_info (m, dev_priv, guc->execbuf_client); +seq_printf (m,"\nGuC preempt client @ %p:\n" , guc->preempt_client); +i915_guc_client_info (m, dev_priv, guc->preempt_client);i915_guc_log_info (m, dev_priv);drivers/gpu/drm/i915/i915_guc_submission.c +0
-0@@ -33,10 +33,11 @@* * GuC client: * A i915_guc_client refers to a submission path through GuC. Currently, there
- * is only one of these (theexecbuf_client) and this one is charged with all
- * submissions to the GuC. This struct is the owner of a doorbell, a process
- * descriptor and a workqueue (all of them inside a single gem object that
- * contains all required pages for these elements).
+ * are two clients. One of them (theexecbuf_client) is charged with all
+ * submissions to the GuC, the other one (preempt_client) is responsible for
+ * preempting theexecbuf_client. This struct is the owner of a doorbell, a
+ * process descriptor and a workqueue (all of them inside a single gem object
+ * that contains all required pages for these elements).
*
* GuC stage descriptor:
* During initialization, the driver allocates a static pool of 1024
such
drivers/gpu/drm/i915/i915_guc_submission.c static void guc_stage_desc_init(struct intel_guc · line 368
Quoted 5 lines
drivers/gpu/drm/i915/i915_guc_submission.c +1
-0@@ -363,6 +364,8 @@ static void guc_stage_desc_init(struct intel_guc *guc,memset (desc,0 ,sizeof (*desc)); desc->attribute =GUC_STAGE_DESC_ATTR_ACTIVE |GUC_STAGE_DESC_ATTR_KERNEL ; +if (client->priority <=GUC_CLIENT_PRIORITY_HIGH )
Can we use
if (client->priority == GUC_CLIENT_PRIORITY_KMD_HIGH || client->priority == GUC_CLIENT_PRIORITY_HIGH)
I assume compiler will optimize it to the same asm, but then we can avoid condition that may be read wrong (priority less then high)
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +1
-0+ desc->attribute |=GUC_STAGE_DESC_ATTR_PREEMPT ; desc->stage_id = client->stage_id; desc->priority = client->priority;drivers/gpu/drm/i915/i915_guc_submission.c +0
-2desc->db_id = client->doorbell_id;@@ -763,14 +766,18 @@ static int guc_init_doorbell_hw(struct intel_guc *guc)/* Now for every client (and not only execbuf_client) make sure their * doorbells are known by the GuC */ -//for (client = client_list; client != NULL; client = client->next) - {
drivers/gpu/drm/i915/i915_guc_submission.c static int guc_init_doorbell_hw(struct intel_guc · line 772
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +4
-6- ret =__create_doorbell (client); -if (ret) { -DRM_ERROR ("Couldn't recreate client %u doorbell: %d\n" , - client->stage_id, ret); -return ret; - } + ret =__create_doorbell (guc->execbuf_client); +if (ret) { +DRM_ERROR ("Couldn't recreate client %u doorbell: %d\n" , + guc->execbuf_client->stage_id, ret);
This
drivers/gpu/drm/i915/i915_guc_submission.c static int guc_init_doorbell_hw(struct intel_guc *guc) · line 779
Quoted 7 lines
drivers/gpu/drm/i915/i915_guc_submission.c +7
-0+return ret; + } + + ret =__create_doorbell (guc->preempt_client); +if (ret) { +DRM_ERROR ("Couldn't recreate client %u doorbell: %d\n" , + guc->preempt_client->stage_id, ret);
And this error can be moved to __create_doorbell to avoid duplication
Quoted 73 lines
drivers/gpu/drm/i915/i915_guc_submission.c +42
-1+return ret; }/* Read back & verify all (used & unused) doorbell registers */ @@ -895,6 +902,47 @@ static void guc_client_free(struct i915_guc_client *client)kfree (client); } +static int guc_clients_create (struct intel_guc *guc) +{ +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct i915_guc_client *client; + + client =guc_client_alloc (dev_priv, +INTEL_INFO (dev_priv)->ring_mask, +GUC_CLIENT_PRIORITY_KMD_NORMAL , + dev_priv->kernel_context); +if (IS_ERR (client)) { +DRM_ERROR ("Failed to create GuC client for submission!\n" ); +return PTR_ERR (client); + } + 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); + } + guc->preempt_client = client; + +return 0 ; +} + +static void guc_clients_destroy (struct intel_guc *guc) +{ +struct i915_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 ;@@ -1134,7 +1182,6 @@ static void i915_guc_submission_unpark(struct intel_engine_cs *engine)int i915_guc_submission_enable (struct drm_i915_private *dev_priv) {struct intel_guc *guc = &dev_priv->guc; -struct i915_guc_client *client = guc->execbuf_client;struct intel_engine_cs *engine;enum intel_engine_id id;drivers/gpu/drm/i915/i915_guc_submission.c +0
-9int err;@@ -1152,28 +1199,29 @@ int i915_guc_submission_enable(struct drm_i915_private *dev_priv)sizeof (struct guc_wq_item ) * I915_NUM_ENGINES > GUC_WQ_SIZE); -if (!client) { - client =guc_client_alloc (dev_priv, -INTEL_INFO (dev_priv)->ring_mask, -GUC_CLIENT_PRIORITY_KMD_NORMAL , - dev_priv->kernel_context); -if (IS_ERR (client)) { -DRM_ERROR ("Failed to create GuC client for execbuf!\n" ); -return PTR_ERR (client); - }
drivers/gpu/drm/i915/i915_guc_submission.c int i915_guc_submission_enable(struct · line 1209
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +8
-2- - guc->execbuf_client = client; +/* +* We're being called on both module initialization and on reset, +* until this flow is changed, we're using regular client presence to +* determine which case are we in, and whether we should allocate new +* clients or just reset their workqueues. +*/ +if (!guc->execbuf_client) { +GEM_BUG_ON (guc->preempt_client);
You can move this GEM_BUG_ON to guc_clients_create and check both
Quoted 42 lines
drivers/gpu/drm/i915/i915_guc_submission.c +11
-9+ err =guc_clients_create (guc); +if (err) +return err; + }else { +guc_reset_wq (guc->execbuf_client); +guc_reset_wq (guc->preempt_client); } err =intel_guc_sample_forcewake (guc);if (err) - goto err_execbuf_client; - -guc_reset_wq (client); + goto err_free_clients; err =guc_init_doorbell_hw (guc);if (err) - goto err_execbuf_client; + goto err_free_clients;/* Take over from manual control of ELSP (execlists) */ guc_interrupts_capture (dev_priv);@@ -1187,9 +1235,8 @@ int i915_guc_submission_enable(struct drm_i915_private *dev_priv)return 0 ; -err_execbuf_client: -guc_client_free (guc->execbuf_client); - guc->execbuf_client =NULL ; +err_free_clients: +guc_clients_destroy (guc);return err; }@@ -1204,6 +1251,5 @@ void i915_guc_submission_disable(struct drm_i915_private *dev_priv)/* Revert back to manual ELSP submission */ intel_engines_reset_default_submission (dev_priv ); -guc_client_free (guc->execbuf_client); - guc->execbuf_client =NULL ; +guc_clients_destroy (guc); }drivers/gpu/drm/i915/intel_guc.h +0
-0
drivers/gpu/drm/i915/intel_guc.h struct intel_guc { · line 62
Quoted 5 lines
drivers/gpu/drm/i915/intel_guc.h +1
-0@@ -58,6 +58,7 @@ struct intel_guc {void *shared_data_vaddr;struct i915_guc_client *execbuf_client; +struct i915_guc_client *preempt_client;DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 1/1 drm/i915/guc: Add a second client, to be used for preemption
In reply to Michał WiniarskiFrom: Dave Gordon <[email protected]>
This second client is created with priority KMD_HIGH, and marked as preemptive. This will allow us to request preemption using GuC actions.
v2: Extract clients creation into a helper, debugfs fixups. (Michał) Recreate doorbell on init. (Daniele)
Move clients into an array.
v3: And move clients back from an array, to get rid of the enum (Michał)
v4: Use is_high_priority, move DRM_ERROR into __create_doorbell, move GEM_BUG_ON inside guc_clients_create (Michał)
--- drivers/gpu/drm/i915/i915_debugfs.c | 2 + drivers/gpu/drm/i915/i915_guc_submission.c | 117 ++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc.h | 1 + 3 files changed, 84 insertions(+), 36 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +2 -0
@@ -2484,6 +2484,8 @@ static int i915_guc_info(struct seq_file *m, void *data)seq_printf (m,"\nGuC execbuf client @ %p:\n" , guc->execbuf_client);i915_guc_client_info (m, dev_priv, guc->execbuf_client); +seq_printf (m,"\nGuC preempt client @ %p:\n" , guc->preempt_client); +i915_guc_client_info (m, dev_priv, guc->preempt_client);i915_guc_log_info (m, dev_priv);
drivers/gpu/drm/i915/i915_guc_submission.c +81 -36
183 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +1 -0
@@ -58,6 +58,7 @@ struct intel_guc {void *shared_data_vaddr;struct i915_guc_client *execbuf_client; +struct i915_guc_client *preempt_client;DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v5 1/1 drm/i915/guc: Add a second client, to be used for preemption
From: Dave Gordon <[email protected]>
This second client is created with priority KMD_HIGH, and marked as preemptive. This will allow us to request preemption using GuC actions.
v2: Extract clients creation into a helper, debugfs fixups. (Michał) Recreate doorbell on init. (Daniele)
Move clients into an array.
v3: And move clients back from an array, to get rid of the enum (Michał)
v4: Use is_high_priority, move DRM_ERROR into __create_doorbell, move GEM_BUG_ON inside guc_clients_create (Michał)
v5: Split the BUG_ON (Michał)
--- drivers/gpu/drm/i915/i915_debugfs.c | 2 + drivers/gpu/drm/i915/i915_guc_submission.c | 118 ++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc.h | 1 + 3 files changed, 85 insertions(+), 36 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +2 -0
@@ -2484,6 +2484,8 @@ static int i915_guc_info(struct seq_file *m, void *data)seq_printf (m,"\nGuC execbuf client @ %p:\n" , guc->execbuf_client);i915_guc_client_info (m, dev_priv, guc->execbuf_client); +seq_printf (m,"\nGuC preempt client @ %p:\n" , guc->preempt_client); +i915_guc_client_info (m, dev_priv, guc->preempt_client);i915_guc_log_info (m, dev_priv);
drivers/gpu/drm/i915/i915_guc_submission.c +82 -36
184 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +1 -0
@@ -58,6 +58,7 @@ struct intel_guc {void *shared_data_vaddr;struct i915_guc_client *execbuf_client; +struct i915_guc_client *preempt_client;DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v5] drm/i915/guc: Add a second client, to be used for preemption
Quoted 16 lines · Michał Winiarski
From: Dave Gordon <[email protected]>
This second client is created with priority
KMD_HIGH, and marked as preemptive. This will allow us to request preemption using GuC actions.v2: Extract clients creation into a helper, debugfs fixups. (Michał) Recreate doorbell on init. (Daniele)
Move clients into an array.v3: And move clients back from an array, to get rid of the enum (Michał)
v4: Use
is_high_priority, moveDRM_ERRORinto__create_doorbell, moveGEM_BUG_ONinsideguc_clients_create(Michał)v5: Split the
BUG_ON(Michał)
Quoted 8 lines
Signed-off-by Dave Gordon <[email protected]>Signed-off-by Michał Winiarski <[email protected]>Cc Chris Wilson <[email protected]>Cc Daniele Ceraolo Spurio <[email protected]>Cc Jeff McGee <[email protected]>Cc Michal Wajdeczko <[email protected]>Cc Oscar Mateo <[email protected]>---
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 1/1 drm/i915/guc: Add a second client, to be used for preemption
In reply to Michał WiniarskiFrom: Dave Gordon <[email protected]>
This second client is created with priority KMD_HIGH, and marked as preemptive. This will allow us to request preemption using GuC actions.
v2: Extract clients creation into a helper, debugfs fixups. (Michał) Recreate doorbell on init. (Daniele)
Move clients into an array.
v3: And move clients back from an array, to get rid of the enum (Michał)
v4: Use is_high_priority, move DRM_ERROR into __create_doorbell, move GEM_BUG_ON inside guc_clients_create (Michał)
v5: Split the BUG_ON (Michał)
v6: Cleanup after error during doorbell reinit (Michał)
--- drivers/gpu/drm/i915/i915_debugfs.c | 2 + drivers/gpu/drm/i915/i915_guc_submission.c | 118 ++++++++++++++++++++--------- drivers/gpu/drm/i915/intel_guc.h | 1 + 3 files changed, 86 insertions(+), 35 deletions(-)
drivers/gpu/drm/i915/i915_debugfs.c +2 -0
@@ -2484,6 +2484,8 @@ static int i915_guc_info(struct seq_file *m, void *data)seq_printf (m,"\nGuC execbuf client @ %p:\n" , guc->execbuf_client);i915_guc_client_info (m, dev_priv, guc->execbuf_client); +seq_printf (m,"\nGuC preempt client @ %p:\n" , guc->preempt_client); +i915_guc_client_info (m, dev_priv, guc->preempt_client);i915_guc_log_info (m, dev_priv);
drivers/gpu/drm/i915/i915_guc_submission.c +83 -35
184 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +1 -0
@@ -58,6 +58,7 @@ struct intel_guc {void *shared_data_vaddr;struct i915_guc_client *execbuf_client; +struct i915_guc_client *preempt_client;DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v6] drm/i915/guc: Add a second client, to be used for preemption
Quoted 11 lines · Michał Winiarski
drivers/gpu/drm/i915/i915_guc_submission.c +0
-7@@ -763,14 +770,14 @@ static int guc_init_doorbell_hw(struct intel_guc *guc)/* Now for every client (and not only execbuf_client) make sure their * doorbells are known by the GuC */ -//for (client = client_list; client != NULL; client = client->next) - { - ret =__create_doorbell (client); -if (ret) { -DRM_ERROR ("Couldn't recreate client %u doorbell: %d\n" , - client->stage_id, ret); -return ret;
drivers/gpu/drm/i915/i915_guc_submission.c static int guc_init_doorbell_hw(struct intel_guc *guc) · line 781
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +8
-1- } + ret =__create_doorbell (guc->execbuf_client); +if (ret) +return ret; + + ret =__create_doorbell (guc->preempt_client); +if (ret) { +__destroy_doorbell (guc->execbuf_client); +return ret; }
I'm pretty sure there's an old client left behind after this, e.g.:
static int guc_init_doorbell_hw(struct intel_guc *guc) {
- struct i915_guc_client *client = guc->execbuf_client;
...
-Michel
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v6] drm/i915/guc: Add a second client, to be used for preemption
Quoting Michel Thierry (2017-10-26 19:49:06)
Quoted 25 lines · Michał Winiarski
Quoted 11 lines
drivers/gpu/drm/i915/i915_guc_submission.c +0
-7@@ -763,14 +770,14 @@ static int guc_init_doorbell_hw(struct intel_guc *guc)/* Now for every client (and not only execbuf_client) make sure their * doorbells are known by the GuC */ -//for (client = client_list; client != NULL; client = client->next) - { - ret =__create_doorbell (client); -if (ret) { -DRM_ERROR ("Couldn't recreate client %u doorbell: %d\n" , - client->stage_id, ret); -return ret;drivers/gpu/drm/i915/i915_guc_submission.c static int guc_init_doorbell_hw(struct intel_guc *guc) · line 781
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +8
-1- } + ret =__create_doorbell (guc->execbuf_client); +if (ret) +return ret; + + ret =__create_doorbell (guc->preempt_client); +if (ret) { +__destroy_doorbell (guc->execbuf_client); +return ret; }I'm pretty sure there's an old client left behind after this, e.g.:
Quoted 3 lines
static int
guc_init_doorbell_hw(struct intel_guc *guc){
- structi915_guc_client*client =guc->execbuf_client;
That doesn't look like it's a problem, or that confusing. This client local var is just a placeholder used to clear the unwanted doorbells. We then walk over the known kernel clients to reset them.
So, I think the suggestion is to split this function again, but I don't see a blocker for merging.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v6] drm/i915/guc: Add a second client, to be used for preemption
Quoted 30 lines · Chris Wilson
Quoting Michel Thierry (2017-10-26 19:49:06)
Quoted 25 lines · Michał Winiarski
Quoted 11 lines
drivers/gpu/drm/i915/i915_guc_submission.c +0
-7@@ -763,14 +770,14 @@ static int guc_init_doorbell_hw(struct intel_guc *guc)/* Now for every client (and not only execbuf_client) make sure their * doorbells are known by the GuC */ -//for (client = client_list; client != NULL; client = client->next) - { - ret =__create_doorbell (client); -if (ret) { -DRM_ERROR ("Couldn't recreate client %u doorbell: %d\n" , - client->stage_id, ret); -return ret;drivers/gpu/drm/i915/i915_guc_submission.c static int guc_init_doorbell_hw(struct intel_guc *guc) · line 781
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +8
-1- } + ret =__create_doorbell (guc->execbuf_client); +if (ret) +return ret; + + ret =__create_doorbell (guc->preempt_client); +if (ret) { +__destroy_doorbell (guc->execbuf_client); +return ret; }I'm pretty sure there's an old client left behind after this, e.g.:
Quoted 3 lines
static int
guc_init_doorbell_hw(struct intel_guc *guc){
- structi915_guc_client*client =guc->execbuf_client;
Quoted 6 lines
That doesn't look like it's a problem, or that confusing. This client local var is just a placeholder used to clear the unwanted doorbells. We then walk over the known kernel clients to reset them.
So, I think the suggestion is to split this function again, but I don't see a blocker for merging.
Ok, I'll need to split the function (or pass some flags to execute just parts of it) for the selftest anyway.
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 6/12 drm/i915/guc: Split guc_wq_item_append
In reply to Michał WiniarskiWe're using a special preempt context for HW to preempt into. We don't want to emit any requests there, but we still need to wrap this context into a valid GuC work item.
Let's cleanup the functions operating on GuC work items.
We can extract guc_request_add - responsible for adding GuC work item and ringing the doorbell, and guc_wq_item_append - used by the function above, not tied to the concept of gem request.
--- drivers/gpu/drm/i915/i915_guc_submission.c | 55 ++++++++++++++++-------------- 1 file changed, 30 insertions(+), 25 deletions(-)
drivers/gpu/drm/i915/i915_guc_submission.c +30 -25
@@ -71,7 +71,7 @@* WQ_TYPE_INORDER is needed to support legacy submission via GuC, which * represents in-order queue. The kernel driver packs ring tail pointer and an * ELSP context descriptor dword into Work Item. -* See guc_wq_item_append() +* See guc_add_request() * * ADS: * The Additional Data Struct (ADS) has pointers for different buffers used by
@@ -390,7 +390,7 @@ static void guc_stage_desc_init(struct intel_guc *guc,* submission or, in other words, not using a direct submission * model) the KMD's LRCA is not used for any work submission. * Instead, the GuC uses the LRCA of the user mode context (see -* guc_wq_item_append below). +* guc_add_request below). */ lrc->context_desc =lower_32_bits (ce->lrc_desc);
@@ -469,22 +469,18 @@ static void guc_shared_data_destroy(struct intel_guc *guc)/* Construct a Work Item and append it to the GuC's Work Queue */ static void guc_wq_item_append (struct i915_guc_client *client, -struct drm_i915_gem_request *rq) +u32 target_engine,u32 context_desc, +u32 ring_tail,u32 fence_id) {/* wqi_len is in DWords, and does not include the one-word header */ const size_t wqi_size =sizeof (struct guc_wq_item );const u32 wqi_len = wqi_size /sizeof (u32) -1 ; -struct intel_engine_cs *engine = rq->engine; -struct i915_gem_context *ctx = rq->ctx;struct guc_process_desc *desc =__get_process_desc (client);struct guc_wq_item *wqi; -u32 ring_tail, wq_off; +u32 wq_off;lockdep_assert_held (&client->wq_lock); - ring_tail =intel_ring_set_tail (rq->ring, rq->tail) /sizeof (u64); -GEM_BUG_ON (ring_tail >WQ_RING_TAIL_MAX ); -/* For now workqueue item is 4 DWs; workqueue buffer is 2 pages. So we * should not have the case where structure wqi is across page, neither * wrapped to the beginning. This simplifies the implementation below.
@@ -506,15 +502,14 @@ static void guc_wq_item_append(struct i915_guc_client *client,/* Now fill in the 4-word work queue item */ wqi->header =WQ_TYPE_INORDER | (wqi_len <<WQ_LEN_SHIFT ) | - (engine->guc_id <<WQ_TARGET_SHIFT ) | + (target_engine <<WQ_TARGET_SHIFT ) |WQ_NO_WCFLUSH_WAIT ; - - wqi->context_desc =lower_32_bits (intel_lr_context_descriptor (ctx, engine)); - + wqi->context_desc = context_desc; wqi->submit_element_info = ring_tail <<WQ_RING_TAIL_SHIFT ; - wqi->fence_id = rq->global_seqno; +GEM_BUG_ON (ring_tail >WQ_RING_TAIL_MAX ); + wqi->fence_id = fence_id; -/* Postincrement WQ tail for next time. */ +/* Make the update visible to GuC */ WRITE_ONCE (desc->tail, (wq_off + wqi_size) & (GUC_WQ_SIZE -1 )); }
@@ -547,6 +542,25 @@ static void guc_ring_doorbell(struct i915_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) +{ +struct i915_guc_client *client = guc->execbuf_client; +struct intel_engine_cs *engine = rq->engine; +u32 ctx_desc =lower_32_bits (intel_lr_context_descriptor (rq->ctx, engine)); +u32 ring_tail =intel_ring_set_tail (rq->ring, rq->tail) /sizeof (u64); + +spin_lock (&client->wq_lock); + +guc_wq_item_append (client, engine->guc_id, ctx_desc, + ring_tail, rq->global_seqno); +guc_ring_doorbell (client); + + client->submissions[engine->id] +=1 ; + +spin_unlock (&client->wq_lock); +} +/** * i915_guc_submit() - Submit commands through GuC * @engine: engine associated with the commands
@@ -558,10 +572,8 @@ static void i915_guc_submit(struct intel_engine_cs *engine) {struct drm_i915_private *dev_priv = engine->i915;struct intel_guc *guc = &dev_priv->guc; -struct i915_guc_client *client = guc->execbuf_client;struct intel_engine_execlists *const execlists = &engine->execlists;struct execlist_port *port = execlists->port; -const unsigned int engine_id = engine->id;unsigned int n;for (n =0 ; n <execlists_num_ports (execlists); n++) {
@@ -575,14 +587,7 @@ static void i915_guc_submit(struct intel_engine_cs *engine)if (i915_vma_is_map_and_fenceable (rq->ring->vma))POSTING_READ_FW (GUC_STATUS ); -spin_lock (&client->wq_lock); - -guc_wq_item_append (client, rq); -guc_ring_doorbell (client); - - client->submissions[engine_id] +=1 ; - -spin_unlock (&client->wq_lock); +guc_add_request (guc, rq); } } }
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 7/12 drm/i915: Extract "emit write" part of emit breadcrumb functions
In reply to Michał WiniarskiLet's separate the "emit" part from touching any internal structures, this way we can have a generic "emit coherent GGTT write" function. We would like to reuse this functionality for emitting HWSP write, to confirm that preempt-to-idle has finished.
v2: Reorder args to match emit_pipe_control, s/render/rcs (Chris)
--- drivers/gpu/drm/i915/intel_lrc.c | 28 +++++++----------------- drivers/gpu/drm/i915/intel_ringbuffer.h | 38 +++++++++++++++++++++++++++++++++ 2 files changed, 46 insertions(+), 20 deletions(-)
drivers/gpu/drm/i915/intel_lrc.c +8 -20
@@ -1794,10 +1794,8 @@ static void gen8_emit_breadcrumb(struct drm_i915_gem_request *request, u32 *cs)/* w/a: bit 5 needs to be zero for MI_FLUSH_DW address. */ BUILD_BUG_ON (I915_GEM_HWS_INDEX_ADDR & (1 <<5 )); - *cs++ = (MI_FLUSH_DW +1 ) |MI_FLUSH_DW_OP_STOREDW ; - *cs++ =intel_hws_seqno_address (request->engine) |MI_FLUSH_DW_USE_GTT ; - *cs++ =0 ; - *cs++ = request->global_seqno; + cs =gen8_emit_ggtt_write (cs, request->global_seqno, +intel_hws_seqno_address (request->engine)); *cs++ =MI_USER_INTERRUPT ; *cs++ =MI_NOOP ; request->tail =intel_ring_offset (request, cs);
@@ -1807,24 +1805,14 @@ static void gen8_emit_breadcrumb(struct drm_i915_gem_request *request, u32 *cs) }static const int gen8_emit_breadcrumb_sz =6 +WA_TAIL_DWORDS ; -static void gen8_emit_breadcrumb_render (struct drm_i915_gem_request *request, +static void gen8_emit_breadcrumb_rcs (struct drm_i915_gem_request *request,u32 *cs) {/* We're using qword write, seqno should be aligned to 8 bytes. */ BUILD_BUG_ON (I915_GEM_HWS_INDEX &1 ); -/* w/a for post sync ops following a GPGPU operation we -* need a prior CS_STALL, which is emitted by the flush -* following the batch. -*/ - *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 (request->engine); - *cs++ =0 ; - *cs++ = request->global_seqno; -/* We're thrashing one dword of HWS. */ - *cs++ =0 ; + cs =gen8_emit_ggtt_write_rcs (cs, request->global_seqno, +intel_hws_seqno_address (request->engine)); *cs++ =MI_USER_INTERRUPT ; *cs++ =MI_NOOP ; request->tail =intel_ring_offset (request, cs);
@@ -1832,7 +1820,7 @@ static void gen8_emit_breadcrumb_render(struct drm_i915_gem_request *request,gen8_emit_wa_tail (request, cs); } -static const int gen8_emit_breadcrumb_render_sz =8 +WA_TAIL_DWORDS ; +static const int gen8_emit_breadcrumb_rcs_sz =8 +WA_TAIL_DWORDS ;static int gen8_init_rcs_context (struct drm_i915_gem_request *req) {
@@ -1991,8 +1979,8 @@ int logical_render_ring_init(struct intel_engine_cs *engine) engine->init_hw = gen8_init_render_ring; engine->init_context = gen8_init_rcs_context; engine->emit_flush = gen8_emit_flush_render; - engine->emit_breadcrumb = gen8_emit_breadcrumb_render; - engine->emit_breadcrumb_sz = gen8_emit_breadcrumb_render_sz; + engine->emit_breadcrumb = gen8_emit_breadcrumb_rcs; + engine->emit_breadcrumb_sz = gen8_emit_breadcrumb_rcs_sz; ret =intel_engine_create_scratch (engine,PAGE_SIZE );if (ret)
drivers/gpu/drm/i915/intel_ringbuffer.h +38 -0
@@ -869,6 +869,44 @@ static inline u32 *gen8_emit_pipe_control(u32 *batch, u32 flags, u32 offset)return batch +6 ; } +static inline u32 * +gen8_emit_ggtt_write_rcs (u32 *cs,u32 value,u32 gtt_offset) +{ +/* We're using qword write, offset should be aligned to 8 bytes. */ +GEM_BUG_ON (!IS_ALIGNED (gtt_offset,8 )); + +/* w/a for post sync ops following a GPGPU operation we +* need a prior CS_STALL, which is emitted by the flush +* following the batch. +*/ + *cs++ =GFX_OP_PIPE_CONTROL (6 ); + *cs++ =PIPE_CONTROL_GLOBAL_GTT_IVB |PIPE_CONTROL_CS_STALL | +PIPE_CONTROL_QW_WRITE ; + *cs++ = gtt_offset; + *cs++ =0 ; + *cs++ = value; +/* We're thrashing one dword of HWS. */ + *cs++ =0 ; + +return cs; +} + +static inline u32 * +gen8_emit_ggtt_write (u32 *cs,u32 value,u32 gtt_offset) +{ +/* w/a: bit 5 needs to be zero for MI_FLUSH_DW address. */ +GEM_BUG_ON (gtt_offset & (1 <<5 )); +/* Offset should be aligned to 8 bytes for both (QW/DW) write types */ +GEM_BUG_ON (!IS_ALIGNED (gtt_offset,8 )); + + *cs++ = (MI_FLUSH_DW +1 ) |MI_FLUSH_DW_OP_STOREDW ; + *cs++ = gtt_offset |MI_FLUSH_DW_USE_GTT ; + *cs++ =0 ; + *cs++ = value; + +return cs; +} +bool intel_engine_is_idle (struct intel_engine_cs *engine);bool intel_engines_are_idle (struct drm_i915_private *dev_priv);
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 8/12 drm/i915: Add information needed to track engine preempt state
In reply to Michał WiniarskiWe shouldn't inspect ELSP context status (or any other bits depending on specific submission backend) when using GuC submission. Let's use another piece of HWSP for preempt context, to write its bit of information, meaning that preemption has finished, and hardware is now idle.
--- drivers/gpu/drm/i915/intel_ringbuffer.h | 7 +++++++ 1 file changed, 7 insertions(+)
drivers/gpu/drm/i915/intel_ringbuffer.h +7 -0
@@ -626,6 +626,8 @@ intel_write_status_page(struct intel_engine_cs *engine, int reg, u32 value)*/ #define I915_GEM_HWS_INDEX 0x30#define I915_GEM_HWS_INDEX_ADDR (I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT) +#define I915_GEM_HWS_PREEMPT_INDEX 0x32 +#define I915_GEM_HWS_PREEMPT_ADDR (I915_GEM_HWS_PREEMPT_INDEX << MI_STORE_DWORD_INDEX_SHIFT)#define I915_GEM_HWS_SCRATCH_INDEX 0x40#define I915_GEM_HWS_SCRATCH_ADDR (I915_GEM_HWS_SCRATCH_INDEX << MI_STORE_DWORD_INDEX_SHIFT)
@@ -778,6 +780,11 @@ static inline u32 intel_hws_seqno_address(struct intel_engine_cs *engine)return engine->status_page.ggtt_offset +I915_GEM_HWS_INDEX_ADDR ; } +static inline u32 intel_hws_preempt_done_address (struct intel_engine_cs *engine) +{ +return engine->status_page.ggtt_offset +I915_GEM_HWS_PREEMPT_ADDR ; +} +/* intel_breadcrumbs.c -- user interrupt bottom-half for waiters */ int intel_engine_init_breadcrumbs (struct intel_engine_cs *engine);
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v2 9/12 drm/i915/guc: Keep request->priority for its lifetime
In reply to Michał WiniarskiWe also want to support preemption with GuC submission backend. In order to do that, we need to remember the priority, like we do on execlists path.
v2: Remove completed prio == INT_MAX optimization
--- drivers/gpu/drm/i915/i915_guc_submission.c | 1 - 1 file changed, 1 deletion(-)
drivers/gpu/drm/i915/i915_guc_submission.c +0 -1
@@ -637,7 +637,6 @@ static void i915_guc_dequeue(struct intel_engine_cs *engine) }INIT_LIST_HEAD (&rq->priotree.link); - rq->priotree.priority =INT_MAX ;__i915_gem_request_submit (rq);trace_i915_gem_request_in (rq,port_index (port, execlists));
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v3 10/12 drm/i915: Rename helpers used for unwinding, use macro for can_preempt
In reply to Michał WiniarskiWe would also like to make use of execlist_cancel_port_requests and unwind_incomplete_requests in GuC preemption backend.
Let's rename the functions to use the correct prefixes, so that we can simply add the declarations in the following patch.
Similar thing for applies for can_preempt, except we're introducing HAS_LOGICAL_RING_PREEMPTION macro instad, converting other users that were previously touching device info directly.
v2: s/intel_engine/execlists and pass execlists to unwind (Chris)
v3: use locked version for exporting, drop const qual (Chris)
--- drivers/gpu/drm/i915/i915_drv.c | 2 +- drivers/gpu/drm/i915/i915_drv.h | 2 ++ drivers/gpu/drm/i915/intel_engine_cs.c | 6 +++--- drivers/gpu/drm/i915/intel_lrc.c | 35 ++++++++++++++++++---------------- 4 files changed, 25 insertions(+), 20 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +1 -1
@@ -372,7 +372,7 @@ static int i915_getparam(struct drm_device *dev, void *data, value |=I915_SCHEDULER_CAP_ENABLED ; value |=I915_SCHEDULER_CAP_PRIORITY ; -if (INTEL_INFO (dev_priv)->has_logical_ring_preemption && +if (HAS_LOGICAL_RING_PREEMPTION (dev_priv) && i915_modparams.enable_execlists && !i915_modparams.enable_guc_submission) value |=I915_SCHEDULER_CAP_PREEMPTION ;
drivers/gpu/drm/i915/i915_drv.h +2 -0
@@ -3140,6 +3140,8 @@ intel_info(const struct drm_i915_private *dev_priv)#define HAS_LOGICAL_RING_CONTEXTS (dev_priv) \ ((dev_priv)->info.has_logical_ring_contexts) +#define HAS_LOGICAL_RING_PREEMPTION (dev_priv) \ + ((dev_priv)->info.has_logical_ring_preemption)#define USES_PPGTT (dev_priv) (i915_modparams.enable_ppgtt)#define USES_FULL_PPGTT (dev_priv) (i915_modparams.enable_ppgtt >= 2)#define USES_FULL_48BIT_PPGTT (dev_priv) (i915_modparams.enable_ppgtt == 3)
drivers/gpu/drm/i915/intel_engine_cs.c +3 -3
@@ -620,7 +620,7 @@ int intel_engine_init_common(struct intel_engine_cs *engine)* Similarly the preempt context must always be available so that * we can interrupt the engine at any time. */ -if (INTEL_INFO (engine->i915)->has_logical_ring_preemption) { +if (HAS_LOGICAL_RING_PREEMPTION (engine->i915)) { ring = engine->context_pin (engine, engine->i915->preempt_context);if (IS_ERR (ring)) {
@@ -651,7 +651,7 @@ int intel_engine_init_common(struct intel_engine_cs *engine) err_breadcrumbs:intel_engine_fini_breadcrumbs (engine); err_unpin_preempt: -if (INTEL_INFO (engine->i915)->has_logical_ring_preemption) +if (HAS_LOGICAL_RING_PREEMPTION (engine->i915)) engine->context_unpin (engine, engine->i915->preempt_context); err_unpin_kernel: engine->context_unpin (engine, engine->i915->kernel_context);
@@ -679,7 +679,7 @@ void intel_engine_cleanup_common(struct intel_engine_cs *engine)intel_engine_cleanup_cmd_parser (engine);i915_gem_batch_pool_fini (&engine->batch_pool); -if (INTEL_INFO (engine->i915)->has_logical_ring_preemption) +if (HAS_LOGICAL_RING_PREEMPTION (engine->i915)) engine->context_unpin (engine, engine->i915->preempt_context); engine->context_unpin (engine, engine->i915->kernel_context); }
drivers/gpu/drm/i915/intel_lrc.c +19 -16
@@ -354,7 +354,7 @@ static void unwind_wa_tail(struct drm_i915_gem_request *rq)assert_ring_tail_valid (rq->ring, rq->tail); } -static void unwind_incomplete_requests (struct intel_engine_cs *engine) +static void __unwind_incomplete_requests (struct intel_engine_cs *engine) {struct drm_i915_gem_request *rq, *rn;struct i915_priolist *uninitialized_var (p );
@@ -385,6 +385,17 @@ static void unwind_incomplete_requests(struct intel_engine_cs *engine) } } +static void +execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists) +{ +struct intel_engine_cs *engine = +container_of (execlists,typeof (*engine), execlists); + +spin_lock_irq (&engine->timeline->lock); +__unwind_incomplete_requests (engine); +spin_unlock_irq (&engine->timeline->lock); +} +static inline void execlists_context_status_change (struct drm_i915_gem_request *rq,unsigned long status)
@@ -515,11 +526,6 @@ static void inject_preempt_context(struct intel_engine_cs *engine)elsp_write (ce->lrc_desc, elsp); } -static bool can_preempt (struct intel_engine_cs *engine) -{ -return INTEL_INFO (engine->i915)->has_logical_ring_preemption; -} -static void execlists_dequeue (struct intel_engine_cs *engine) {struct intel_engine_execlists *const execlists = &engine->execlists;
@@ -567,7 +573,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine)if (port_count (&port[0 ]) >1 ) goto unlock; -if (can_preempt (engine) && +if (HAS_LOGICAL_RING_PREEMPTION (engine->i915) && rb_entry(rb, struct i915_priolist, node)->priority >max (last->priotree.priority,0 )) {/*
@@ -691,7 +697,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine) }static void -execlist_cancel_port_requests (struct intel_engine_execlists *execlists) +execlists_cancel_port_requests (struct intel_engine_execlists *const execlists) {struct execlist_port *port = execlists->port;unsigned int num_ports =execlists_num_ports (execlists);
@@ -718,7 +724,7 @@ static void execlists_cancel_requests(struct intel_engine_cs *engine)spin_lock_irqsave (&engine->timeline->lock, flags);/* Cancel the requests on the HW and clear the ELSP tracker. */ -execlist_cancel_port_requests (execlists); +execlists_cancel_port_requests (execlists);/* Mark all executing requests as skipped. */ list_for_each_entry (rq, &engine->timeline->requests, link) {
@@ -858,11 +864,8 @@ static void intel_lrc_irq_handler(unsigned long data)if (status &GEN8_CTX_STATUS_ACTIVE_IDLE && buf[2 *head +1 ] ==PREEMPT_ID ) { -execlist_cancel_port_requests (execlists); - -spin_lock_irq (&engine->timeline->lock); -unwind_incomplete_requests (engine); -spin_unlock_irq (&engine->timeline->lock); +execlists_cancel_port_requests (execlists); +execlists_unwind_incomplete_requests (execlists);GEM_BUG_ON (!execlists_is_active (execlists,EXECLISTS_ACTIVE_PREEMPT ));
@@ -1531,10 +1534,10 @@ static void reset_common_ring(struct intel_engine_cs *engine,* guessing the missed context-switch events by looking at what * requests were completed. */ -execlist_cancel_port_requests (execlists); +execlists_cancel_port_requests (execlists);/* Push back any incomplete requests for replay after the reset. */ -unwind_incomplete_requests (engine); +__unwind_incomplete_requests (engine);spin_unlock_irqrestore (&engine->timeline->lock, flags);
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v3 10/12] drm/i915: Rename helpers used for unwinding, use macro for can_preempt
Quoting Michał Winiarski (2017-10-25 21:00:18)
Quoted 8 lines
We would also like to make use of
execlist_cancel_port_requestsandunwind_incomplete_requestsin GuC preemption backend.
Let's rename the functions to use the correct prefixes, so that we can simply add the declarations in the following patch.
Similar thing for applies forcan_preempt, except we're introducingHAS_LOGICAL_RING_PREEMPTIONmacro instad, converting other users that were previously touching device info directly.
Quoted 7 lines
v2: s/intel_engine/execlists and pass execlists to unwind (Chris)
v3: use locked version for exporting, drop const qual (Chris)Signed-off-by Michał Winiarski <[email protected]>Cc Chris Wilson <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Michal Wajdeczko <[email protected]>
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v4 11/12 drm/i915/guc: Preemption! With GuC
In reply to Michał WiniarskiPretty similar to what we have on execlists.
We're reusing most of the GEM code, however, due to GuC quirks we need a couple of extra bits.
Preemption is implemented as GuC action, and actions can be pretty slow. Because of that, we're using a mutex to serialize them. Since we're requesting preemption from the tasklet, the task of creating a workitem and wrapping it in GuC action is delegated to a worker.
To distinguish that preemption has finished, we're using additional piece of HWSP, and since we're not getting context switch interrupts, we're also adding a user interrupt.
The fact that our special preempt context has completed unfortunately doesn't mean that we're ready to submit new work. We also need to wait for GuC to finish its own processing.
v2: Don't compile out the wait for GuC, handle workqueue flush on reset, no need for ordered workqueue, put on a reviewer hat when looking at my own patches (Chris)
Move struct work around in intel_guc, move user interruput outside of conditional (Michał)
Keep ring around rather than chase though intel_context
v3: Extract WA for flushing ggtt writes to a helper (Chris) Keep work_struct in intel_guc rather than engine (Michał) Use ordered workqueue for inject_preempt worker to avoid GuC quirks.
v4: Drop now unused INTEL_GUC_PREEMPT_OPTION_IMMEDIATE (Daniele) Drop stray newlines, use container_of for intel_guc in worker, check for presence of workqueue when flushing it, rather than enable_guc_submission modparam, reorder preempt postprocessing (Chris)
--- drivers/gpu/drm/i915/i915_drv.c | 3 +- drivers/gpu/drm/i915/i915_gem.c | 10 ++ drivers/gpu/drm/i915/i915_guc_submission.c | 208 +++++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_guc.h | 8 ++ drivers/gpu/drm/i915/intel_lrc.c | 4 +- drivers/gpu/drm/i915/intel_lrc.h | 1 - drivers/gpu/drm/i915/intel_ringbuffer.h | 6 + 7 files changed, 222 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +1 -2
@@ -373,8 +373,7 @@ static int i915_getparam(struct drm_device *dev, void *data, value |=I915_SCHEDULER_CAP_PRIORITY ;if (HAS_LOGICAL_RING_PREEMPTION (dev_priv) && - i915_modparams.enable_execlists && - !i915_modparams.enable_guc_submission) + i915_modparams.enable_execlists) value |=I915_SCHEDULER_CAP_PREEMPTION ; }break ;
drivers/gpu/drm/i915/i915_gem.c +10 -0
@@ -2921,6 +2921,16 @@ i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)tasklet_kill (&engine->execlists.irq_tasklet);tasklet_disable (&engine->execlists.irq_tasklet); +/* +* We're using worker to queue preemption requests from the tasklet in +* GuC submission mode. +* Even though tasklet was disabled, we may still have a worker queued. +* Let's make sure that all workers scheduled before disabling the +* tasklet are completed before continuing with the reset. +*/ +if (engine->i915->guc.preempt_wq) +flush_workqueue (engine->i915->guc.preempt_wq); +if (engine->irq_seqno_barrier) engine->irq_seqno_barrier (engine);
drivers/gpu/drm/i915/i915_guc_submission.c +195 -13
282 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +8 -0
@@ -34,6 +34,11 @@#include "i915_guc_reg.h" #include "i915_vma.h" +struct guc_preempt_work { +struct intel_engine_cs *engine; +struct work_struct work; +}; +/* * Top level structure of GuC. It handles firmware loading and manages client * pool and doorbells. intel_guc owns a i915_guc_client to replace the legacy
@@ -60,6 +65,9 @@ struct intel_guc {struct i915_guc_client *execbuf_client;struct i915_guc_client *preempt_client; +struct guc_preempt_work preempt_work[I915_NUM_ENGINES ]; +struct workqueue_struct *preempt_wq; +DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */ u32 db_cacheline;
drivers/gpu/drm/i915/intel_lrc.c +2 -2
@@ -385,7 +385,7 @@ static void __unwind_incomplete_requests(struct intel_engine_cs *engine) } } -static void +void execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists) { struct intel_engine_cs *engine =
@@ -696,7 +696,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine) } } -static void +void execlists_cancel_port_requests (struct intel_engine_execlists *const execlists) {struct execlist_port *port = execlists->port;
drivers/gpu/drm/i915/intel_lrc.h +0 -1
@@ -107,7 +107,6 @@ intel_lr_context_descriptor(struct i915_gem_context *ctx,return ctx->engine[engine->id].lrc_desc; } -/* Execlists */ int intel_sanitize_enable_execlists (struct drm_i915_private *dev_priv,int enable_execlists);
drivers/gpu/drm/i915/intel_ringbuffer.h +6 -0
@@ -557,6 +557,12 @@ execlists_is_active(const struct intel_engine_execlists *execlists,return test_bit (bit, (unsigned long *)&execlists->active); } +void +execlists_cancel_port_requests (struct intel_engine_execlists *const execlists); + +void +execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists); +static inline unsigned int execlists_num_ports (const struct intel_engine_execlists *const execlists) {
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v4 11/12] drm/i915/guc: Preemption! With GuC
Quoting Michał Winiarski (2017-10-25 21:00:19)
Quoted 50 lines
Pretty similar to what we have on execlists.
We're reusing most of the GEM code, however, due to GuC quirks we need a couple of extra bits.
Preemption is implemented as GuC action, and actions can be pretty slow. Because of that, we're using a mutex to serialize them. Since we're requesting preemption from the tasklet, the task of creating a workitem and wrapping it in GuC action is delegated to a worker.To distinguish that preemption has finished, we're using additional piece of HWSP, and since we're not getting context switch interrupts, we're also adding a user interrupt.
The fact that our special preempt context has completed unfortunately doesn't mean that we're ready to submit new work. We also need to wait for GuC to finish its own processing.
v2: Don't compile out the wait for GuC, handle workqueue flush on reset, no need for ordered workqueue, put on a reviewer hat when looking at my own patches (Chris)
Move struct work around inintel_guc, move user interruput outside of conditional (Michał)
Keep ring around rather than chase thoughintel_contextv3: Extract WA for flushing ggtt writes to a helper (Chris) Keep
work_structinintel_gucrather than engine (Michał) Use ordered workqueue forinject_preemptworker to avoid GuC quirks.v4: Drop now unused
INTEL_GUC_PREEMPT_OPTION_IMMEDIATE(Daniele) Drop stray newlines, usecontainer_offorintel_gucin worker, check for presence of workqueue when flushing it, rather thanenable_guc_submissionmodparam, reorder preempt postprocessing (Chris)Signed-off-by Michał Winiarski <[email protected]>Cc Chris Wilson <[email protected]>Cc Jeff McGee <[email protected]>Cc Joonas Lahtinen <[email protected]>Cc Oscar Mateo <[email protected]>Cc Tvrtko Ursulin <[email protected]>---
drivers/gpu/drm/i915/i915_guc_submission.c +11
-0+#define GUC_PREEMPT_POSTPROCESS_DELAY_MS 10 +static void wait_for_guc_preempt_report (struct intel_engine_cs *engine) +{ +struct intel_guc *guc = &engine->i915->guc; +struct guc_shared_ctx_data *data = guc->shared_data_vaddr; +struct guc_ctx_report *report = + &data->preempt_ctx_report[engine->guc_id]; + +WARN_ON (wait_for_atomic (report->report_return_status == +INTEL_GUC_REPORT_STATUS_COMPLETE , +GUC_PREEMPT_POSTPROCESS_DELAY_MS ));
drivers/gpu/drm/i915/i915_guc_submission.c static void guc_add_request(struct intel_guc *guc, · line 666
Quoted 10 lines
drivers/gpu/drm/i915/i915_guc_submission.c +10
-0+/* +* GuC is expecting that we're also going to clear the affected context +* counter, let's also reset the return status to not depend on GuC +* resetting it after recieving another preempt action +*/ + report->affected_count =0 ; + report->report_return_status =INTEL_GUC_REPORT_STATUS_UNKNOWN ; +} + +
The horror! ;)
Looks like everything is in order. Preemptive
I shall ponder the question of whether we can stress these (guc/execlists) preemption paths more. We have checks that we can preempt and reorder in-flight requests; we have simple smoketests. What I feel is like we probably need more of the latter to find timing issues. (A new mode for gem_concurrent_blit? There's nothing like running a test for a few days to shake out timing bugs.) Unless I've missed a rule we can test in gem_exec_schedule.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Re: [PATCH v4 11/12] drm/i915/guc: Preemption! With GuC
In reply to Michał WiniarskiQuoting Michał Winiarski (2017-10-25 21:00:19)
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +2
-0@@ -2921,6 +2921,16 @@ i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)tasklet_kill (&engine->execlists.irq_tasklet);tasklet_disable (&engine->execlists.irq_tasklet); +/* +* We're using worker to queue preemption requests from the tasklet in
drivers/gpu/drm/i915/i915_gem.c i915_gem_reset_prepare_engine(struct intel_engine_cs *engine) · line 2935
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +8
-0+* GuC submission mode. +* Even though tasklet was disabled, we may still have a worker queued. +* Let's make sure that all workers scheduled before disabling the +* tasklet are completed before continuing with the reset. +*/ +if (engine->i915->guc.preempt_wq) +flush_workqueue (engine->i915->guc.preempt_wq); +if (engine->irq_seqno_barrier) engine->irq_seqno_barrier (engine);
https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6192/fi-glk-dsi/igt@[email protected]
[ 31.382512] i915 0000:00:02.0: Resetting rcs0 after gpu hang [ 31.382864] INFO: trying to register non-static key. [ 31.382871] the code is fine but needs lockdep annotation. [ 31.382874] turning off the locking correctness validator. [ 31.382879] CPU: 0 PID: 1473 Comm: drv_hangman Not tainted 4.14.0-rc6-CI-Patchwork_6192+ #1 [ 31.382884] Hardware name: Intel Corp. Geminilake/GLK RVP2 LP4SD (07), BIOS GELKRVPA.X64.0062.B30.1708222146 08/22/2017 [ 31.382890] Call Trace: [ 31.382898] dump_stack+0x68/0x9f [ 31.382903] register_lock_class+0x3fd/0x580 [ 31.382907] ? __bfs+0x129/0x210 [ 31.382912] __lock_acquire+0xa4/0x1b00 [ 31.382916] ? flush_workqueue+0x75/0x520 [ 31.382921] ? __raw_spin_lock_init+0x21/0x60 [ 31.382925] lock_acquire+0xb0/0x200 [ 31.382929] ? lock_acquire+0xb0/0x200 [ 31.382932] ? flush_workqueue+0x75/0x520 [ 31.382936] flush_workqueue+0x98/0x520 [ 31.382939] ? flush_workqueue+0x75/0x520 [ 31.382944] ? _raw_spin_unlock_irq+0x37/0x50 [ 31.382991] i915_gem_reset_prepare_engine+0x5b/0x90 [i915] [ 31.383024] ? i915_gem_reset_prepare_engine+0x5b/0x90 [i915] [ 31.383055] i915_reset_engine+0x43/0x100 [i915] [ 31.383087] i915_handle_error+0x1ee/0x430 [i915] [ 31.383095] ? __might_fault+0x3e/0x90 [ 31.383127] i915_wedged_set+0x84/0xd0 [i915] [ 31.383132] simple_attr_write+0xb4/0xd0 [ 31.383138] full_proxy_write+0x54/0x80 [ 31.383143] __vfs_write+0x28/0x130 [ 31.383148] ? rcu_read_lock_sched_held+0x7a/0x90 [ 31.383152] ? rcu_sync_lockdep_assert+0x2f/0x60 [ 31.383157] ? __sb_start_write+0x10c/0x200 [ 31.383161] vfs_write+0xc8/0x1c0 [ 31.383164] SyS_write+0x49/0xb0 [ 31.383169] entry_SYSCALL_64_fastpath+0x1c/0xb1 [ 31.383173] RIP: 0033:0x7f3492708670 [ 31.383177] RSP: 002b:00007fffcbef7258 EFLAGS: 00000246 ORIG_RAX: 0000000000000001 [ 31.383182] RAX: ffffffffffffffda RBX: ffffffff81491ef3 RCX: 00007f3492708670 [ 31.383186] RDX: 0000000000000002 RSI: 00005637752962c5 RDI: 0000000000000007 [ 31.383190] RBP: ffffc900022abf88 R08: 0000563776b75df0 R09: 0000000000000000 [ 31.383195] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000 [ 31.383199] R13: 00007fffcbef7660 R14: 000056377527ea10 R15: 0000000000000007 [ 31.383205] ? __this_cpu_preempt_check+0x13/0x20 [ 31.383232] BUG: unable to handle kernel paging request at 0000000100000058
drivers/gpu/drm/i915/i915_gem.c i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)
Quoted 10 lines
drivers/gpu/drm/i915/i915_gem.c +10
-0+static void guc_preempt_work_destroy (struct intel_guc *guc) +{ +struct drm_i915_private *dev_priv =guc_to_i915 (guc); +struct intel_engine_cs *engine; +enum intel_engine_id id; + +for_each_engine (engine, dev_priv, id) +cancel_work_sync (&guc->preempt_work[id].work); + +destroy_workqueue (guc->preempt_wq);
guc->preempt_wq = NULL;
glk is special in that CI run for trying and failing to enable the GuC due to no fw. (Do I hear the noise of a combinatorial explosion in the background?)
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v5 1/1 drm/i915/guc: Preemption! With GuC
In reply to Michał WiniarskiPretty similar to what we have on execlists.
We're reusing most of the GEM code, however, due to GuC quirks we need a couple of extra bits.
Preemption is implemented as GuC action, and actions can be pretty slow. Because of that, we're using a mutex to serialize them. Since we're requesting preemption from the tasklet, the task of creating a workitem and wrapping it in GuC action is delegated to a worker.
To distinguish that preemption has finished, we're using additional piece of HWSP, and since we're not getting context switch interrupts, we're also adding a user interrupt.
The fact that our special preempt context has completed unfortunately doesn't mean that we're ready to submit new work. We also need to wait for GuC to finish its own processing.
v2: Don't compile out the wait for GuC, handle workqueue flush on reset, no need for ordered workqueue, put on a reviewer hat when looking at my own patches (Chris)
Move struct work around in intel_guc, move user interruput outside of conditional (Michał)
Keep ring around rather than chase though intel_context
v3: Extract WA for flushing ggtt writes to a helper (Chris) Keep work_struct in intel_guc rather than engine (Michał) Use ordered workqueue for inject_preempt worker to avoid GuC quirks.
v4: Drop now unused INTEL_GUC_PREEMPT_OPTION_IMMEDIATE (Daniele) Drop stray newlines, use container_of for intel_guc in worker, check for presence of workqueue when flushing it, rather than enable_guc_submission modparam, reorder preempt postprocessing (Chris)
v5: Make wq NULL after destroying it
--- drivers/gpu/drm/i915/i915_drv.c | 3 +- drivers/gpu/drm/i915/i915_gem.c | 10 ++ drivers/gpu/drm/i915/i915_guc_submission.c | 208 +++++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_guc.h | 8 ++ drivers/gpu/drm/i915/intel_lrc.c | 4 +- drivers/gpu/drm/i915/intel_lrc.h | 1 - drivers/gpu/drm/i915/intel_ringbuffer.h | 6 + 7 files changed, 222 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +1 -2
@@ -373,8 +373,7 @@ static int i915_getparam(struct drm_device *dev, void *data, value |=I915_SCHEDULER_CAP_PRIORITY ;if (HAS_LOGICAL_RING_PREEMPTION (dev_priv) && - i915_modparams.enable_execlists && - !i915_modparams.enable_guc_submission) + i915_modparams.enable_execlists) value |=I915_SCHEDULER_CAP_PREEMPTION ; }break ;
drivers/gpu/drm/i915/i915_gem.c +10 -0
@@ -2921,6 +2921,16 @@ i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)tasklet_kill (&engine->execlists.irq_tasklet);tasklet_disable (&engine->execlists.irq_tasklet); +/* +* We're using worker to queue preemption requests from the tasklet in +* GuC submission mode. +* Even though tasklet was disabled, we may still have a worker queued. +* Let's make sure that all workers scheduled before disabling the +* tasklet are completed before continuing with the reset. +*/ +if (engine->i915->guc.preempt_wq) +flush_workqueue (engine->i915->guc.preempt_wq); +if (engine->irq_seqno_barrier) engine->irq_seqno_barrier (engine);
drivers/gpu/drm/i915/i915_guc_submission.c +195 -13
282 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +8 -0
@@ -34,6 +34,11 @@#include "i915_guc_reg.h" #include "i915_vma.h" +struct guc_preempt_work { +struct intel_engine_cs *engine; +struct work_struct work; +}; +/* * Top level structure of GuC. It handles firmware loading and manages client * pool and doorbells. intel_guc owns a i915_guc_client to replace the legacy
@@ -60,6 +65,9 @@ struct intel_guc {struct i915_guc_client *execbuf_client;struct i915_guc_client *preempt_client; +struct guc_preempt_work preempt_work[I915_NUM_ENGINES ]; +struct workqueue_struct *preempt_wq; +DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */ u32 db_cacheline;
drivers/gpu/drm/i915/intel_lrc.c +2 -2
@@ -385,7 +385,7 @@ static void __unwind_incomplete_requests(struct intel_engine_cs *engine) } } -static void +void execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists) { struct intel_engine_cs *engine =
@@ -696,7 +696,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine) } } -static void +void execlists_cancel_port_requests (struct intel_engine_execlists *const execlists) {struct execlist_port *port = execlists->port;
drivers/gpu/drm/i915/intel_lrc.h +0 -1
@@ -107,7 +107,6 @@ intel_lr_context_descriptor(struct i915_gem_context *ctx,return ctx->engine[engine->id].lrc_desc; } -/* Execlists */ int intel_sanitize_enable_execlists (struct drm_i915_private *dev_priv,int enable_execlists);
drivers/gpu/drm/i915/intel_ringbuffer.h +6 -0
@@ -557,6 +557,12 @@ execlists_is_active(const struct intel_engine_execlists *execlists,return test_bit (bit, (unsigned long *)&execlists->active); } +void +execlists_cancel_port_requests (struct intel_engine_execlists *const execlists); + +void +execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists); +static inline unsigned int execlists_num_ports (const struct intel_engine_execlists *const execlists) {
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH v6 1/1 drm/i915/guc: Preemption! With GuC
Pretty similar to what we have on execlists.
We're reusing most of the GEM code, however, due to GuC quirks we need a couple of extra bits.
Preemption is implemented as GuC action, and actions can be pretty slow. Because of that, we're using a mutex to serialize them. Since we're requesting preemption from the tasklet, the task of creating a workitem and wrapping it in GuC action is delegated to a worker.
To distinguish that preemption has finished, we're using additional piece of HWSP, and since we're not getting context switch interrupts, we're also adding a user interrupt.
The fact that our special preempt context has completed unfortunately doesn't mean that we're ready to submit new work. We also need to wait for GuC to finish its own processing.
v2: Don't compile out the wait for GuC, handle workqueue flush on reset, no need for ordered workqueue, put on a reviewer hat when looking at my own patches (Chris)
Move struct work around in intel_guc, move user interruput outside of conditional (Michał)
Keep ring around rather than chase though intel_context
v3: Extract WA for flushing ggtt writes to a helper (Chris) Keep work_struct in intel_guc rather than engine (Michał) Use ordered workqueue for inject_preempt worker to avoid GuC quirks.
v4: Drop now unused INTEL_GUC_PREEMPT_OPTION_IMMEDIATE (Daniele) Drop stray newlines, use container_of for intel_guc in worker, check for presence of workqueue when flushing it, rather than enable_guc_submission modparam, reorder preempt postprocessing (Chris)
v5: Make wq NULL after destroying it
v6: Swap struct guc_preempt_work members (Michał)
--- drivers/gpu/drm/i915/i915_drv.c | 3 +- drivers/gpu/drm/i915/i915_gem.c | 10 ++ drivers/gpu/drm/i915/i915_guc_submission.c | 208 +++++++++++++++++++++++++++-- drivers/gpu/drm/i915/intel_guc.h | 8 ++ drivers/gpu/drm/i915/intel_lrc.c | 4 +- drivers/gpu/drm/i915/intel_lrc.h | 1 - drivers/gpu/drm/i915/intel_ringbuffer.h | 6 + 7 files changed, 222 insertions(+), 18 deletions(-)
drivers/gpu/drm/i915/i915_drv.c +1 -2
@@ -373,8 +373,7 @@ static int i915_getparam(struct drm_device *dev, void *data, value |=I915_SCHEDULER_CAP_PRIORITY ;if (HAS_LOGICAL_RING_PREEMPTION (dev_priv) && - i915_modparams.enable_execlists && - !i915_modparams.enable_guc_submission) + i915_modparams.enable_execlists) value |=I915_SCHEDULER_CAP_PREEMPTION ; }break ;
drivers/gpu/drm/i915/i915_gem.c +10 -0
@@ -2921,6 +2921,16 @@ i915_gem_reset_prepare_engine(struct intel_engine_cs *engine)tasklet_kill (&engine->execlists.irq_tasklet);tasklet_disable (&engine->execlists.irq_tasklet); +/* +* We're using worker to queue preemption requests from the tasklet in +* GuC submission mode. +* Even though tasklet was disabled, we may still have a worker queued. +* Let's make sure that all workers scheduled before disabling the +* tasklet are completed before continuing with the reset. +*/ +if (engine->i915->guc.preempt_wq) +flush_workqueue (engine->i915->guc.preempt_wq); +if (engine->irq_seqno_barrier) engine->irq_seqno_barrier (engine);
drivers/gpu/drm/i915/i915_guc_submission.c +195 -13
282 lines not on this page · raw message
drivers/gpu/drm/i915/intel_guc.h +8 -0
@@ -34,6 +34,11 @@#include "i915_guc_reg.h" #include "i915_vma.h" +struct guc_preempt_work { +struct work_struct work; +struct intel_engine_cs *engine; +}; +/* * Top level structure of GuC. It handles firmware loading and manages client * pool and doorbells. intel_guc owns a i915_guc_client to replace the legacy
@@ -60,6 +65,9 @@ struct intel_guc {struct i915_guc_client *execbuf_client;struct i915_guc_client *preempt_client; +struct guc_preempt_work preempt_work[I915_NUM_ENGINES ]; +struct workqueue_struct *preempt_wq; +DECLARE_BITMAP (doorbell_bitmap,GUC_NUM_DOORBELLS );/* Cyclic counter mod pagesize */ u32 db_cacheline;
drivers/gpu/drm/i915/intel_lrc.c +2 -2
@@ -385,7 +385,7 @@ static void __unwind_incomplete_requests(struct intel_engine_cs *engine) } } -static void +void execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists) { struct intel_engine_cs *engine =
@@ -696,7 +696,7 @@ static void execlists_dequeue(struct intel_engine_cs *engine) } } -static void +void execlists_cancel_port_requests (struct intel_engine_execlists *const execlists) {struct execlist_port *port = execlists->port;
drivers/gpu/drm/i915/intel_lrc.h +0 -1
@@ -107,7 +107,6 @@ intel_lr_context_descriptor(struct i915_gem_context *ctx,return ctx->engine[engine->id].lrc_desc; } -/* Execlists */ int intel_sanitize_enable_execlists (struct drm_i915_private *dev_priv,int enable_execlists);
drivers/gpu/drm/i915/intel_ringbuffer.h +6 -0
@@ -557,6 +557,12 @@ execlists_is_active(const struct intel_engine_execlists *execlists,return test_bit (bit, (unsigned long *)&execlists->active); } +void +execlists_cancel_port_requests (struct intel_engine_execlists *const execlists); + +void +execlists_unwind_incomplete_requests (struct intel_engine_execlists *execlists); +static inline unsigned int execlists_num_ports (const struct intel_engine_execlists *const execlists) {
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
PATCH 12/12 HAX Enable GuC Submission for CI
In reply to Michał WiniarskiAlso:
Revert "drm/i915/guc: Assert that we switch between known ggtt->invalidate functions"
This reverts commit 04f7b24eccdfae680a36e9825fe0d61dcd5ed528.
--- drivers/gpu/drm/i915/i915_gem_gtt.c | 8 ++------ drivers/gpu/drm/i915/i915_params.h | 4 ++-- 2 files changed, 4 insertions(+), 8 deletions(-)
drivers/gpu/drm/i915/i915_gem_gtt.c +2 -6
@@ -3568,17 +3568,13 @@ int i915_ggtt_enable_hw(struct drm_i915_private *dev_priv)void i915_ggtt_enable_guc (struct drm_i915_private *i915) { -GEM_BUG_ON (i915->ggtt.invalidate != gen6_ggtt_invalidate); - i915->ggtt.invalidate = guc_ggtt_invalidate; }void i915_ggtt_disable_guc (struct drm_i915_private *i915) { -/* We should only be called after i915_ggtt_enable_guc() */ -GEM_BUG_ON (i915->ggtt.invalidate != guc_ggtt_invalidate); - - i915->ggtt.invalidate = gen6_ggtt_invalidate; +if (i915->ggtt.invalidate == guc_ggtt_invalidate) + i915->ggtt.invalidate = gen6_ggtt_invalidate; }void i915_gem_restore_gtt_mappings (struct drm_i915_private *dev_priv)
drivers/gpu/drm/i915/i915_params.h +2 -2
@@ -44,8 +44,8 @@ param(int, disable_power_well, -1) \ param(int, enable_ips, 1) \ param(int, invert_brightness, 0) \ - param(int, enable_guc_loading, 0) \ - param(int, enable_guc_submission, 0) \ + param(int, enable_guc_loading, 1) \ + param(int, enable_guc_submission, 1) \ param(int, guc_log_level, -1) \ param(char *, guc_firmware_path, NULL) \param (char *, huc_firmware_path, NULL) \
Signature · 6 lines
2.13.6 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: failure for Preemption with GuC, fourth try
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try
URL : https://patchwork.freedesktop.org/series/32654/
State : failure
== Summary ==
Series 32654v1 Preemption with GuC, fourth try
https://patchwork.freedesktop.org/api/1.0/series/32654/revisions/1/mbox/
Test chamelium:
Subgroup dp-crc-fast:
fail -> PASS (fi-kbl-7500u) fdo#102514
Test debugfs_test:
Subgroup read_all_entries:
pass -> DMESG-WARN (fi-cfl-s)
Test drv_getparams_basic:
Subgroup basic-eu-total:
pass -> DMESG-WARN (fi-cfl-s)
Subgroup basic-subslice-total:
pass -> DMESG-WARN (fi-cfl-s)
Test drv_hangman:
Subgroup error-state-basic:
pass -> DMESG-FAIL (fi-glk-1)
pass -> DMESG-FAIL (fi-glk-dsi)
pass -> DMESG-WARN (fi-cfl-s)
Test gem_basic:
Subgroup bad-close:
pass -> INCOMPLETE (fi-glk-1)
pass -> INCOMPLETE (fi-glk-dsi)
Test gem_exec_reloc:
Subgroup basic-gtt-active:
pass -> FAIL (fi-gdg-551) fdo#102582 +2
Subgroup basic-write-cpu-active:
pass -> FAIL (fi-gdg-551)
Subgroup basic-write-gtt-active:
pass -> FAIL (fi-gdg-551) fdo#102582
Test gem_sync:
Subgroup basic-many-each:
pass -> FAIL (fi-kbl-7567u) fdo#103165
Test kms_flip:
Subgroup basic-flip-vs-wf_vblank:
fail -> PASS (fi-skl-6770hq)
Test kms_setmode:
Subgroup basic-clone-single-crtc:
pass -> INCOMPLETE (fi-cnl-y)fdo#102514 https://bugs.freedesktop.org/show_bug.cgi?id=102514
fdo#102582 https://bugs.freedesktop.org/show_bug.cgi?id=102582
fdo#102582 https://bugs.freedesktop.org/show_bug.cgi?id=102582
fdo#103165 https://bugs.freedesktop.org/show_bug.cgi?id=103165
fi-bdw-5557u total:289 pass:268 dwarn:0 dfail:0 fail:0 skip:21 time:441s fi-bdw-gvtdvm total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:451s fi-blb-e6850 total:289 pass:223 dwarn:1 dfail:0 fail:0 skip:65 time:377s fi-bsw-n3050 total:289 pass:243 dwarn:0 dfail:0 fail:0 skip:46 time:537s fi-bwr-2160 total:289 pass:183 dwarn:0 dfail:0 fail:0 skip:106 time:262s fi-bxt-dsi total:289 pass:259 dwarn:0 dfail:0 fail:0 skip:30 time:500s fi-bxt-j4205 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:508s fi-byt-j1900 total:289 pass:253 dwarn:1 dfail:0 fail:0 skip:35 time:488s fi-byt-n2820 total:289 pass:249 dwarn:1 dfail:0 fail:0 skip:39 time:487s fi-cfl-s total:289 pass:249 dwarn:8 dfail:0 fail:0 skip:32 time:562s fi-cnl-y total:289 pass:224 dwarn:0 dfail:0 fail:0 skip:24 fi-elk-e7500 total:289 pass:229 dwarn:0 dfail:0 fail:0 skip:60 time:417s fi-gdg-551 total:289 pass:173 dwarn:1 dfail:0 fail:6 skip:109 time:250s fi-glk-1 total:16 pass:5 dwarn:0 dfail:1 fail:0 skip:9 fi-glk-dsi total:16 pass:5 dwarn:0 dfail:1 fail:0 skip:9 fi-hsw-4770 total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:429s fi-hsw-4770r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:424s fi-ilk-650 total:289 pass:228 dwarn:0 dfail:0 fail:0 skip:61 time:434s fi-ivb-3520m total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:493s fi-ivb-3770 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:462s fi-kbl-7500u total:289 pass:264 dwarn:1 dfail:0 fail:0 skip:24 time:487s fi-kbl-7560u total:289 pass:270 dwarn:0 dfail:0 fail:0 skip:19 time:571s fi-kbl-7567u total:289 pass:268 dwarn:0 dfail:0 fail:1 skip:20 time:483s fi-kbl-r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:582s fi-pnv-d510 total:289 pass:222 dwarn:1 dfail:0 fail:0 skip:66 time:549s fi-skl-6260u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:447s fi-skl-6600u total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:597s fi-skl-6700hq total:289 pass:263 dwarn:0 dfail:0 fail:0 skip:26 time:651s fi-skl-6700k total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:529s fi-skl-6770hq total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:498s fi-skl-gvtdvm total:289 pass:266 dwarn:0 dfail:0 fail:0 skip:23 time:458s fi-snb-2520m total:289 pass:250 dwarn:0 dfail:0 fail:0 skip:39 time:565s fi-snb-2600 total:289 pass:249 dwarn:0 dfail:0 fail:0 skip:40 time:423s
be67da218126edb88de92c391a48964c89219c11 drm-tip: 2017y-10m-25d-17h-36m-48s UTC integration manifest 51a3ae75372d HAX Enable GuC Submission for CI30e2ce646bf4 drm/i915/guc: Preemption! With GuCee2027e659d3 drm/i915: Rename helpers used for unwinding, use macro for can_preempt 24e661557be6 drm/i915/guc: Keep request->priority for its lifetimedbb9d7ee0e47 drm/i915: Add information needed to track engine preempt state 047a7e6c2786 drm/i915: Extract "emit write" part of emit breadcrumb functions c6dc70581913 drm/i915/guc: Split guc_wq_item_append2fb58671aa1c drm/i915/guc: Add a second client, to be used for preemption ce8846251ead drm/i915/guc: Add preemption action to GuC firmware interface 6565d972ea35 drm/i915/guc: Allocate separate shared data object for GuC communication 336ed6330546 drm/i915/guc: Extract GuC stage desc pool creation into a helper b84d980555d1 drm/i915/guc: Do not use 0 for GuC doorbell cookie
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6192/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: warning for Preemption with GuC, fourth try (rev2)
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try (rev2)
URL : https://patchwork.freedesktop.org/series/32654/
State : warning
== Summary ==
Series 32654v2 Preemption with GuC, fourth try
https://patchwork.freedesktop.org/api/1.0/series/32654/revisions/2/mbox/
Test drv_module_reload:
Subgroup basic-reload:
pass -> DMESG-WARN (fi-cnl-y)
Subgroup basic-no-display:
pass -> DMESG-WARN (fi-cnl-y)
Subgroup basic-reload-inject:
pass -> DMESG-WARN (fi-cnl-y)fi-bdw-5557u total:289 pass:268 dwarn:0 dfail:0 fail:0 skip:21 time:439s fi-bdw-gvtdvm total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:444s fi-blb-e6850 total:289 pass:223 dwarn:1 dfail:0 fail:0 skip:65 time:370s fi-bsw-n3050 total:289 pass:243 dwarn:0 dfail:0 fail:0 skip:46 time:527s fi-bwr-2160 total:289 pass:183 dwarn:0 dfail:0 fail:0 skip:106 time:264s fi-bxt-dsi total:289 pass:259 dwarn:0 dfail:0 fail:0 skip:30 time:503s fi-bxt-j4205 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:503s fi-byt-j1900 total:289 pass:253 dwarn:1 dfail:0 fail:0 skip:35 time:492s fi-byt-n2820 total:289 pass:249 dwarn:1 dfail:0 fail:0 skip:39 time:470s fi-cfl-s total:289 pass:253 dwarn:4 dfail:0 fail:0 skip:32 time:551s fi-cnl-y total:289 pass:259 dwarn:3 dfail:0 fail:0 skip:27 time:620s fi-elk-e7500 total:289 pass:229 dwarn:0 dfail:0 fail:0 skip:60 time:419s fi-gdg-551 total:289 pass:178 dwarn:1 dfail:0 fail:1 skip:109 time:246s fi-glk-1 total:289 pass:261 dwarn:0 dfail:0 fail:0 skip:28 time:576s fi-glk-dsi total:289 pass:258 dwarn:0 dfail:0 fail:1 skip:30 time:482s fi-hsw-4770 total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:431s fi-hsw-4770r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:431s fi-ilk-650 total:289 pass:228 dwarn:0 dfail:0 fail:0 skip:61 time:442s fi-ivb-3520m total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:494s fi-ivb-3770 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:461s fi-kbl-7500u total:289 pass:264 dwarn:1 dfail:0 fail:0 skip:24 time:492s fi-kbl-7560u total:289 pass:270 dwarn:0 dfail:0 fail:0 skip:19 time:574s fi-kbl-7567u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:479s fi-kbl-r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:584s fi-pnv-d510 total:289 pass:222 dwarn:1 dfail:0 fail:0 skip:66 time:544s fi-skl-6260u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:449s fi-skl-6600u total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:590s fi-skl-6700hq total:289 pass:263 dwarn:0 dfail:0 fail:0 skip:26 time:648s fi-skl-6700k total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:519s fi-skl-6770hq total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:495s fi-skl-gvtdvm total:289 pass:266 dwarn:0 dfail:0 fail:0 skip:23 time:454s fi-snb-2520m total:289 pass:250 dwarn:0 dfail:0 fail:0 skip:39 time:555s fi-snb-2600 total:289 pass:249 dwarn:0 dfail:0 fail:0 skip:40 time:415s
2ea0b3d47030274c97624258e09fc7d1ffd0e0f2 drm-tip: 2017y-10m-25d-18h-42m-20s UTC integration manifest d8525b3f8fda HAX Enable GuC Submission for CIca766427cf88 drm/i915/guc: Preemption! With GuC5c250d581667 drm/i915: Rename helpers used for unwinding, use macro for can_preempt dee14752e8a2 drm/i915/guc: Keep request->priority for its lifetime8be229d8c330 drm/i915: Add information needed to track engine preempt state ebe667124190 drm/i915: Extract "emit write" part of emit breadcrumb functions e03bf3d7119c drm/i915/guc: Split guc_wq_item_appendf4bdf01a19d4 drm/i915/guc: Add a second client, to be used for preemption 55ce81eb7f4b drm/i915/guc: Add preemption action to GuC firmware interface 47cc469a334c drm/i915/guc: Allocate separate shared data object for GuC communication ced55050645b drm/i915/guc: Extract GuC stage desc pool creation into a helper 814910338e4f drm/i915/guc: Do not use 0 for GuC doorbell cookie
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6198/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: warning for Preemption with GuC, fourth try (rev3)
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try (rev3)
URL : https://patchwork.freedesktop.org/series/32654/
State : warning
== Summary ==
Series 32654v3 Preemption with GuC, fourth try
https://patchwork.freedesktop.org/api/1.0/series/32654/revisions/3/mbox/
Test chamelium:
Subgroup dp-crc-fast:
pass -> FAIL (fi-kbl-7500u) fdo#102514
Test kms_pipe_crc_basic:
Subgroup read-crc-pipe-b-frame-sequence:
skip -> PASS (fi-hsw-4770r) fdo#102332
Subgroup suspend-read-crc-pipe-b:
pass -> DMESG-WARN (fi-byt-j1900) fdo#101705
Test drv_module_reload:
Subgroup basic-reload:
pass -> DMESG-WARN (fi-cnl-y)
Subgroup basic-no-display:
pass -> DMESG-WARN (fi-cnl-y)
Subgroup basic-reload-inject:
pass -> DMESG-WARN (fi-cnl-y)fdo#102514 https://bugs.freedesktop.org/show_bug.cgi?id=102514
fdo#102332 https://bugs.freedesktop.org/show_bug.cgi?id=102332
fdo#101705 https://bugs.freedesktop.org/show_bug.cgi?id=101705
fi-bdw-5557u total:289 pass:268 dwarn:0 dfail:0 fail:0 skip:21 time:438s fi-bdw-gvtdvm total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:451s fi-blb-e6850 total:289 pass:223 dwarn:1 dfail:0 fail:0 skip:65 time:369s fi-bsw-n3050 total:289 pass:243 dwarn:0 dfail:0 fail:0 skip:46 time:528s fi-bwr-2160 total:289 pass:183 dwarn:0 dfail:0 fail:0 skip:106 time:263s fi-bxt-dsi total:289 pass:259 dwarn:0 dfail:0 fail:0 skip:30 time:503s fi-bxt-j4205 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:502s fi-byt-j1900 total:289 pass:253 dwarn:1 dfail:0 fail:0 skip:35 time:494s fi-byt-n2820 total:289 pass:249 dwarn:1 dfail:0 fail:0 skip:39 time:482s fi-cfl-s total:289 pass:253 dwarn:4 dfail:0 fail:0 skip:32 time:552s fi-cnl-y total:289 pass:259 dwarn:3 dfail:0 fail:0 skip:27 time:618s fi-elk-e7500 total:289 pass:229 dwarn:0 dfail:0 fail:0 skip:60 time:418s fi-gdg-551 total:289 pass:178 dwarn:1 dfail:0 fail:1 skip:109 time:249s fi-glk-1 total:289 pass:261 dwarn:0 dfail:0 fail:0 skip:28 time:574s fi-glk-dsi total:289 pass:258 dwarn:0 dfail:0 fail:1 skip:30 time:487s fi-hsw-4770 total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:426s fi-hsw-4770r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:425s fi-ilk-650 total:289 pass:228 dwarn:0 dfail:0 fail:0 skip:61 time:434s fi-ivb-3520m total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:492s fi-ivb-3770 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:463s fi-kbl-7500u total:289 pass:263 dwarn:1 dfail:0 fail:1 skip:24 time:480s fi-kbl-7560u total:289 pass:270 dwarn:0 dfail:0 fail:0 skip:19 time:575s fi-kbl-7567u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:478s fi-kbl-r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:581s fi-pnv-d510 total:289 pass:222 dwarn:1 dfail:0 fail:0 skip:66 time:545s fi-skl-6260u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:454s fi-skl-6600u total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:589s fi-skl-6700hq total:289 pass:263 dwarn:0 dfail:0 fail:0 skip:26 time:659s fi-skl-6700k total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:522s fi-skl-6770hq total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:501s fi-skl-gvtdvm total:289 pass:266 dwarn:0 dfail:0 fail:0 skip:23 time:456s fi-snb-2520m total:289 pass:250 dwarn:0 dfail:0 fail:0 skip:39 time:560s fi-snb-2600 total:289 pass:249 dwarn:0 dfail:0 fail:0 skip:40 time:418s
df3033b174059a59aa0c890f81de8af037abd11f drm-tip: 2017y-10m-26d-11h-03m-59s UTC integration manifest 3218d17f96a7 HAX Enable GuC Submission for CIa621438797f4 drm/i915/guc: Preemption! With GuCcac38ec21a03 drm/i915: Rename helpers used for unwinding, use macro for can_preempt d8b26d6919a9 drm/i915/guc: Keep request->priority for its lifetime794b8ad2405a drm/i915: Add information needed to track engine preempt state fd083843a919 drm/i915: Extract "emit write" part of emit breadcrumb functions 5b871dda8364 drm/i915/guc: Split guc_wq_item_appendb1bd6d916ab2 drm/i915/guc: Add a second client, to be used for preemption cb79905d0954 drm/i915/guc: Add preemption action to GuC firmware interface 18c35c003dcb drm/i915/guc: Allocate separate shared data object for GuC communication 5bda4fcf07ad drm/i915/guc: Extract GuC stage desc pool creation into a helper 87f3d391fb9f drm/i915/guc: Do not use 0 for GuC doorbell cookie
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6204/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.BAT: success for Preemption with GuC, fourth try (rev5)
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try (rev5)
URL : https://patchwork.freedesktop.org/series/32654/
State : success
== Summary ==
Series 32654v5 Preemption with GuC, fourth try
https://patchwork.freedesktop.org/api/1.0/series/32654/revisions/5/mbox/
Test kms_cursor_legacy:
Subgroup basic-flip-before-cursor-atomic:
pass -> INCOMPLETE (fi-cnl-y) fdo#102035
Test kms_pipe_crc_basic:
Subgroup read-crc-pipe-b-frame-sequence:
skip -> PASS (fi-hsw-4770r) fdo#102332
Subgroup suspend-read-crc-pipe-b:
pass -> DMESG-WARN (fi-byt-j1900) fdo#101705
Test drv_module_reload:
Subgroup basic-reload:
pass -> DMESG-WARN (fi-bsw-n3050) fdo#103165fdo#102035 https://bugs.freedesktop.org/show_bug.cgi?id=102035
fdo#102332 https://bugs.freedesktop.org/show_bug.cgi?id=102332
fdo#101705 https://bugs.freedesktop.org/show_bug.cgi?id=101705
fdo#103165 https://bugs.freedesktop.org/show_bug.cgi?id=103165
fi-bdw-5557u total:289 pass:268 dwarn:0 dfail:0 fail:0 skip:21 time:441s fi-bdw-gvtdvm total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:453s fi-blb-e6850 total:289 pass:223 dwarn:1 dfail:0 fail:0 skip:65 time:370s fi-bsw-n3050 total:289 pass:242 dwarn:1 dfail:0 fail:0 skip:46 time:520s fi-bwr-2160 total:289 pass:183 dwarn:0 dfail:0 fail:0 skip:106 time:261s fi-bxt-dsi total:289 pass:259 dwarn:0 dfail:0 fail:0 skip:30 time:499s fi-bxt-j4205 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:500s fi-byt-j1900 total:289 pass:253 dwarn:1 dfail:0 fail:0 skip:35 time:495s fi-byt-n2820 total:289 pass:249 dwarn:1 dfail:0 fail:0 skip:39 time:476s fi-cfl-s total:289 pass:253 dwarn:4 dfail:0 fail:0 skip:32 time:554s fi-cnl-y total:289 pass:193 dwarn:0 dfail:0 fail:0 skip:20 fi-elk-e7500 total:289 pass:229 dwarn:0 dfail:0 fail:0 skip:60 time:417s fi-gdg-551 total:289 pass:178 dwarn:1 dfail:0 fail:1 skip:109 time:251s fi-glk-1 total:289 pass:261 dwarn:0 dfail:0 fail:0 skip:28 time:582s fi-glk-dsi total:289 pass:258 dwarn:0 dfail:0 fail:1 skip:30 time:488s fi-hsw-4770 total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:425s fi-hsw-4770r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:425s fi-ilk-650 total:289 pass:228 dwarn:0 dfail:0 fail:0 skip:61 time:433s fi-ivb-3520m total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:484s fi-ivb-3770 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:459s fi-kbl-7500u total:289 pass:264 dwarn:1 dfail:0 fail:0 skip:24 time:490s fi-kbl-7560u total:289 pass:270 dwarn:0 dfail:0 fail:0 skip:19 time:573s fi-kbl-7567u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:483s fi-kbl-r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:585s fi-pnv-d510 total:289 pass:222 dwarn:1 dfail:0 fail:0 skip:66 time:550s fi-skl-6260u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:444s fi-skl-6600u total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:597s fi-skl-6700hq total:289 pass:263 dwarn:0 dfail:0 fail:0 skip:26 time:647s fi-skl-6700k total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:522s fi-skl-6770hq total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:502s fi-skl-gvtdvm total:289 pass:266 dwarn:0 dfail:0 fail:0 skip:23 time:457s fi-snb-2520m total:289 pass:250 dwarn:0 dfail:0 fail:0 skip:39 time:560s fi-snb-2600 total:289 pass:249 dwarn:0 dfail:0 fail:0 skip:40 time:413s
df3033b174059a59aa0c890f81de8af037abd11f drm-tip: 2017y-10m-26d-11h-03m-59s UTC integration manifest 57f4372b843b HAX Enable GuC Submission for CIbd453e53421a drm/i915/guc: Preemption! With GuCc230cfaaeec4 drm/i915: Rename helpers used for unwinding, use macro for can_preempt be74df89d1cb drm/i915/guc: Keep request->priority for its lifetimec65600f6018d drm/i915: Add information needed to track engine preempt state 05053f4249e6 drm/i915: Extract "emit write" part of emit breadcrumb functions 44056d81e278 drm/i915/guc: Split guc_wq_item_appende6230d13d5b1 drm/i915/guc: Add a second client, to be used for preemption 46cdb7035ad8 drm/i915/guc: Add preemption action to GuC firmware interface 05352d789d91 drm/i915/guc: Allocate separate shared data object for GuC communication 79c012264126 drm/i915/guc: Extract GuC stage desc pool creation into a helper a535408d82c7 drm/i915/guc: Do not use 0 for GuC doorbell cookie
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6205/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✗ Fi.CI.BAT: warning for Preemption with GuC, fourth try (rev6)
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try (rev6)
URL : https://patchwork.freedesktop.org/series/32654/
State : warning
== Summary ==
Series 32654v6 Preemption with GuC, fourth try
https://patchwork.freedesktop.org/api/1.0/series/32654/revisions/6/mbox/
Test kms_pipe_crc_basic:
Subgroup read-crc-pipe-b-frame-sequence:
skip -> PASS (fi-hsw-4770r) fdo#102332
Subgroup suspend-read-crc-pipe-b:
pass -> DMESG-WARN (fi-byt-j1900) fdo#101705
Test drv_module_reload:
Subgroup basic-reload:
pass -> DMESG-WARN (fi-cnl-y)
Subgroup basic-no-display:
dmesg-warn -> INCOMPLETE (fi-cfl-s) fdo#103206
pass -> DMESG-WARN (fi-cnl-y)
Subgroup basic-reload-inject:
pass -> DMESG-WARN (fi-cnl-y)fdo#102332 https://bugs.freedesktop.org/show_bug.cgi?id=102332
fdo#101705 https://bugs.freedesktop.org/show_bug.cgi?id=101705
fdo#103206 https://bugs.freedesktop.org/show_bug.cgi?id=103206
fi-bdw-5557u total:289 pass:268 dwarn:0 dfail:0 fail:0 skip:21 time:445s fi-bdw-gvtdvm total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:455s fi-blb-e6850 total:289 pass:223 dwarn:1 dfail:0 fail:0 skip:65 time:370s fi-bsw-n3050 total:289 pass:243 dwarn:0 dfail:0 fail:0 skip:46 time:525s fi-bwr-2160 total:289 pass:183 dwarn:0 dfail:0 fail:0 skip:106 time:263s fi-bxt-dsi total:289 pass:259 dwarn:0 dfail:0 fail:0 skip:30 time:504s fi-bxt-j4205 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:498s fi-byt-j1900 total:289 pass:253 dwarn:1 dfail:0 fail:0 skip:35 time:490s fi-byt-n2820 total:289 pass:249 dwarn:1 dfail:0 fail:0 skip:39 time:482s fi-cfl-s total:287 pass:253 dwarn:2 dfail:0 fail:0 skip:31 fi-cnl-y total:289 pass:259 dwarn:3 dfail:0 fail:0 skip:27 time:606s fi-elk-e7500 total:289 pass:229 dwarn:0 dfail:0 fail:0 skip:60 time:418s fi-gdg-551 total:289 pass:178 dwarn:1 dfail:0 fail:1 skip:109 time:250s fi-glk-1 total:289 pass:261 dwarn:0 dfail:0 fail:0 skip:28 time:575s fi-glk-dsi total:289 pass:258 dwarn:0 dfail:0 fail:1 skip:30 time:483s fi-hsw-4770 total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:424s fi-hsw-4770r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:428s fi-ilk-650 total:289 pass:228 dwarn:0 dfail:0 fail:0 skip:61 time:434s fi-ivb-3520m total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:493s fi-ivb-3770 total:289 pass:260 dwarn:0 dfail:0 fail:0 skip:29 time:455s fi-kbl-7500u total:289 pass:264 dwarn:1 dfail:0 fail:0 skip:24 time:491s fi-kbl-7560u total:289 pass:270 dwarn:0 dfail:0 fail:0 skip:19 time:572s fi-kbl-7567u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:485s fi-kbl-r total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:586s fi-pnv-d510 total:289 pass:222 dwarn:1 dfail:0 fail:0 skip:66 time:541s fi-skl-6260u total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:449s fi-skl-6600u total:289 pass:262 dwarn:0 dfail:0 fail:0 skip:27 time:598s fi-skl-6700hq total:289 pass:263 dwarn:0 dfail:0 fail:0 skip:26 time:648s fi-skl-6700k total:289 pass:265 dwarn:0 dfail:0 fail:0 skip:24 time:523s fi-skl-6770hq total:289 pass:269 dwarn:0 dfail:0 fail:0 skip:20 time:496s fi-skl-gvtdvm total:289 pass:266 dwarn:0 dfail:0 fail:0 skip:23 time:461s fi-snb-2520m total:289 pass:250 dwarn:0 dfail:0 fail:0 skip:39 time:561s fi-snb-2600 total:289 pass:249 dwarn:0 dfail:0 fail:0 skip:40 time:420s
df3033b174059a59aa0c890f81de8af037abd11f drm-tip: 2017y-10m-26d-11h-03m-59s UTC integration manifest 7172b849c490 HAX Enable GuC Submission for CI996b8a088c11 drm/i915/guc: Preemption! With GuC7bebd9377fe5 drm/i915: Rename helpers used for unwinding, use macro for can_preempt c358e5f30565 drm/i915/guc: Keep request->priority for its lifetime815b33b8e308 drm/i915: Add information needed to track engine preempt state 364832f82afd drm/i915: Extract "emit write" part of emit breadcrumb functions 274c59afc863 drm/i915/guc: Split guc_wq_item_appendea7d6bdcc394 drm/i915/guc: Add a second client, to be used for preemption 2de790e9c225 drm/i915/guc: Add preemption action to GuC firmware interface 881c8236e90b drm/i915/guc: Allocate separate shared data object for GuC communication db5247f8c3a1 drm/i915/guc: Extract GuC stage desc pool creation into a helper ae0602d37052 drm/i915/guc: Do not use 0 for GuC doorbell cookie
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6207/
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.IGT: success for Preemption with GuC, fourth try (rev5)
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try (rev5)
URL : https://patchwork.freedesktop.org/series/32654/
State : success
== Summary ==
Test kms_busy:
Subgroup extended-modeset-hang-oldfb-with-reset-render-A:
pass -> DMESG-WARN (shard-hsw) fdo#102249 +1
Subgroup extended-modeset-hang-newfb-with-reset-render-B:
pass -> DMESG-WARN (shard-hsw) fdo#103038
Test kms_plane:
Subgroup plane-panning-bottom-right-suspend-pipe-A-planes:
skip -> PASS (shard-hsw)
Test kms_flip:
Subgroup dpms-vs-vblank-race-interruptible:
pass -> FAIL (shard-hsw) fdo#103060
Test kms_setmode:
Subgroup basic:
pass -> FAIL (shard-hsw) fdo#99912fdo#102249 https://bugs.freedesktop.org/show_bug.cgi?id=102249
fdo#103038 https://bugs.freedesktop.org/show_bug.cgi?id=103038
fdo#103060 https://bugs.freedesktop.org/show_bug.cgi?id=103060
fdo#99912 https://bugs.freedesktop.org/show_bug.cgi?id=99912
shard-hsw total:2539 pass:1430 dwarn:2 dfail:0 fail:10 skip:1097 time:9281s
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6205/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Patchwork
CI✓ Fi.CI.IGT: success for Preemption with GuC, fourth try (rev6)
In reply to Michał Winiarski== Series Details ==
== Series Details ==
Series: Preemption with GuC, fourth try (rev6)
URL : https://patchwork.freedesktop.org/series/32654/
State : success
== Summary ==
Test kms_busy:
Subgroup extended-modeset-hang-newfb-with-reset-render-A:
pass -> DMESG-WARN (shard-hsw) fdo#102249 +1
Test kms_plane:
Subgroup plane-panning-bottom-right-suspend-pipe-A-planes:
skip -> PASS (shard-hsw)fdo#102249 https://bugs.freedesktop.org/show_bug.cgi?id=102249
shard-hsw total:2539 pass:1433 dwarn:1 dfail:0 fail:8 skip:1097 time:9240s
== Logs ==
For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_6207/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoting Michał Winiarski (2017-10-25 21:00:08)
Quoted 3 lines
No major changes from previous iteration.
Dropped the workaround for missing interrupt (which turned out to be self-inflicted, now properly fixed by Chris), and applied the review comments.
And pushed. Thanks everyone for the reviews and suggestions; looks like we have a solid foundation on which to build.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx