[Intel-gfx] [PATCH 046/190] drm/i915: Derive GEM requests from dma-fence
Chris Wilson
chris at chris-wilson.co.uk
Mon Jan 11 01:16:57 PST 2016
dma-buf provides a generic fence class for interoperation between
drivers. Internally we use the request structure as a fence, and so with
only a little bit of interfacing we can rebase those requests on top of
dma-buf fences. This will allow us, in the future, to pass those fences
back to userspace or between drivers.
v2: The fence_context needs to be globally unique, not just unique to
this device.
Signed-off-by: Chris Wilson <chris at chris-wilson.co.uk>
Cc: Jesse Barnes <jbarnes at virtuousgeek.org>
Cc: Daniel Vetter <daniel.vetter at ffwll.ch>
---
drivers/gpu/drm/i915/i915_debugfs.c | 2 +-
drivers/gpu/drm/i915/i915_gem_request.c | 111 +++++++++++++++++++++++++----
drivers/gpu/drm/i915/i915_gem_request.h | 33 ++++-----
drivers/gpu/drm/i915/i915_gpu_error.c | 2 +-
drivers/gpu/drm/i915/i915_guc_submission.c | 2 +-
drivers/gpu/drm/i915/i915_trace.h | 2 +-
drivers/gpu/drm/i915/intel_breadcrumbs.c | 3 +-
drivers/gpu/drm/i915/intel_lrc.c | 3 +-
drivers/gpu/drm/i915/intel_ringbuffer.c | 15 ++--
drivers/gpu/drm/i915/intel_ringbuffer.h | 1 +
10 files changed, 133 insertions(+), 41 deletions(-)
diff --git a/drivers/gpu/drm/i915/i915_debugfs.c b/drivers/gpu/drm/i915/i915_debugfs.c
index 6172649b7e56..b82482573a8f 100644
--- a/drivers/gpu/drm/i915/i915_debugfs.c
+++ b/drivers/gpu/drm/i915/i915_debugfs.c
@@ -710,7 +710,7 @@ static int i915_gem_request_info(struct seq_file *m, void *data)
if (req->pid)
task = pid_task(req->pid, PIDTYPE_PID);
seq_printf(m, " %x @ %d: %s [%d]\n",
- req->seqno,
+ req->fence.seqno,
(int) (jiffies - req->emitted_jiffies),
task ? task->comm : "<unknown>",
task ? task->pid : -1);
diff --git a/drivers/gpu/drm/i915/i915_gem_request.c b/drivers/gpu/drm/i915/i915_gem_request.c
index 1c4f4d83a3c2..e366ca0dcd99 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.c
+++ b/drivers/gpu/drm/i915/i915_gem_request.c
@@ -24,6 +24,92 @@
#include "i915_drv.h"
+static inline struct drm_i915_gem_request *
+to_i915_request(struct fence *fence)
+{
+ return container_of(fence, struct drm_i915_gem_request, fence);
+}
+
+static const char *i915_fence_get_driver_name(struct fence *fence)
+{
+ return "i915";
+}
+
+static const char *i915_fence_get_timeline_name(struct fence *fence)
+{
+ return to_i915_request(fence)->ring->name;
+}
+
+static bool i915_fence_signaled(struct fence *fence)
+{
+ return i915_gem_request_completed(to_i915_request(fence));
+}
+
+static bool i915_fence_enable_signaling(struct fence *fence)
+{
+ if (i915_fence_signaled(fence))
+ return false;
+
+ return intel_engine_enable_signaling(to_i915_request(fence)) == 0;
+}
+
+static signed long i915_fence_wait(struct fence *fence,
+ bool interruptible,
+ signed long timeout_jiffies)
+{
+ s64 timeout_ns, *timeout;
+ int ret;
+
+ if (timeout_jiffies != MAX_SCHEDULE_TIMEOUT) {
+ timeout_ns = jiffies_to_nsecs(timeout_jiffies);
+ timeout = &timeout_ns;
+ } else
+ timeout = NULL;
+
+ ret = __i915_wait_request(to_i915_request(fence),
+ interruptible, timeout,
+ NULL);
+ if (ret == -ETIME)
+ return 0;
+
+ if (ret < 0)
+ return ret;
+
+ if (timeout_jiffies != MAX_SCHEDULE_TIMEOUT)
+ timeout_jiffies = nsecs_to_jiffies(timeout_ns);
+
+ return timeout_jiffies;
+}
+
+static void i915_fence_value_str(struct fence *fence, char *str, int size)
+{
+ snprintf(str, size, "%u", fence->seqno);
+}
+
+static void i915_fence_timeline_value_str(struct fence *fence, char *str,
+ int size)
+{
+ snprintf(str, size, "%u",
+ intel_ring_get_seqno(to_i915_request(fence)->ring));
+}
+
+static void i915_fence_release(struct fence *fence)
+{
+ struct drm_i915_gem_request *req = to_i915_request(fence);
+ kmem_cache_free(req->i915->requests, req);
+}
+
+static const struct fence_ops i915_fence_ops = {
+ .get_driver_name = i915_fence_get_driver_name,
+ .get_timeline_name = i915_fence_get_timeline_name,
+ .enable_signaling = i915_fence_enable_signaling,
+ .signaled = i915_fence_signaled,
+ .wait = i915_fence_wait,
+ .release = i915_fence_release,
+ .fence_value_str = i915_fence_value_str,
+ .timeline_value_str = i915_fence_timeline_value_str,
+};
+
static int
i915_gem_check_wedge(unsigned reset_counter, bool interruptible)
{
@@ -116,6 +202,7 @@ int i915_gem_request_alloc(struct intel_engine_cs *ring,
struct drm_i915_private *dev_priv = to_i915(ring->dev);
unsigned reset_counter = i915_reset_counter(&dev_priv->gpu_error);
struct drm_i915_gem_request *req;
+ u32 seqno;
int ret;
if (!req_out)
@@ -135,11 +222,17 @@ int i915_gem_request_alloc(struct intel_engine_cs *ring,
if (req == NULL)
return -ENOMEM;
- ret = i915_gem_get_seqno(dev_priv, &req->seqno);
+ ret = i915_gem_get_seqno(dev_priv, &seqno);
if (ret)
goto err;
- kref_init(&req->ref);
+ spin_lock_init(&req->lock);
+ fence_init(&req->fence,
+ &i915_fence_ops,
+ &req->lock,
+ ring->fence_context,
+ seqno);
+
req->i915 = dev_priv;
req->ring = ring;
req->reset_counter = reset_counter;
@@ -377,7 +470,7 @@ void __i915_add_request(struct drm_i915_gem_request *request,
request->emitted_jiffies = jiffies;
request->previous_seqno = ring->last_submitted_seqno;
- ring->last_submitted_seqno = request->seqno;
+ ring->last_submitted_seqno = request->fence.seqno;
list_add_tail(&request->list, &ring->request_list);
trace_i915_gem_request_add(request);
@@ -531,7 +624,7 @@ int __i915_wait_request(struct drm_i915_gem_request *req,
if (INTEL_INFO(req->i915)->gen >= 6)
gen6_rps_boost(req->i915, rps, req->emitted_jiffies);
- intel_wait_init(&wait, req->seqno);
+ intel_wait_init(&wait, req->fence.seqno);
set_task_state(wait.task, state);
/* Optimistic spin for the next ~jiffie before touching IRQs */
@@ -598,7 +691,8 @@ complete:
*timeout = 0;
}
- if (ret == 0 && rps && req->seqno == req->ring->last_submitted_seqno) {
+ if (ret == 0 && rps &&
+ req->fence.seqno == req->ring->last_submitted_seqno) {
/* The GPU is now idle and this client has stalled.
* Since no other client has submitted a request in the
* meantime, assume that this client is the only one
@@ -644,10 +738,3 @@ i915_wait_request(struct drm_i915_gem_request *req)
i915_gem_request_retire_upto(req);
return 0;
}
-
-void i915_gem_request_free(struct kref *req_ref)
-{
- struct drm_i915_gem_request *req =
- container_of(req_ref, typeof(*req), ref);
- kmem_cache_free(req->i915->requests, req);
-}
diff --git a/drivers/gpu/drm/i915/i915_gem_request.h b/drivers/gpu/drm/i915/i915_gem_request.h
index af1b825fce50..b55d0b7c7f2a 100644
--- a/drivers/gpu/drm/i915/i915_gem_request.h
+++ b/drivers/gpu/drm/i915/i915_gem_request.h
@@ -25,6 +25,8 @@
#ifndef I915_GEM_REQUEST_H
#define I915_GEM_REQUEST_H
+#include <linux/fence.h>
+
/**
* Request queue structure.
*
@@ -36,11 +38,11 @@
* emission time to be associated with the request for tracking how far ahead
* of the GPU the submission is.
*
- * The requests are reference counted, so upon creation they should have an
- * initial reference taken using kref_init
+ * The requests are reference counted.
*/
struct drm_i915_gem_request {
- struct kref ref;
+ struct fence fence;
+ spinlock_t lock;
/** On Which ring this request was generated */
struct drm_i915_private *i915;
@@ -53,12 +55,6 @@ struct drm_i915_gem_request {
*/
u32 previous_seqno;
- /** GEM sequence number associated with this request,
- * when the HWS breadcrumb is equal or greater than this the GPU
- * has finished processing this request.
- */
- u32 seqno;
-
/** Position in the ringbuffer of the start of the request */
u32 head;
@@ -126,7 +122,6 @@ int i915_gem_request_alloc(struct intel_engine_cs *ring,
struct intel_context *ctx,
struct drm_i915_gem_request **req_out);
void i915_gem_request_cancel(struct drm_i915_gem_request *req);
-void i915_gem_request_free(struct kref *req_ref);
int i915_gem_request_add_to_client(struct drm_i915_gem_request *req,
struct drm_file *file);
void i915_gem_request_retire_upto(struct drm_i915_gem_request *req);
@@ -134,7 +129,7 @@ void i915_gem_request_retire_upto(struct drm_i915_gem_request *req);
static inline uint32_t
i915_gem_request_get_seqno(struct drm_i915_gem_request *req)
{
- return req ? req->seqno : 0;
+ return req ? req->fence.seqno : 0;
}
static inline struct intel_engine_cs *
@@ -144,17 +139,23 @@ i915_gem_request_get_ring(struct drm_i915_gem_request *req)
}
static inline struct drm_i915_gem_request *
+to_request(struct fence *fence)
+{
+ /* We assume that NULL fence/request are interoperable */
+ BUILD_BUG_ON(offsetof(struct drm_i915_gem_request, fence) != 0);
+ return container_of(fence, struct drm_i915_gem_request, fence);
+}
+
+static inline struct drm_i915_gem_request *
i915_gem_request_reference(struct drm_i915_gem_request *req)
{
- if (req)
- kref_get(&req->ref);
- return req;
+ return to_request(fence_get(&req->fence));
}
static inline void
i915_gem_request_unreference(struct drm_i915_gem_request *req)
{
- kref_put(&req->ref, i915_gem_request_free);
+ fence_put(&req->fence);
}
static inline void i915_gem_request_assign(struct drm_i915_gem_request **pdst,
@@ -203,7 +204,7 @@ static inline bool i915_gem_request_started(struct drm_i915_gem_request *req)
static inline bool i915_gem_request_completed(struct drm_i915_gem_request *req)
{
return i915_seqno_passed(intel_ring_get_seqno(req->ring),
- req->seqno);
+ req->fence.seqno);
}
#endif /* I915_GEM_REQUEST_H */
diff --git a/drivers/gpu/drm/i915/i915_gpu_error.c b/drivers/gpu/drm/i915/i915_gpu_error.c
index 86f582115313..05f054898a95 100644
--- a/drivers/gpu/drm/i915/i915_gpu_error.c
+++ b/drivers/gpu/drm/i915/i915_gpu_error.c
@@ -1092,7 +1092,7 @@ static void i915_gem_record_rings(struct drm_device *dev,
}
erq = &error->ring[i].requests[count++];
- erq->seqno = request->seqno;
+ erq->seqno = request->fence.seqno;
erq->jiffies = request->emitted_jiffies;
erq->tail = request->postfix;
}
diff --git a/drivers/gpu/drm/i915/i915_guc_submission.c b/drivers/gpu/drm/i915/i915_guc_submission.c
index 9c244247c13e..56d3064d32ed 100644
--- a/drivers/gpu/drm/i915/i915_guc_submission.c
+++ b/drivers/gpu/drm/i915/i915_guc_submission.c
@@ -616,7 +616,7 @@ int i915_guc_submit(struct i915_guc_client *client,
client->retcode = 0;
}
guc->submissions[ring_id] += 1;
- guc->last_seqno[ring_id] = rq->seqno;
+ guc->last_seqno[ring_id] = rq->fence.seqno;
return q_ret;
}
diff --git a/drivers/gpu/drm/i915/i915_trace.h b/drivers/gpu/drm/i915/i915_trace.h
index 43bb2e0bb949..dc2ff5cac2f4 100644
--- a/drivers/gpu/drm/i915/i915_trace.h
+++ b/drivers/gpu/drm/i915/i915_trace.h
@@ -503,7 +503,7 @@ TRACE_EVENT(i915_gem_ring_dispatch,
__entry->ring = ring->id;
__entry->seqno = i915_gem_request_get_seqno(req);
__entry->flags = flags;
- intel_engine_enable_signaling(req);
+ fence_enable_sw_signaling(&req->fence);
),
TP_printk("dev=%u, ring=%u, seqno=%u, flags=%x",
diff --git a/drivers/gpu/drm/i915/intel_breadcrumbs.c b/drivers/gpu/drm/i915/intel_breadcrumbs.c
index f6731aac7fcf..61e18cb90850 100644
--- a/drivers/gpu/drm/i915/intel_breadcrumbs.c
+++ b/drivers/gpu/drm/i915/intel_breadcrumbs.c
@@ -390,6 +390,7 @@ static int intel_breadcrumbs_signaler(void *arg)
*/
intel_engine_remove_wait(engine, &signal->wait);
+ fence_signal(&signal->request->fence);
i915_gem_request_unreference(signal->request);
/* Find the next oldest signal. Note that as we have
@@ -456,7 +457,7 @@ int intel_engine_enable_signaling(struct drm_i915_gem_request *request)
}
signal->wait.task = task;
- signal->wait.seqno = request->seqno;
+ signal->wait.seqno = request->fence.seqno;
signal->request = i915_gem_request_reference(request);
diff --git a/drivers/gpu/drm/i915/intel_lrc.c b/drivers/gpu/drm/i915/intel_lrc.c
index 7a3069a2beb2..f43a94ae5c76 100644
--- a/drivers/gpu/drm/i915/intel_lrc.c
+++ b/drivers/gpu/drm/i915/intel_lrc.c
@@ -1798,7 +1798,7 @@ static int gen8_emit_request(struct drm_i915_gem_request *request)
(ring->status_page.gfx_addr +
(I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT)));
intel_logical_ring_emit(ringbuf, 0);
- intel_logical_ring_emit(ringbuf, request->seqno);
+ intel_logical_ring_emit(ringbuf, request->fence.seqno);
intel_logical_ring_emit(ringbuf, MI_USER_INTERRUPT);
intel_logical_ring_emit(ringbuf, MI_NOOP);
intel_logical_ring_advance_and_submit(request);
@@ -1909,6 +1909,7 @@ static int logical_ring_init(struct drm_device *dev, struct intel_engine_cs *rin
ring->dev = dev;
ring->i915 = to_i915(dev);
+ ring->fence_context = fence_context_alloc(1);
INIT_LIST_HEAD(&ring->active_list);
INIT_LIST_HEAD(&ring->request_list);
i915_gem_batch_pool_init(dev, &ring->batch_pool);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.c b/drivers/gpu/drm/i915/intel_ringbuffer.c
index d9bb6458fa60..e8a7a1045c06 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.c
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.c
@@ -1218,7 +1218,7 @@ static int gen8_rcs_signal(struct drm_i915_gem_request *signaller_req,
PIPE_CONTROL_FLUSH_ENABLE);
intel_ring_emit(signaller, lower_32_bits(gtt_offset));
intel_ring_emit(signaller, upper_32_bits(gtt_offset));
- intel_ring_emit(signaller, signaller_req->seqno);
+ intel_ring_emit(signaller, signaller_req->fence.seqno);
intel_ring_emit(signaller, 0);
intel_ring_emit(signaller, MI_SEMAPHORE_SIGNAL |
MI_SEMAPHORE_TARGET(waiter->id));
@@ -1256,7 +1256,7 @@ static int gen8_xcs_signal(struct drm_i915_gem_request *signaller_req,
intel_ring_emit(signaller, lower_32_bits(gtt_offset) |
MI_FLUSH_DW_USE_GTT);
intel_ring_emit(signaller, upper_32_bits(gtt_offset));
- intel_ring_emit(signaller, signaller_req->seqno);
+ intel_ring_emit(signaller, signaller_req->fence.seqno);
intel_ring_emit(signaller, MI_SEMAPHORE_SIGNAL |
MI_SEMAPHORE_TARGET(waiter->id));
intel_ring_emit(signaller, 0);
@@ -1289,7 +1289,7 @@ static int gen6_signal(struct drm_i915_gem_request *signaller_req,
if (i915_mmio_reg_valid(mbox_reg)) {
intel_ring_emit(signaller, MI_LOAD_REGISTER_IMM(1));
intel_ring_emit_reg(signaller, mbox_reg);
- intel_ring_emit(signaller, signaller_req->seqno);
+ intel_ring_emit(signaller, signaller_req->fence.seqno);
}
}
@@ -1324,7 +1324,7 @@ gen6_add_request(struct drm_i915_gem_request *req)
intel_ring_emit(ring, MI_STORE_DWORD_INDEX);
intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
- intel_ring_emit(ring, req->seqno);
+ intel_ring_emit(ring, req->fence.seqno);
intel_ring_emit(ring, MI_USER_INTERRUPT);
__intel_ring_advance(ring);
@@ -1448,7 +1448,7 @@ pc_render_add_request(struct drm_i915_gem_request *req)
PIPE_CONTROL_QW_WRITE |
PIPE_CONTROL_WRITE_FLUSH);
intel_ring_emit(ring, addr | PIPE_CONTROL_GLOBAL_GTT);
- intel_ring_emit(ring, req->seqno);
+ intel_ring_emit(ring, req->fence.seqno);
intel_ring_emit(ring, 0);
PIPE_CONTROL_FLUSH(ring, scratch_addr);
scratch_addr += 2 * CACHELINE_BYTES; /* write to separate cachelines */
@@ -1467,7 +1467,7 @@ pc_render_add_request(struct drm_i915_gem_request *req)
PIPE_CONTROL_WRITE_FLUSH |
PIPE_CONTROL_NOTIFY);
intel_ring_emit(ring, addr | PIPE_CONTROL_GLOBAL_GTT);
- intel_ring_emit(ring, req->seqno);
+ intel_ring_emit(ring, req->fence.seqno);
intel_ring_emit(ring, 0);
__intel_ring_advance(ring);
@@ -1577,7 +1577,7 @@ i9xx_add_request(struct drm_i915_gem_request *req)
intel_ring_emit(ring, MI_STORE_DWORD_INDEX);
intel_ring_emit(ring, I915_GEM_HWS_INDEX << MI_STORE_DWORD_INDEX_SHIFT);
- intel_ring_emit(ring, req->seqno);
+ intel_ring_emit(ring, req->fence.seqno);
intel_ring_emit(ring, MI_USER_INTERRUPT);
__intel_ring_advance(ring);
@@ -2010,6 +2010,7 @@ static int intel_init_ring_buffer(struct drm_device *dev,
ring->dev = dev;
ring->i915 = to_i915(dev);
+ ring->fence_context = fence_context_alloc(1);
INIT_LIST_HEAD(&ring->active_list);
INIT_LIST_HEAD(&ring->request_list);
INIT_LIST_HEAD(&ring->execlist_queue);
diff --git a/drivers/gpu/drm/i915/intel_ringbuffer.h b/drivers/gpu/drm/i915/intel_ringbuffer.h
index eecf9c7ae2b8..a1fcb6c7501f 100644
--- a/drivers/gpu/drm/i915/intel_ringbuffer.h
+++ b/drivers/gpu/drm/i915/intel_ringbuffer.h
@@ -159,6 +159,7 @@ struct intel_engine_cs {
} id;
#define I915_NUM_RINGS 5
#define LAST_USER_RING (VECS + 1)
+ unsigned fence_context;
u32 mmio_base;
struct drm_device *dev;
struct drm_i915_private *i915;
--
2.7.0.rc3
More information about the Intel-gfx
mailing list