From: John Harrison <john.c.harrison@intel.com>
To: Matthew Brost <matthew.brost@intel.com>,
<intel-gfx@lists.freedesktop.org>,
<dri-devel@lists.freedesktop.org>
Cc: <daniel.vetter@ffwll.ch>, <tony.ye@intel.com>, <zhengguo.xu@intel.com>
Subject: Re: [Intel-gfx] [PATCH 15/27] drm/i915/guc: Implement multi-lrc submission
Date: Mon, 20 Sep 2021 14:48:52 -0700 [thread overview]
Message-ID: <c840bf5b-c5e9-872d-e04f-2dfe1852e555@intel.com> (raw)
In-Reply-To: <20210820224446.30620-16-matthew.brost@intel.com>
On 8/20/2021 15:44, Matthew Brost wrote:
> Implement multi-lrc submission via a single workqueue entry and single
> H2G. The workqueue entry contains an updated tail value for each
> request, of all the contexts in the multi-lrc submission, and updates
> these values simultaneously. As such, the tasklet and bypass path have
> been updated to coalesce requests into a single submission.
>
> Signed-off-by: Matthew Brost <matthew.brost@intel.com>
> ---
> drivers/gpu/drm/i915/gt/uc/intel_guc.c | 21 ++
> drivers/gpu/drm/i915/gt/uc/intel_guc.h | 8 +
> drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c | 24 +-
> drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h | 6 +-
> .../gpu/drm/i915/gt/uc/intel_guc_submission.c | 312 +++++++++++++++---
> drivers/gpu/drm/i915/i915_request.h | 8 +
> 6 files changed, 317 insertions(+), 62 deletions(-)
>
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.c b/drivers/gpu/drm/i915/gt/uc/intel_guc.c
> index fbfcae727d7f..879aef662b2e 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.c
> @@ -748,3 +748,24 @@ void intel_guc_load_status(struct intel_guc *guc, struct drm_printer *p)
> }
> }
> }
> +
> +void intel_guc_write_barrier(struct intel_guc *guc)
> +{
> + struct intel_gt *gt = guc_to_gt(guc);
> +
> + if (i915_gem_object_is_lmem(guc->ct.vma->obj)) {
> + GEM_BUG_ON(guc->send_regs.fw_domains);
Granted, this patch is just moving code from one file to another not
changing it. However, I think it would be worth adding a blank line in
here. Otherwise the 'this register' comment below can be confusingly
read as referring to the send_regs.fw_domain entry above.
And maybe add a comment why it is a bug for the send_regs value to be
set? I'm not seeing any obvious connection between it and the reset of
this code.
> + /*
> + * This register is used by the i915 and GuC for MMIO based
> + * communication. Once we are in this code CTBs are the only
> + * method the i915 uses to communicate with the GuC so it is
> + * safe to write to this register (a value of 0 is NOP for MMIO
> + * communication). If we ever start mixing CTBs and MMIOs a new
> + * register will have to be chosen.
> + */
> + intel_uncore_write_fw(gt->uncore, GEN11_SOFT_SCRATCH(0), 0);
> + } else {
> + /* wmb() sufficient for a barrier if in smem */
> + wmb();
> + }
> +}
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc.h b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> index 3f95b1b4f15c..0ead2406d03c 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc.h
> @@ -37,6 +37,12 @@ struct intel_guc {
> /* Global engine used to submit requests to GuC */
> struct i915_sched_engine *sched_engine;
> struct i915_request *stalled_request;
> + enum {
> + STALL_NONE,
> + STALL_REGISTER_CONTEXT,
> + STALL_MOVE_LRC_TAIL,
> + STALL_ADD_REQUEST,
> + } submission_stall_reason;
>
> /* intel_guc_recv interrupt related state */
> spinlock_t irq_lock;
> @@ -332,4 +338,6 @@ void intel_guc_submission_cancel_requests(struct intel_guc *guc);
>
> void intel_guc_load_status(struct intel_guc *guc, struct drm_printer *p);
>
> +void intel_guc_write_barrier(struct intel_guc *guc);
> +
> #endif
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> index 20c710a74498..10d1878d2826 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_ct.c
> @@ -377,28 +377,6 @@ static u32 ct_get_next_fence(struct intel_guc_ct *ct)
> return ++ct->requests.last_fence;
> }
>
> -static void write_barrier(struct intel_guc_ct *ct)
> -{
> - struct intel_guc *guc = ct_to_guc(ct);
> - struct intel_gt *gt = guc_to_gt(guc);
> -
> - if (i915_gem_object_is_lmem(guc->ct.vma->obj)) {
> - GEM_BUG_ON(guc->send_regs.fw_domains);
> - /*
> - * This register is used by the i915 and GuC for MMIO based
> - * communication. Once we are in this code CTBs are the only
> - * method the i915 uses to communicate with the GuC so it is
> - * safe to write to this register (a value of 0 is NOP for MMIO
> - * communication). If we ever start mixing CTBs and MMIOs a new
> - * register will have to be chosen.
> - */
> - intel_uncore_write_fw(gt->uncore, GEN11_SOFT_SCRATCH(0), 0);
> - } else {
> - /* wmb() sufficient for a barrier if in smem */
> - wmb();
> - }
> -}
> -
> static int ct_write(struct intel_guc_ct *ct,
> const u32 *action,
> u32 len /* in dwords */,
> @@ -468,7 +446,7 @@ static int ct_write(struct intel_guc_ct *ct,
> * make sure H2G buffer update and LRC tail update (if this triggering a
> * submission) are visible before updating the descriptor tail
> */
> - write_barrier(ct);
> + intel_guc_write_barrier(ct_to_guc(ct));
>
> /* update local copies */
> ctb->tail = tail;
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h b/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h
> index 0e600a3b8f1e..6cd26dc060d1 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_fwif.h
> @@ -65,12 +65,14 @@
> #define WQ_TYPE_PSEUDO (0x2 << WQ_TYPE_SHIFT)
> #define WQ_TYPE_INORDER (0x3 << WQ_TYPE_SHIFT)
> #define WQ_TYPE_NOOP (0x4 << WQ_TYPE_SHIFT)
> -#define WQ_TARGET_SHIFT 10
> +#define WQ_TYPE_MULTI_LRC (0x5 << WQ_TYPE_SHIFT)
> +#define WQ_TARGET_SHIFT 8
> #define WQ_LEN_SHIFT 16
> #define WQ_NO_WCFLUSH_WAIT (1 << 27)
> #define WQ_PRESENT_WORKLOAD (1 << 28)
>
> -#define WQ_RING_TAIL_SHIFT 20
> +#define WQ_GUC_ID_SHIFT 0
> +#define WQ_RING_TAIL_SHIFT 18
Presumably all of these API changes are not actually new? They really
came in with the reset of the v40 re-write? It's just that this is the
first time we are using them and therefore need to finally update the
defines?
> #define WQ_RING_TAIL_MAX 0x7FF /* 2^11 QWords */
> #define WQ_RING_TAIL_MASK (WQ_RING_TAIL_MAX << WQ_RING_TAIL_SHIFT)
>
> diff --git a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> index e9dfd43d29a0..b107ad095248 100644
> --- a/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> +++ b/drivers/gpu/drm/i915/gt/uc/intel_guc_submission.c
> @@ -391,6 +391,29 @@ __get_process_desc(struct intel_context *ce)
> LRC_STATE_OFFSET) / sizeof(u32)));
> }
>
> +static u32 *get_wq_pointer(struct guc_process_desc *desc,
> + struct intel_context *ce,
> + u32 wqi_size)
> +{
> + /*
> + * Check for space in work queue. Caching a value of head pointer in
> + * intel_context structure in order reduce the number accesses to shared
> + * GPU memory which may be across a PCIe bus.
> + */
> +#define AVAILABLE_SPACE \
> + CIRC_SPACE(ce->guc_wqi_tail, ce->guc_wqi_head, GUC_WQ_SIZE)
> + if (wqi_size > AVAILABLE_SPACE) {
> + ce->guc_wqi_head = READ_ONCE(desc->head);
> +
> + if (wqi_size > AVAILABLE_SPACE)
> + return NULL;
> + }
> +#undef AVAILABLE_SPACE
> +
> + return ((u32 *)__get_process_desc(ce)) +
> + ((WQ_OFFSET + ce->guc_wqi_tail) / sizeof(u32));
> +}
> +
> static struct guc_lrc_desc *__get_lrc_desc(struct intel_guc *guc, u32 index)
> {
> struct guc_lrc_desc *base = guc->lrc_desc_pool_vaddr;
> @@ -547,10 +570,10 @@ int intel_guc_wait_for_idle(struct intel_guc *guc, long timeout)
>
> static int guc_lrc_desc_pin(struct intel_context *ce, bool loop);
>
> -static int guc_add_request(struct intel_guc *guc, struct i915_request *rq)
> +static int __guc_add_request(struct intel_guc *guc, struct i915_request *rq)
> {
> int err = 0;
> - struct intel_context *ce = rq->context;
> + struct intel_context *ce = request_to_scheduling_context(rq);
> u32 action[3];
> int len = 0;
> u32 g2h_len_dw = 0;
> @@ -571,26 +594,17 @@ static int guc_add_request(struct intel_guc *guc, struct i915_request *rq)
> GEM_BUG_ON(!atomic_read(&ce->guc_id.ref));
> GEM_BUG_ON(context_guc_id_invalid(ce));
>
> - /*
> - * Corner case where the GuC firmware was blown away and reloaded while
> - * this context was pinned.
> - */
> - if (unlikely(!lrc_desc_registered(guc, ce->guc_id.id))) {
> - err = guc_lrc_desc_pin(ce, false);
> - if (unlikely(err))
> - return err;
> - }
> -
> spin_lock(&ce->guc_state.lock);
>
> /*
> * The request / context will be run on the hardware when scheduling
> - * gets enabled in the unblock.
> + * gets enabled in the unblock. For multi-lrc we still submit the
> + * context to move the LRC tails.
> */
> - if (unlikely(context_blocked(ce)))
> + if (unlikely(context_blocked(ce) && !intel_context_is_parent(ce)))
> goto out;
>
> - enabled = context_enabled(ce);
> + enabled = context_enabled(ce) || context_blocked(ce);
Would be better to say '|| is_parent(ce)' rather than blocked? The
reason for reason for claiming enabled when not is because it's a
multi-LRC parent, right? Or can there be a parent that is neither
enabled nor blocked for which we don't want to do the processing? But
why would that make sense/be possible?
>
> if (!enabled) {
> action[len++] = INTEL_GUC_ACTION_SCHED_CONTEXT_MODE_SET;
> @@ -609,6 +623,18 @@ static int guc_add_request(struct intel_guc *guc, struct i915_request *rq)
> trace_intel_context_sched_enable(ce);
> atomic_inc(&guc->outstanding_submission_g2h);
> set_context_enabled(ce);
> +
> + /*
> + * Without multi-lrc KMD does the submission step (moving the
> + * lrc tail) so enabling scheduling is sufficient to submit the
> + * context. This isn't the case in multi-lrc submission as the
> + * GuC needs to move the tails, hence the need for another H2G
> + * to submit a multi-lrc context after enabling scheduling.
> + */
> + if (intel_context_is_parent(ce)) {
> + action[0] = INTEL_GUC_ACTION_SCHED_CONTEXT;
> + err = intel_guc_send_nb(guc, action, len - 1, 0);
> + }
> } else if (!enabled) {
> clr_context_pending_enable(ce);
> intel_context_put(ce);
> @@ -621,6 +647,18 @@ static int guc_add_request(struct intel_guc *guc, struct i915_request *rq)
> return err;
> }
>
> +static int guc_add_request(struct intel_guc *guc, struct i915_request *rq)
> +{
> + int ret = __guc_add_request(guc, rq);
> +
> + if (unlikely(ret == -EBUSY)) {
> + guc->stalled_request= rq;
> + guc->submission_stall_reason = STALL_ADD_REQUEST;
> + }
> +
> + return ret;
> +}
> +
> static void guc_set_lrc_tail(struct i915_request *rq)
> {
> rq->context->lrc_reg_state[CTX_RING_TAIL] =
> @@ -632,6 +670,127 @@ static int rq_prio(const struct i915_request *rq)
> return rq->sched.attr.priority;
> }
>
> +static bool is_multi_lrc_rq(struct i915_request *rq)
> +{
> + return intel_context_is_child(rq->context) ||
> + intel_context_is_parent(rq->context);
> +}
> +
> +static bool can_merge_rq(struct i915_request *rq,
> + struct i915_request *last)
> +{
> + return request_to_scheduling_context(rq) ==
> + request_to_scheduling_context(last);
> +}
> +
> +static u32 wq_space_until_wrap(struct intel_context *ce)
> +{
> + return (GUC_WQ_SIZE - ce->guc_wqi_tail);
> +}
> +
> +static void write_wqi(struct guc_process_desc *desc,
> + struct intel_context *ce,
> + u32 wqi_size)
> +{
> + /*
> + * Ensure WQE are visible before updating tail
WQE or WQI?
> + */
> + intel_guc_write_barrier(ce_to_guc(ce));
> +
> + ce->guc_wqi_tail = (ce->guc_wqi_tail + wqi_size) & (GUC_WQ_SIZE - 1);
> + WRITE_ONCE(desc->tail, ce->guc_wqi_tail);
> +}
> +
> +static int guc_wq_noop_append(struct intel_context *ce)
> +{
> + struct guc_process_desc *desc = __get_process_desc(ce);
> + u32 *wqi = get_wq_pointer(desc, ce, wq_space_until_wrap(ce));
> +
> + if (!wqi)
> + return -EBUSY;
> +
> + *wqi = WQ_TYPE_NOOP |
> + ((wq_space_until_wrap(ce) / sizeof(u32) - 1) << WQ_LEN_SHIFT);
This should have a BUG_ON check that the requested size fits within the
WQ_LEN field?
Indeed, would be better to use the FIELD macros as they do that kind of
thing for you.
> + ce->guc_wqi_tail = 0;
> +
> + return 0;
> +}
> +
> +static int __guc_wq_item_append(struct i915_request *rq)
> +{
> + struct intel_context *ce = request_to_scheduling_context(rq);
> + struct intel_context *child;
> + struct guc_process_desc *desc = __get_process_desc(ce);
> + unsigned int wqi_size = (ce->guc_number_children + 4) *
> + sizeof(u32);
> + u32 *wqi;
> + int ret;
> +
> + /* Ensure context is in correct state updating work queue */
> + GEM_BUG_ON(!atomic_read(&ce->guc_id.ref));
> + GEM_BUG_ON(context_guc_id_invalid(ce));
> + GEM_BUG_ON(context_wait_for_deregister_to_register(ce));
> + GEM_BUG_ON(!lrc_desc_registered(ce_to_guc(ce), ce->guc_id.id));
> +
> + /* Insert NOOP if this work queue item will wrap the tail pointer. */
> + if (wqi_size > wq_space_until_wrap(ce)) {
> + ret = guc_wq_noop_append(ce);
> + if (ret)
> + return ret;
> + }
> +
> + wqi = get_wq_pointer(desc, ce, wqi_size);
> + if (!wqi)
> + return -EBUSY;
> +
> + *wqi++ = WQ_TYPE_MULTI_LRC |
> + ((wqi_size / sizeof(u32) - 1) << WQ_LEN_SHIFT);
> + *wqi++ = ce->lrc.lrca;
> + *wqi++ = (ce->guc_id.id << WQ_GUC_ID_SHIFT) |
> + ((ce->ring->tail / sizeof(u64)) << WQ_RING_TAIL_SHIFT);
As above, would be better to use FIELD macros instead of manual shifting.
John.
> + *wqi++ = 0; /* fence_id */
> + for_each_child(ce, child)
> + *wqi++ = child->ring->tail / sizeof(u64);
> +
> + write_wqi(desc, ce, wqi_size);
> +
> + return 0;
> +}
> +
> +static int guc_wq_item_append(struct intel_guc *guc,
> + struct i915_request *rq)
> +{
> + struct intel_context *ce = request_to_scheduling_context(rq);
> + int ret = 0;
> +
> + if (likely(!intel_context_is_banned(ce))) {
> + ret = __guc_wq_item_append(rq);
> +
> + if (unlikely(ret == -EBUSY)) {
> + guc->stalled_request = rq;
> + guc->submission_stall_reason = STALL_MOVE_LRC_TAIL;
> + }
> + }
> +
> + return ret;
> +}
> +
> +static bool multi_lrc_submit(struct i915_request *rq)
> +{
> + struct intel_context *ce = request_to_scheduling_context(rq);
> +
> + intel_ring_set_tail(rq->ring, rq->tail);
> +
> + /*
> + * We expect the front end (execbuf IOCTL) to set this flag on the last
> + * request generated from a multi-BB submission. This indicates to the
> + * backend (GuC interface) that we should submit this context thus
> + * submitting all the requests generated in parallel.
> + */
> + return test_bit(I915_FENCE_FLAG_SUBMIT_PARALLEL, &rq->fence.flags) ||
> + intel_context_is_banned(ce);
> +}
> +
> static int guc_dequeue_one_context(struct intel_guc *guc)
> {
> struct i915_sched_engine * const sched_engine = guc->sched_engine;
> @@ -645,7 +804,17 @@ static int guc_dequeue_one_context(struct intel_guc *guc)
> if (guc->stalled_request) {
> submit = true;
> last = guc->stalled_request;
> - goto resubmit;
> +
> + switch (guc->submission_stall_reason) {
> + case STALL_REGISTER_CONTEXT:
> + goto register_context;
> + case STALL_MOVE_LRC_TAIL:
> + goto move_lrc_tail;
> + case STALL_ADD_REQUEST:
> + goto add_request;
> + default:
> + MISSING_CASE(guc->submission_stall_reason);
> + }
> }
>
> while ((rb = rb_first_cached(&sched_engine->queue))) {
> @@ -653,8 +822,8 @@ static int guc_dequeue_one_context(struct intel_guc *guc)
> struct i915_request *rq, *rn;
>
> priolist_for_each_request_consume(rq, rn, p) {
> - if (last && rq->context != last->context)
> - goto done;
> + if (last && !can_merge_rq(rq, last))
> + goto register_context;
>
> list_del_init(&rq->sched.link);
>
> @@ -662,33 +831,84 @@ static int guc_dequeue_one_context(struct intel_guc *guc)
>
> trace_i915_request_in(rq, 0);
> last = rq;
> - submit = true;
> +
> + if (is_multi_lrc_rq(rq)) {
> + /*
> + * We need to coalesce all multi-lrc requests in
> + * a relationship into a single H2G. We are
> + * guaranteed that all of these requests will be
> + * submitted sequentially.
> + */
> + if (multi_lrc_submit(rq)) {
> + submit = true;
> + goto register_context;
> + }
> + } else {
> + submit = true;
> + }
> }
>
> rb_erase_cached(&p->node, &sched_engine->queue);
> i915_priolist_free(p);
> }
> -done:
> +
> +register_context:
> if (submit) {
> - guc_set_lrc_tail(last);
> -resubmit:
> + struct intel_context *ce = request_to_scheduling_context(last);
> +
> + if (unlikely(!lrc_desc_registered(guc, ce->guc_id.id) &&
> + !intel_context_is_banned(ce))) {
> + ret = guc_lrc_desc_pin(ce, false);
> + if (unlikely(ret == -EPIPE)) {
> + goto deadlk;
> + } else if (ret == -EBUSY) {
> + guc->stalled_request = last;
> + guc->submission_stall_reason =
> + STALL_REGISTER_CONTEXT;
> + goto schedule_tasklet;
> + } else if (ret != 0) {
> + GEM_WARN_ON(ret); /* Unexpected */
> + goto deadlk;
> + }
> + }
> +
> +move_lrc_tail:
> + if (is_multi_lrc_rq(last)) {
> + ret = guc_wq_item_append(guc, last);
> + if (ret == -EBUSY) {
> + goto schedule_tasklet;
> + } else if (ret != 0) {
> + GEM_WARN_ON(ret); /* Unexpected */
> + goto deadlk;
> + }
> + } else {
> + guc_set_lrc_tail(last);
> + }
> +
> +add_request:
> ret = guc_add_request(guc, last);
> - if (unlikely(ret == -EPIPE))
> + if (unlikely(ret == -EPIPE)) {
> + goto deadlk;
> + } else if (ret == -EBUSY) {
> + goto schedule_tasklet;
> + } else if (ret != 0) {
> + GEM_WARN_ON(ret); /* Unexpected */
> goto deadlk;
> - else if (ret == -EBUSY) {
> - tasklet_schedule(&sched_engine->tasklet);
> - guc->stalled_request = last;
> - return false;
> }
> }
>
> guc->stalled_request = NULL;
> + guc->submission_stall_reason = STALL_NONE;
> return submit;
>
> deadlk:
> sched_engine->tasklet.callback = NULL;
> tasklet_disable_nosync(&sched_engine->tasklet);
> return false;
> +
> +schedule_tasklet:
> + tasklet_schedule(&sched_engine->tasklet);
> + return false;
> }
>
> static void guc_submission_tasklet(struct tasklet_struct *t)
> @@ -1227,10 +1447,16 @@ static int guc_bypass_tasklet_submit(struct intel_guc *guc,
>
> trace_i915_request_in(rq, 0);
>
> - guc_set_lrc_tail(rq);
> - ret = guc_add_request(guc, rq);
> - if (ret == -EBUSY)
> - guc->stalled_request = rq;
> + if (is_multi_lrc_rq(rq)) {
> + if (multi_lrc_submit(rq)) {
> + ret = guc_wq_item_append(guc, rq);
> + if (!ret)
> + ret = guc_add_request(guc, rq);
> + }
> + } else {
> + guc_set_lrc_tail(rq);
> + ret = guc_add_request(guc, rq);
> + }
>
> if (unlikely(ret == -EPIPE))
> disable_submission(guc);
> @@ -1238,6 +1464,16 @@ static int guc_bypass_tasklet_submit(struct intel_guc *guc,
> return ret;
> }
>
> +bool need_tasklet(struct intel_guc *guc, struct i915_request *rq)
> +{
> + struct i915_sched_engine *sched_engine = rq->engine->sched_engine;
> + struct intel_context *ce = request_to_scheduling_context(rq);
> +
> + return submission_disabled(guc) || guc->stalled_request ||
> + !i915_sched_engine_is_empty(sched_engine) ||
> + !lrc_desc_registered(guc, ce->guc_id.id);
> +}
> +
> static void guc_submit_request(struct i915_request *rq)
> {
> struct i915_sched_engine *sched_engine = rq->engine->sched_engine;
> @@ -1247,8 +1483,7 @@ static void guc_submit_request(struct i915_request *rq)
> /* Will be called from irq-context when using foreign fences. */
> spin_lock_irqsave(&sched_engine->lock, flags);
>
> - if (submission_disabled(guc) || guc->stalled_request ||
> - !i915_sched_engine_is_empty(sched_engine))
> + if (need_tasklet(guc, rq))
> queue_request(sched_engine, rq, rq_prio(rq));
> else if (guc_bypass_tasklet_submit(guc, rq) == -EBUSY)
> tasklet_hi_schedule(&sched_engine->tasklet);
> @@ -2241,9 +2476,10 @@ static bool new_guc_prio_higher(u8 old_guc_prio, u8 new_guc_prio)
>
> static void add_to_context(struct i915_request *rq)
> {
> - struct intel_context *ce = rq->context;
> + struct intel_context *ce = request_to_scheduling_context(rq);
> u8 new_guc_prio = map_i915_prio_to_guc_prio(rq_prio(rq));
>
> + GEM_BUG_ON(intel_context_is_child(ce));
> GEM_BUG_ON(rq->guc_prio == GUC_PRIO_FINI);
>
> spin_lock(&ce->guc_state.lock);
> @@ -2276,7 +2512,9 @@ static void guc_prio_fini(struct i915_request *rq, struct intel_context *ce)
>
> static void remove_from_context(struct i915_request *rq)
> {
> - struct intel_context *ce = rq->context;
> + struct intel_context *ce = request_to_scheduling_context(rq);
> +
> + GEM_BUG_ON(intel_context_is_child(ce));
>
> spin_lock_irq(&ce->guc_state.lock);
>
> @@ -2692,7 +2930,7 @@ static void guc_init_breadcrumbs(struct intel_engine_cs *engine)
> static void guc_bump_inflight_request_prio(struct i915_request *rq,
> int prio)
> {
> - struct intel_context *ce = rq->context;
> + struct intel_context *ce = request_to_scheduling_context(rq);
> u8 new_guc_prio = map_i915_prio_to_guc_prio(prio);
>
> /* Short circuit function */
> @@ -2715,7 +2953,7 @@ static void guc_bump_inflight_request_prio(struct i915_request *rq,
>
> static void guc_retire_inflight_request_prio(struct i915_request *rq)
> {
> - struct intel_context *ce = rq->context;
> + struct intel_context *ce = request_to_scheduling_context(rq);
>
> spin_lock(&ce->guc_state.lock);
> guc_prio_fini(rq, ce);
> diff --git a/drivers/gpu/drm/i915/i915_request.h b/drivers/gpu/drm/i915/i915_request.h
> index 177eaf55adff..8f0073e19079 100644
> --- a/drivers/gpu/drm/i915/i915_request.h
> +++ b/drivers/gpu/drm/i915/i915_request.h
> @@ -139,6 +139,14 @@ enum {
> * the GPU. Here we track such boost requests on a per-request basis.
> */
> I915_FENCE_FLAG_BOOST,
> +
> + /*
> + * I915_FENCE_FLAG_SUBMIT_PARALLEL - request with a context in a
> + * parent-child relationship (parallel submission, multi-lrc) should
> + * trigger a submission to the GuC rather than just moving the context
> + * tail.
> + */
> + I915_FENCE_FLAG_SUBMIT_PARALLEL,
> };
>
> /**
next prev parent reply other threads:[~2021-09-20 21:49 UTC|newest]
Thread overview: 106+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-08-20 22:44 [PATCH 00/27] Parallel submission aka multi-bb execbuf Matthew Brost
2021-08-20 22:44 ` [PATCH 01/27] drm/i915/guc: Squash Clean up GuC CI failures, simplify locking, and kernel DOC Matthew Brost
2021-08-20 22:44 ` [PATCH 02/27] drm/i915/guc: Allow flexible number of context ids Matthew Brost
2021-09-09 22:13 ` [Intel-gfx] " John Harrison
2021-09-10 0:14 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 03/27] drm/i915/guc: Connect the number of guc_ids to debugfs Matthew Brost
2021-09-09 22:16 ` [Intel-gfx] " John Harrison
2021-09-10 0:16 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 04/27] drm/i915/guc: Take GT PM ref when deregistering context Matthew Brost
2021-09-09 22:28 ` [Intel-gfx] " John Harrison
2021-09-10 0:21 ` Matthew Brost
2021-09-13 9:55 ` Tvrtko Ursulin
2021-09-13 17:12 ` Matthew Brost
2021-09-14 8:41 ` Tvrtko Ursulin
2021-08-20 22:44 ` [PATCH 05/27] drm/i915: Add GT PM unpark worker Matthew Brost
2021-09-09 22:36 ` [Intel-gfx] " John Harrison
2021-09-10 0:34 ` Matthew Brost
2021-09-10 8:36 ` Tvrtko Ursulin
2021-09-10 20:09 ` Matthew Brost
2021-09-13 10:33 ` Tvrtko Ursulin
2021-09-13 17:20 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 06/27] drm/i915/guc: Take engine PM when a context is pinned with GuC submission Matthew Brost
2021-09-09 22:46 ` [Intel-gfx] " John Harrison
2021-09-10 0:41 ` Matthew Brost
2021-09-13 22:26 ` John Harrison
2021-09-14 1:12 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 07/27] drm/i915/guc: Don't call switch_to_kernel_context " Matthew Brost
2021-09-09 22:51 ` [Intel-gfx] " John Harrison
2021-09-13 16:54 ` Matthew Brost
2021-09-13 22:38 ` John Harrison
2021-09-14 5:02 ` Matthew Brost
2021-09-13 16:55 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 08/27] drm/i915: Add logical engine mapping Matthew Brost
2021-09-10 11:12 ` [Intel-gfx] " Tvrtko Ursulin
2021-09-10 19:49 ` Matthew Brost
2021-09-13 9:24 ` Tvrtko Ursulin
2021-09-13 16:50 ` Matthew Brost
2021-09-14 8:34 ` Tvrtko Ursulin
2021-09-14 18:04 ` Matthew Brost
2021-09-15 8:24 ` Tvrtko Ursulin
2021-09-15 16:58 ` Matthew Brost
2021-09-16 8:31 ` Tvrtko Ursulin
2021-08-20 22:44 ` [PATCH 09/27] drm/i915: Expose logical engine instance to user Matthew Brost
2021-09-13 23:06 ` [Intel-gfx] " John Harrison
2021-09-14 1:08 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 10/27] drm/i915/guc: Introduce context parent-child relationship Matthew Brost
2021-09-13 23:19 ` [Intel-gfx] " John Harrison
2021-09-14 1:18 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 11/27] drm/i915/guc: Implement parallel context pin / unpin functions Matthew Brost
2021-08-20 22:44 ` [PATCH 12/27] drm/i915/guc: Add multi-lrc context registration Matthew Brost
2021-09-15 19:21 ` [Intel-gfx] " John Harrison
2021-09-15 19:31 ` Matthew Brost
2021-09-15 20:23 ` John Harrison
2021-09-15 20:33 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 13/27] drm/i915/guc: Ensure GuC schedule operations do not operate on child contexts Matthew Brost
2021-09-15 19:24 ` [Intel-gfx] " John Harrison
2021-09-15 19:34 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 14/27] drm/i915/guc: Assign contexts in parent-child relationship consecutive guc_ids Matthew Brost
2021-09-15 20:04 ` [Intel-gfx] " John Harrison
2021-09-15 20:55 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 15/27] drm/i915/guc: Implement multi-lrc submission Matthew Brost
2021-08-21 14:04 ` kernel test robot
2021-08-22 2:18 ` kernel test robot
2021-09-20 21:48 ` John Harrison [this message]
2021-09-22 16:25 ` [Intel-gfx] " Matthew Brost
2021-09-22 20:15 ` John Harrison
2021-09-23 2:44 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 16/27] drm/i915/guc: Insert submit fences between requests in parent-child relationship Matthew Brost
2021-09-20 21:57 ` [Intel-gfx] " John Harrison
2021-08-20 22:44 ` [PATCH 17/27] drm/i915/guc: Implement multi-lrc reset Matthew Brost
2021-09-20 22:44 ` [Intel-gfx] " John Harrison
2021-09-22 16:16 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 18/27] drm/i915/guc: Update debugfs for GuC multi-lrc Matthew Brost
2021-09-20 22:48 ` [Intel-gfx] " John Harrison
2021-09-21 19:13 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 19/27] drm/i915: Fix bug in user proto-context creation that leaked contexts Matthew Brost
2021-09-20 22:57 ` [Intel-gfx] " John Harrison
2021-09-21 14:49 ` Tvrtko Ursulin
2021-09-21 19:28 ` Matthew Brost
2021-09-21 19:28 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 20/27] drm/i915/guc: Connect UAPI to GuC multi-lrc interface Matthew Brost
2021-08-29 4:00 ` [Intel-gfx] " kernel test robot
2021-08-29 19:59 ` kernel test robot
2021-09-21 0:09 ` John Harrison
2021-09-22 16:38 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 21/27] drm/i915/doc: Update parallel submit doc to point to i915_drm.h Matthew Brost
2021-09-21 0:12 ` [Intel-gfx] " John Harrison
2021-08-20 22:44 ` [PATCH 22/27] drm/i915/guc: Add basic GuC multi-lrc selftest Matthew Brost
2021-09-28 20:47 ` [Intel-gfx] " John Harrison
2021-08-20 22:44 ` [PATCH 23/27] drm/i915/guc: Implement no mid batch preemption for multi-lrc Matthew Brost
2021-09-10 11:25 ` [Intel-gfx] " Tvrtko Ursulin
2021-09-10 20:49 ` Matthew Brost
2021-09-13 10:52 ` Tvrtko Ursulin
2021-09-28 22:20 ` John Harrison
2021-09-28 22:33 ` Matthew Brost
2021-09-28 23:33 ` John Harrison
2021-09-29 0:22 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 24/27] drm/i915: Multi-BB execbuf Matthew Brost
2021-08-21 19:01 ` [Intel-gfx] " kernel test robot
2021-08-30 3:46 ` kernel test robot
2021-09-30 22:16 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 25/27] drm/i915/guc: Handle errors in multi-lrc requests Matthew Brost
2021-09-29 20:44 ` [Intel-gfx] " John Harrison
2021-09-29 20:58 ` Matthew Brost
2021-08-20 22:44 ` [PATCH 26/27] drm/i915: Enable multi-bb execbuf Matthew Brost
2021-08-20 22:44 ` [PATCH 27/27] drm/i915/execlists: Weak parallel submission support for execlists Matthew Brost
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=c840bf5b-c5e9-872d-e04f-2dfe1852e555@intel.com \
--to=john.c.harrison@intel.com \
--cc=daniel.vetter@ffwll.ch \
--cc=dri-devel@lists.freedesktop.org \
--cc=intel-gfx@lists.freedesktop.org \
--cc=matthew.brost@intel.com \
--cc=tony.ye@intel.com \
--cc=zhengguo.xu@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).