[Intel-gfx] [PATCH v5 3/5] drm/i915/guc : Decouple logs and ADS from submission
Michal Wajdeczko
michal.wajdeczko at intel.com
Wed Oct 4 13:50:05 UTC 2017
On Wed, 04 Oct 2017 00:56:38 +0200, Sujaritha Sundaresan
<sujaritha.sundaresan at intel.com> wrote:
> The Additional Data Struct (ADS) contains objects that are required by
> guc post FW load and are not necessarily submission-only (although that's
> our current only use-case). If in the future we load GuC with submission
> disabled to use some other GuC feature we might still end up requiring
> something inside the ADS, so it makes more sense for them to be always
> created if GuC is loaded.
>
> Similarly, we still want to access GuC logs even if GuC submission is
> disable to debug issues with GuC loading or with wathever we're using
> GuC for.
>
> To make a concrete example, the pages used by GuC to save state during
> suspend are allocated as part of the ADS.
>
> v3: Group initialization of GuC objects
>
> v2: Decoupling ADS together with logs (Daniele)
>
> v3: Re-factoring code as per review (Michal)
>
> v4: Rebase
>
> v5: Separating group object initialization into next patch
> Clarifying commit message
>
> Cc: Anusha Srivatsa <anusha.srivatsa at intel.com>
> Cc: Daniele Ceraolo Spurio <daniele.ceraolospurio at intel.com>
> Cc: Michal Wajdeczko <michal.wajdeczko at intel.com>
> Cc: Oscar Mateo <oscar.mateo at intel.com>
> Cc: Sagar Arun Kamble <sagar.a.kamble at intel.com>
> Signed-off-by: Sujaritha Sundaresan <sujaritha.sundaresan at intel.com>
> ---
> drivers/gpu/drm/i915/i915_guc_submission.c | 114
> +---------------------------
> drivers/gpu/drm/i915/intel_guc_log.c | 6 +-
> drivers/gpu/drm/i915/intel_uc.c | 115
> ++++++++++++++++++++++++++++-
> drivers/gpu/drm/i915/intel_uc.h | 1 +
> 4 files changed, 121 insertions(+), 115 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c
> b/drivers/gpu/drm/i915/i915_guc_submission.c
> index 04f1281..c456c55 100644
> --- a/drivers/gpu/drm/i915/i915_guc_submission.c
> +++ b/drivers/gpu/drm/i915/i915_guc_submission.c
> @@ -71,13 +71,6 @@
> * ELSP context descriptor dword into Work Item.
> * See guc_wq_item_append()
> *
> - * ADS:
> - * The Additional Data Struct (ADS) has pointers for different buffers
> used by
> - * the GuC. One single gem object contains the ADS struct itself
> (guc_ads), the
> - * scheduling policies (guc_policies), a structure describing a
> collection of
> - * register sets (guc_mmio_reg_state) and some extra pages for the GuC
> to save
> - * its internal state for sleep.
> - *
> */
> static inline bool is_high_priority(struct i915_guc_client* client)
> @@ -904,7 +897,7 @@ static void guc_policy_init(struct guc_policy
> *policy)
> policy->policy_flags = 0;
> }
> -static void guc_policies_init(struct guc_policies *policies)
> +void i915_guc_policies_init(struct guc_policies *policies)
> {
> struct guc_policy *policy;
> u32 p, i;
> @@ -924,88 +917,6 @@ static void guc_policies_init(struct guc_policies
> *policies)
> }
> /*
> - * The first 80 dwords of the register state context, containing the
> - * execlists and ppgtt registers.
> - */
> -#define LR_HW_CONTEXT_SIZE (80 * sizeof(u32))
> -
> -static int guc_ads_create(struct intel_guc *guc)
> -{
> - struct drm_i915_private *dev_priv = guc_to_i915(guc);
> - struct i915_vma *vma;
> - struct page *page;
> - /* The ads obj includes the struct itself and buffers passed to GuC */
> - struct {
> - struct guc_ads ads;
> - struct guc_policies policies;
> - struct guc_mmio_reg_state reg_state;
> - u8 reg_state_buffer[GUC_S3_SAVE_SPACE_PAGES * PAGE_SIZE];
> - } __packed *blob;
> - struct intel_engine_cs *engine;
> - enum intel_engine_id id;
> - const u32 skipped_offset = LRC_HEADER_PAGES * PAGE_SIZE;
> - const u32 skipped_size = LRC_PPHWSP_SZ * PAGE_SIZE +
> LR_HW_CONTEXT_SIZE;
> - u32 base;
> -
> - GEM_BUG_ON(guc->ads_vma);
> -
> - vma = intel_guc_allocate_vma(guc, PAGE_ALIGN(sizeof(*blob)));
> - if (IS_ERR(vma))
> - return PTR_ERR(vma);
> -
> - guc->ads_vma = vma;
> -
> - page = i915_vma_first_page(vma);
> - blob = kmap(page);
> -
> - /* GuC scheduling policies */
> - guc_policies_init(&blob->policies);
> -
> - /* MMIO reg state */
> - for_each_engine(engine, dev_priv, id) {
> - blob->reg_state.white_list[engine->guc_id].mmio_start =
> - engine->mmio_base + GUC_MMIO_WHITE_LIST_START;
> -
> - /* Nothing to be saved or restored for now. */
> - blob->reg_state.white_list[engine->guc_id].count = 0;
> - }
> -
> - /*
> - * The GuC requires a "Golden Context" when it reinitialises
> - * engines after a reset. Here we use the Render ring default
> - * context, which must already exist and be pinned in the GGTT,
> - * so its address won't change after we've told the GuC where
> - * to find it. Note that we have to skip our header (1 page),
> - * because our GuC shared data is there.
> - */
> - blob->ads.golden_context_lrca =
> - guc_ggtt_offset(dev_priv->kernel_context->engine[RCS].state) +
> skipped_offset;
> -
> - /*
> - * The GuC expects us to exclude the portion of the context image that
> - * it skips from the size it is to read. It starts reading from after
> - * the execlist context (so skipping the first page [PPHWSP] and 80
> - * dwords). Weird guc is weird.
> - */
> - for_each_engine(engine, dev_priv, id)
> - blob->ads.eng_state_size[engine->guc_id] = engine->context_size -
> skipped_size;
> -
> - base = guc_ggtt_offset(vma);
> - blob->ads.scheduler_policies = base + ptr_offset(blob, policies);
> - blob->ads.reg_state_buffer = base + ptr_offset(blob, reg_state_buffer);
> - blob->ads.reg_state_addr = base + ptr_offset(blob, reg_state);
> -
> - kunmap(page);
> -
> - return 0;
> -}
> -
> -static void guc_ads_destroy(struct intel_guc *guc)
> -{
> - i915_vma_unpin_and_release(&guc->ads_vma);
> -}
> -
> -/*
> * Set up the memory resources to be shared with the GuC (via the GGTT)
> * at firmware loading time.
> */
> @@ -1014,7 +925,6 @@ 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;
> @@ -1029,31 +939,15 @@ int i915_guc_submission_init(struct
> drm_i915_private *dev_priv)
> 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;
> + i915_vma_unpin_and_release(&guc->stage_desc_pool);
> + return PTR_ERR(vaddr);
I would stay with 'goto err' pattern
> }
> guc->stage_desc_pool_vaddr = vaddr;
> - ret = intel_guc_log_create(guc);
> - if (ret < 0)
> - goto err_vaddr;
> -
> - 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);
> - return ret;
> }
> void i915_guc_submission_fini(struct drm_i915_private *dev_priv)
> @@ -1061,8 +955,6 @@ 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);
> }
> diff --git a/drivers/gpu/drm/i915/intel_guc_log.c
> b/drivers/gpu/drm/i915/intel_guc_log.c
> index 6571d96..1c054c12 100644
> --- a/drivers/gpu/drm/i915/intel_guc_log.c
> +++ b/drivers/gpu/drm/i915/intel_guc_log.c
> @@ -502,7 +502,7 @@ static void guc_flush_logs(struct intel_guc *guc)
> {
> struct drm_i915_private *dev_priv = guc_to_i915(guc);
> - if (!i915_modparams.enable_guc_submission ||
> + if (!NEEDS_GUC_LOADING(dev_priv) ||
> (i915_modparams.guc_log_level < 0))
> return;
> @@ -642,7 +642,7 @@ int i915_guc_log_control(struct drm_i915_private
> *dev_priv, u64 control_val)
> void i915_guc_log_register(struct drm_i915_private *dev_priv)
> {
> - if (!i915_modparams.enable_guc_submission ||
> + if (!NEEDS_GUC_LOADING(dev_priv) ||
> (i915_modparams.guc_log_level < 0))
> return;
> @@ -653,7 +653,7 @@ void i915_guc_log_register(struct drm_i915_private
> *dev_priv)
> void i915_guc_log_unregister(struct drm_i915_private *dev_priv)
> {
> - if (!i915_modparams.enable_guc_submission)
> + if (!NEEDS_GUC_LOADING(dev_priv))
> return;
> mutex_lock(&dev_priv->drm.struct_mutex);
> diff --git a/drivers/gpu/drm/i915/intel_uc.c
> b/drivers/gpu/drm/i915/intel_uc.c
> index 4290b35..732f188 100644
> --- a/drivers/gpu/drm/i915/intel_uc.c
> +++ b/drivers/gpu/drm/i915/intel_uc.c
> @@ -328,6 +328,112 @@ static void guc_disable_communication(struct
> intel_guc *guc)
> guc->send = intel_guc_send_nop;
> }
> +/*
> + * The first 80 dwords of the register state context, containing the
> + * execlists and ppgtt registers.
> + */
> + #define LR_HW_CONTEXT_SIZE (80 * sizeof(u32))
> +
> +static int guc_ads_create(struct intel_guc *guc)
Hmm, I'm wondering if maybe better place for this function will
be intel_guc_loader.c or even intel_guc_ads.c ?
> + {
> + struct drm_i915_private *dev_priv = guc_to_i915(guc);
> + struct i915_vma *vma;
> + struct page *page;
> + /* The ads obj includes the struct itself and buffers passed to GuC */
> + struct {
> + struct guc_ads ads;
> + struct guc_policies policies;
> + struct guc_mmio_reg_state reg_state;
> + u8 reg_state_buffer[GUC_S3_SAVE_SPACE_PAGES * PAGE_SIZE];
> + } __packed *blob;
> + struct intel_engine_cs *engine;
> + enum intel_engine_id id;
> + const u32 skipped_offset = LRC_HEADER_PAGES * PAGE_SIZE;
> + const u32 skipped_size = LRC_PPHWSP_SZ * PAGE_SIZE +
> LR_HW_CONTEXT_SIZE;
> + u32 base;
> +
> + GEM_BUG_ON(guc->ads_vma);
> +
> + vma = intel_guc_allocate_vma(guc, PAGE_ALIGN(sizeof(*blob)));
> + if (IS_ERR(vma))
> + return PTR_ERR(vma);
> +
> + guc->ads_vma = vma;
> +
> + page = i915_vma_first_page(vma);
> + blob = kmap(page);
> +
> + /* GuC scheduling policies */
> + i915_guc_policies_init(&blob->policies);
btw, do we need all ads (like these policies) when guc submission is off?
> +
> + /* MMIO reg state */
> + for_each_engine(engine, dev_priv, id) {
> + blob->reg_state.white_list[engine->guc_id].mmio_start =
> + engine->mmio_base + GUC_MMIO_WHITE_LIST_START;
> +
> + /* Nothing to be saved or restored for now. */
> + blob->reg_state.white_list[engine->guc_id].count = 0;
> + }
> +
> + /*
> + * The GuC requires a "Golden Context" when it reinitialises
> + * engines after a reset. Here we use the Render ring default
> + * context, which must already exist and be pinned in the GGTT,
> + * so its address won't change after we've told the GuC where
> + * to find it. Note that we have to skip our header (1 page),
> + * because our GuC shared data is there.
> + */
> + blob->ads.golden_context_lrca =
> + guc_ggtt_offset(dev_priv->kernel_context->engine[RCS].state) +
> skipped_offset;
> +
> + /*
> + * The GuC expects us to exclude the portion of the context image that
> + * it skips from the size it is to read. It starts reading from after
> + * the execlist context (so skipping the first page [PPHWSP] and 80
> + * dwords). Weird guc is weird.
> + */
> + for_each_engine(engine, dev_priv, id)
> + blob->ads.eng_state_size[engine->guc_id] = engine->context_size -
> skipped_size;
> +
> + base = guc_ggtt_offset(vma);
> + blob->ads.scheduler_policies = base + ptr_offset(blob, policies);
> + blob->ads.reg_state_buffer = base + ptr_offset(blob,
> reg_state_buffer);
> + blob->ads.reg_state_addr = base + ptr_offset(blob, reg_state);
> +
> + kunmap(page);
> +
> + return 0;
> +}
> +
> +static void guc_ads_destroy(struct intel_guc *guc)
> +{
> + i915_vma_unpin_and_release(&guc->ads_vma);
> +}
> +
> +static int guc_shared_objects_init(struct intel_guc *guc)
> +{
> + int ret;
> +
> + if (guc->ads_vma)
> + return 0;
> +
> + ret = intel_guc_log_create(guc);
> + if (ret < 0)
> + return ret;
> +
> + ret = guc_ads_create(guc);
> + if (ret < 0)
> + intel_guc_log_destroy(guc);
Try to always use this pattern:
if (ret)
goto err_x;
err_x:
x_cleanup();
return ret;
> +
> + return ret;
> +}
> +
> +static void guc_shared_objects_fini(struct intel_guc *guc)
> +{
> + guc_ads_destroy(guc);
> + intel_guc_log_destroy(guc);
> +}
> +
> int intel_uc_init_hw(struct drm_i915_private *dev_priv)
> {
> struct intel_guc *guc = &dev_priv->guc;
> @@ -342,6 +448,10 @@ int intel_uc_init_hw(struct drm_i915_private
> *dev_priv)
> /* We need to notify the guc whenever we change the GGTT */
> i915_ggtt_enable_guc(dev_priv);
> + ret = guc_shared_objects_init(guc);
> + if (ret)
> + goto err_guc;
> +
> if (i915_modparams.enable_guc_submission) {
> /*
> * This is stuff we need to have available at fw load time
> @@ -349,7 +459,7 @@ int intel_uc_init_hw(struct drm_i915_private
> *dev_priv)
> */
> ret = i915_guc_submission_init(dev_priv);
> if (ret)
> - goto err_guc;
> + goto err_shared;
> }
> /* init WOPCM */
> @@ -419,6 +529,8 @@ int intel_uc_init_hw(struct drm_i915_private
> *dev_priv)
> err_submission:
> if (i915_modparams.enable_guc_submission)
> i915_guc_submission_fini(dev_priv);
> +err_shared:
> + guc_shared_objects_fini(guc);
> err_guc:
> i915_ggtt_disable_guc(dev_priv);
> @@ -458,6 +570,7 @@ void intel_uc_fini_hw(struct drm_i915_private
> *dev_priv)
> i915_guc_submission_fini(dev_priv);
> }
> + guc_shared_objects_fini(&dev_priv->guc);
> i915_ggtt_disable_guc(dev_priv);
> }
> diff --git a/drivers/gpu/drm/i915/intel_uc.h
> b/drivers/gpu/drm/i915/intel_uc.h
> index 58d787e..0d12ff4 100644
> --- a/drivers/gpu/drm/i915/intel_uc.h
> +++ b/drivers/gpu/drm/i915/intel_uc.h
> @@ -235,6 +235,7 @@ static inline void intel_guc_notify(struct intel_guc
> *guc)
> void i915_guc_submission_disable(struct drm_i915_private *dev_priv);
> void i915_guc_submission_fini(struct drm_i915_private *dev_priv);
> struct i915_vma *intel_guc_allocate_vma(struct intel_guc *guc, u32
> size);
> +void i915_guc_policies_init(struct guc_policies *policies);
> /* intel_guc_log.c */
> int intel_guc_log_create(struct intel_guc *guc);
More information about the Intel-gfx
mailing list