return &base[id];
}
-static int guc_workqueue_create(struct intel_guc *guc)
-{
- return intel_guc_allocate_and_map_vma(guc, GUC_WQ_SIZE, &guc->workqueue,
- &guc->workqueue_vaddr);
-}
-
-static void guc_workqueue_destroy(struct intel_guc *guc)
-{
- i915_vma_unpin_and_release(&guc->workqueue, I915_VMA_RELEASE_MAP);
-}
-
-/*
- * Initialise the process descriptor shared with the GuC firmware.
- */
-static int guc_proc_desc_create(struct intel_guc *guc)
-{
- const u32 size = PAGE_ALIGN(sizeof(struct guc_process_desc));
-
- return intel_guc_allocate_and_map_vma(guc, size, &guc->proc_desc,
- &guc->proc_desc_vaddr);
-}
-
-static void guc_proc_desc_destroy(struct intel_guc *guc)
-{
- i915_vma_unpin_and_release(&guc->proc_desc, I915_VMA_RELEASE_MAP);
-}
-
-static void guc_proc_desc_init(struct intel_guc *guc)
-{
- struct guc_process_desc *desc;
-
- desc = memset(guc->proc_desc_vaddr, 0, sizeof(*desc));
-
- /*
- * XXX: pDoorbell and WQVBaseAddress are pointers in process address
- * space for ring3 clients (set them as in mmap_ioctl) or kernel
- * space for kernel clients (map on demand instead? May make debug
- * easier to have it mapped).
- */
- desc->wq_base_addr = 0;
- desc->db_base_addr = 0;
-
- desc->wq_size_bytes = GUC_WQ_SIZE;
- desc->wq_status = WQ_STATUS_ACTIVE;
- desc->priority = GUC_CLIENT_PRIORITY_KMD_NORMAL;
-}
-
-static void guc_proc_desc_fini(struct intel_guc *guc)
-{
- memset(guc->proc_desc_vaddr, 0, sizeof(struct guc_process_desc));
-}
-
static int guc_stage_desc_pool_create(struct intel_guc *guc)
{
u32 size = PAGE_ALIGN(sizeof(struct guc_stage_desc) *
desc->stage_id = 0;
desc->priority = GUC_CLIENT_PRIORITY_KMD_NORMAL;
- desc->process_desc = intel_guc_ggtt_offset(guc, guc->proc_desc);
- desc->wq_addr = intel_guc_ggtt_offset(guc, guc->workqueue);
desc->wq_size = GUC_WQ_SIZE;
}
memset(desc, 0, sizeof(*desc));
}
-/* Construct a Work Item and append it to the GuC's Work Queue */
-static void guc_wq_item_append(struct intel_guc *guc,
- 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 guc_process_desc *desc = guc->proc_desc_vaddr;
- struct guc_wq_item *wqi;
- u32 wq_off;
-
- lockdep_assert_held(&guc->wq_lock);
-
- /* 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.
- *
- * XXX: if not the case, we need save data to a temp wqi and copy it to
- * workqueue buffer dw by dw.
- */
- BUILD_BUG_ON(wqi_size != 16);
-
- /* We expect the WQ to be active if we're appending items to it */
- GEM_BUG_ON(desc->wq_status != WQ_STATUS_ACTIVE);
-
- /* Free space is guaranteed. */
- wq_off = READ_ONCE(desc->tail);
- GEM_BUG_ON(CIRC_SPACE(wq_off, READ_ONCE(desc->head),
- GUC_WQ_SIZE) < wqi_size);
- GEM_BUG_ON(wq_off & (wqi_size - 1));
-
- wqi = guc->workqueue_vaddr + wq_off;
-
- /* Now fill in the 4-word work queue item */
- wqi->header = WQ_TYPE_INORDER |
- (wqi_len << WQ_LEN_SHIFT) |
- (target_engine << WQ_TARGET_SHIFT) |
- WQ_NO_WCFLUSH_WAIT;
- wqi->context_desc = context_desc;
- wqi->submit_element_info = ring_tail << WQ_RING_TAIL_SHIFT;
- GEM_BUG_ON(ring_tail > WQ_RING_TAIL_MAX);
- wqi->fence_id = fence_id;
-
- /* Make the update visible to GuC */
- WRITE_ONCE(desc->tail, (wq_off + wqi_size) & (GUC_WQ_SIZE - 1));
-}
-
static void guc_add_request(struct intel_guc *guc, struct i915_request *rq)
{
- struct intel_engine_cs *engine = rq->engine;
- u32 ctx_desc = rq->context->lrc.ccid;
- u32 ring_tail = intel_ring_set_tail(rq->ring, rq->tail) / sizeof(u64);
-
- guc_wq_item_append(guc, engine->guc_id, ctx_desc,
- ring_tail, rq->fence.seqno);
+ /* Leaving stub as this function will be used in future patches */
}
/*
{
struct intel_guc *guc = &engine->gt->uc.guc;
- spin_lock(&guc->wq_lock);
-
do {
struct i915_request *rq = *out++;
flush_ggtt_writes(rq->ring->vma);
guc_add_request(guc, rq);
} while (out != end);
-
- spin_unlock(&guc->wq_lock);
}
static inline int rq_prio(const struct i915_request *rq)
__tasklet_disable_sync_once(&execlists->tasklet);
}
-static void
-cancel_port_requests(struct intel_engine_execlists * const execlists)
-{
- struct i915_request * const *port, *rq;
-
- /* Note we are only using the inflight and not the pending queue */
-
- for (port = execlists->active; (rq = *port); port++)
- schedule_out(rq);
- execlists->active =
- memset(execlists->inflight, 0, sizeof(execlists->inflight));
-}
-
static void guc_reset_state(struct intel_context *ce,
struct intel_engine_cs *engine,
u32 head,
spin_lock_irqsave(&engine->active.lock, flags);
- cancel_port_requests(execlists);
-
/* Push back any incomplete requests for replay after the reset. */
rq = execlists_unwind_incomplete_requests(execlists);
if (!rq)
*/
spin_lock_irqsave(&engine->active.lock, flags);
- /* Cancel the requests on the HW and clear the ELSP tracker. */
- cancel_port_requests(execlists);
-
/* Mark all executing requests as skipped. */
list_for_each_entry(rq, &engine->active.requests, sched.link) {
i915_request_set_error_once(rq, -EIO);
atomic_read(&execlists->tasklet.count));
}
-/*
- * Everything below here is concerned with setup & teardown, and is
- * therefore not part of the somewhat time-critical batch-submission
- * path of guc_submit() above.
- */
-
/*
* Set up the memory resources to be shared with the GuC (via the GGTT)
* at firmware loading time.
*/
GEM_BUG_ON(!guc->stage_desc_pool);
- ret = guc_workqueue_create(guc);
- if (ret)
- goto err_pool;
-
- ret = guc_proc_desc_create(guc);
- if (ret)
- goto err_workqueue;
-
- spin_lock_init(&guc->wq_lock);
-
return 0;
-
-err_workqueue:
- guc_workqueue_destroy(guc);
-err_pool:
- guc_stage_desc_pool_destroy(guc);
- return ret;
}
void intel_guc_submission_fini(struct intel_guc *guc)
{
if (guc->stage_desc_pool) {
- guc_proc_desc_destroy(guc);
- guc_workqueue_destroy(guc);
guc_stage_desc_pool_destroy(guc);
}
}
struct intel_engine_cs *engine;
enum intel_engine_id id;
- /*
- * We're using GuC work items for submitting work through GuC. Since
- * we're coalescing multiple requests from a single context into a
- * single work item prior to assigning it to execlist_port, we can
- * never have more work items than the total number of ports (for all
- * engines). The GuC firmware is controlling the HEAD of work queue,
- * and it is guaranteed that it will remove the work item from the
- * queue before our request is completed.
- */
- BUILD_BUG_ON(ARRAY_SIZE(engine->execlists.inflight) *
- sizeof(struct guc_wq_item) *
- I915_NUM_ENGINES > GUC_WQ_SIZE);
-
- guc_proc_desc_init(guc);
guc_stage_desc_init(guc);
/* Take over from manual control of ELSP (execlists) */
guc_interrupts_release(gt);
guc_stage_desc_fini(guc);
- guc_proc_desc_fini(guc);
}
static bool __guc_submission_selected(struct intel_guc *guc)