[PATCH 1/9] drm/radeon: wake up all fences on manual reset

[PATCH 1/9] drm/radeon: wake up all fences on manual reset

[Christian König]

From: Christian König <christian.koenig@amd.com>

Wake up all fences when we manually trigger an reset.

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/radeon_fence.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 9137870..5bd837a 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -833,6 +833,7 @@ static int radeon_debugfs_gpu_reset(struct seq_file *m, void *data)
 	down_read(&rdev->exclusive_lock);
 	seq_printf(m, "%d\n", rdev->needs_reset);
 	rdev->needs_reset = true;
+	wake_up_all(&rdev->fence_queue);
 	up_read(&rdev->exclusive_lock);
 
 	return 0;
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 2/9] drm/radeon: fix pm handling in radeon_gpu_reset

[Christian König]

From: Alex Deucher <alexander.deucher@amd.com>

pm_suspend is handled in the radeon_suspend callbacks.
pm_resume has special handling depending on whether
dpm or legacy pm is enabled.  Change radeon_gpu_reset
to mirror the behavior in the suspend and resume
pathes.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/radeon/radeon_device.c | 18 ++++++++++++++++--
 1 file changed, 16 insertions(+), 2 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 152181d..d2ff12e 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1682,7 +1682,6 @@ int radeon_gpu_reset(struct radeon_device *rdev)
 	radeon_save_bios_scratch_regs(rdev);
 	/* block TTM */
 	resched = ttm_bo_lock_delayed_workqueue(&rdev->mman.bdev);
-	radeon_pm_suspend(rdev);
 	radeon_suspend(rdev);
 
 	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
@@ -1728,9 +1727,24 @@ retry:
 		}
 	}
 
-	radeon_pm_resume(rdev);
+	if ((rdev->pm.pm_method == PM_METHOD_DPM) && rdev->pm.dpm_enabled) {
+		/* do dpm late init */
+		r = radeon_pm_late_init(rdev);
+		if (r) {
+			rdev->pm.dpm_enabled = false;
+			DRM_ERROR("radeon_pm_late_init failed, disabling dpm\n");
+		}
+	} else {
+		/* resume old pm late */
+		radeon_pm_resume(rdev);
+	}
+
 	drm_helper_resume_force_mode(rdev->ddev);
 
+	/* set the power state here in case we are a PX system or headless */
+	if ((rdev->pm.pm_method == PM_METHOD_DPM) && rdev->pm.dpm_enabled)
+		radeon_pm_compute_clocks(rdev);
+
 	ttm_bo_unlock_delayed_workqueue(&rdev->mman.bdev, resched);
 	if (r) {
 		/* bad news, how to tell it to userspace ? */
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 3/9] drm/radeon: fix display handling in radeon_gpu_reset

[Christian König]

From: Alex Deucher <alexander.deucher@amd.com>

If the display hw was reset or a hard reset was used,
we need to re-init some of the common display hardware as well.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
---
 drivers/gpu/drm/radeon/radeon_device.c | 16 ++++++++++++++++
 1 file changed, 16 insertions(+)

diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index d2ff12e..b6aee40 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1683,6 +1683,7 @@ int radeon_gpu_reset(struct radeon_device *rdev)
 	/* block TTM */
 	resched = ttm_bo_lock_delayed_workqueue(&rdev->mman.bdev);
 	radeon_suspend(rdev);
+	radeon_hpd_fini(rdev);
 
 	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
 		ring_sizes[i] = radeon_ring_backup(rdev, &rdev->ring[i],
@@ -1739,6 +1740,21 @@ retry:
 		radeon_pm_resume(rdev);
 	}
 
+	/* init dig PHYs, disp eng pll */
+	if (rdev->is_atom_bios) {
+		radeon_atom_encoder_init(rdev);
+		radeon_atom_disp_eng_pll_init(rdev);
+		/* turn on the BL */
+		if (rdev->mode_info.bl_encoder) {
+			u8 bl_level = radeon_get_backlight_level(rdev,
+								 rdev->mode_info.bl_encoder);
+			radeon_set_backlight_level(rdev, rdev->mode_info.bl_encoder,
+						   bl_level);
+		}
+	}
+	/* reset hpd state */
+	radeon_hpd_init(rdev);
+
 	drm_helper_resume_force_mode(rdev->ddev);
 
 	/* set the power state here in case we are a PX system or headless */
-- 
1.9.1

[PATCH 4/9] drm/radeon: force fence completion only on problematic rings

[Christian König]

From: Christian König <christian.koenig@amd.com>

Instead of resetting all fence numbers, only reset the
number of the problematic ring. Split out from a patch
from Maarten Lankhorst <maarten.lankhorst@canonical.com>

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/radeon.h        |  2 +-
 drivers/gpu/drm/radeon/radeon_device.c |  6 +-----
 drivers/gpu/drm/radeon/radeon_fence.c  | 12 ++++--------
 drivers/gpu/drm/radeon/radeon_ib.c     |  1 +
 4 files changed, 7 insertions(+), 14 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 8810df3..821bf51 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -370,7 +370,7 @@ struct radeon_fence {
 int radeon_fence_driver_start_ring(struct radeon_device *rdev, int ring);
 int radeon_fence_driver_init(struct radeon_device *rdev);
 void radeon_fence_driver_fini(struct radeon_device *rdev);
-void radeon_fence_driver_force_completion(struct radeon_device *rdev);
+void radeon_fence_driver_force_completion(struct radeon_device *rdev, int ring);
 int radeon_fence_emit(struct radeon_device *rdev, struct radeon_fence **fence, int ring);
 void radeon_fence_process(struct radeon_device *rdev, int ring);
 bool radeon_fence_signaled(struct radeon_fence *fence);
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index b6aee40..bca7d45 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1488,7 +1488,6 @@ int radeon_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
 	struct drm_crtc *crtc;
 	struct drm_connector *connector;
 	int i, r;
-	bool force_completion = false;
 
 	if (dev == NULL || dev->dev_private == NULL) {
 		return -ENODEV;
@@ -1532,12 +1531,9 @@ int radeon_suspend_kms(struct drm_device *dev, bool suspend, bool fbcon)
 		r = radeon_fence_wait_empty(rdev, i);
 		if (r) {
 			/* delay GPU reset to resume */
-			force_completion = true;
+			radeon_fence_driver_force_completion(rdev, i);
 		}
 	}
-	if (force_completion) {
-		radeon_fence_driver_force_completion(rdev);
-	}
 
 	radeon_save_bios_scratch_regs(rdev);
 
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 5bd837a..e8a28e7 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -758,7 +758,7 @@ void radeon_fence_driver_fini(struct radeon_device *rdev)
 		r = radeon_fence_wait_empty(rdev, ring);
 		if (r) {
 			/* no need to trigger GPU reset as we are unloading */
-			radeon_fence_driver_force_completion(rdev);
+			radeon_fence_driver_force_completion(rdev, ring);
 		}
 		wake_up_all(&rdev->fence_queue);
 		radeon_scratch_free(rdev, rdev->fence_drv[ring].scratch_reg);
@@ -771,19 +771,15 @@ void radeon_fence_driver_fini(struct radeon_device *rdev)
  * radeon_fence_driver_force_completion - force all fence waiter to complete
  *
  * @rdev: radeon device pointer
+ * @ring: the ring to complete
  *
  * In case of GPU reset failure make sure no process keep waiting on fence
  * that will never complete.
  */
-void radeon_fence_driver_force_completion(struct radeon_device *rdev)
+void radeon_fence_driver_force_completion(struct radeon_device *rdev, int ring)
 {
-	int ring;
-
-	for (ring = 0; ring < RADEON_NUM_RINGS; ring++) {
-		if (!rdev->fence_drv[ring].initialized)
-			continue;
+	if (rdev->fence_drv[ring].initialized)
 		radeon_fence_write(rdev, rdev->fence_drv[ring].sync_seq[ring], ring);
-	}
 }
 
 
diff --git a/drivers/gpu/drm/radeon/radeon_ib.c b/drivers/gpu/drm/radeon/radeon_ib.c
index 65b0c21..bc77e01 100644
--- a/drivers/gpu/drm/radeon/radeon_ib.c
+++ b/drivers/gpu/drm/radeon/radeon_ib.c
@@ -268,6 +268,7 @@ int radeon_ib_ring_tests(struct radeon_device *rdev)
 
 		r = radeon_ib_test(rdev, i, ring);
 		if (r) {
+			radeon_fence_driver_force_completion(rdev, i);
 			ring->ready = false;
 			rdev->needs_reset = false;
 
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 5/9] drm/radeon: take exclusive_lock in read mode during ring tests, v4

[Christian König]

From: Maarten Lankhorst <maarten.lankhorst@canonical.com>

This is needed for the next commit, because the lockup detection
will need the read lock to run.

v4 (chk): split out forced fence completion, remove unrelated changes,
          add and handle in_reset flag

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/radeon.h         |  2 +-
 drivers/gpu/drm/radeon/radeon_cs.c      |  7 ++++++
 drivers/gpu/drm/radeon/radeon_device.c  | 39 ++++++++++++++-------------------
 drivers/gpu/drm/radeon/radeon_display.c |  4 +++-
 4 files changed, 28 insertions(+), 24 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index 821bf51..f528ae8 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -2326,7 +2326,7 @@ struct radeon_device {
 	bool				need_dma32;
 	bool				accel_working;
 	bool				fastfb_working; /* IGP feature*/
-	bool				needs_reset;
+	bool				needs_reset, in_reset;
 	struct radeon_surface_reg surface_regs[RADEON_GEM_MAX_SURFACES];
 	const struct firmware *me_fw;	/* all family ME firmware */
 	const struct firmware *pfp_fw;	/* r6/700 PFP firmware */
diff --git a/drivers/gpu/drm/radeon/radeon_cs.c b/drivers/gpu/drm/radeon/radeon_cs.c
index c91d8c8..06cbc9a 100644
--- a/drivers/gpu/drm/radeon/radeon_cs.c
+++ b/drivers/gpu/drm/radeon/radeon_cs.c
@@ -652,6 +652,13 @@ int radeon_cs_ioctl(struct drm_device *dev, void *data, struct drm_file *filp)
 		up_read(&rdev->exclusive_lock);
 		return -EBUSY;
 	}
+	if (rdev->in_reset) {
+		up_read(&rdev->exclusive_lock);
+		r = radeon_gpu_reset(rdev);
+		if (!r)
+			r = -EAGAIN;
+		return r;
+	}
 	/* initialize parser */
 	memset(&parser, 0, sizeof(struct radeon_cs_parser));
 	parser.filp = filp;
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index bca7d45..9f66637 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1673,6 +1673,7 @@ int radeon_gpu_reset(struct radeon_device *rdev)
 		return 0;
 	}
 
+	rdev->in_reset = true;
 	rdev->needs_reset = false;
 
 	radeon_save_bios_scratch_regs(rdev);
@@ -1691,7 +1692,6 @@ int radeon_gpu_reset(struct radeon_device *rdev)
 		}
 	}
 
-retry:
 	r = radeon_asic_reset(rdev);
 	if (!r) {
 		dev_info(rdev->dev, "GPU reset succeeded, trying to resume\n");
@@ -1700,26 +1700,12 @@ retry:
 
 	radeon_restore_bios_scratch_regs(rdev);
 
-	if (!r) {
-		for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+		if (!r && ring_data[i]) {
 			radeon_ring_restore(rdev, &rdev->ring[i],
 					    ring_sizes[i], ring_data[i]);
-			ring_sizes[i] = 0;
-			ring_data[i] = NULL;
-		}
-
-		r = radeon_ib_ring_tests(rdev);
-		if (r) {
-			dev_err(rdev->dev, "ib ring test failed (%d).\n", r);
-			if (saved) {
-				saved = false;
-				radeon_suspend(rdev);
-				goto retry;
-			}
-		}
-	} else {
-		radeon_fence_driver_force_completion(rdev);
-		for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+		} else {
+			radeon_fence_driver_force_completion(rdev, i);
 			kfree(ring_data[i]);
 		}
 	}
@@ -1751,19 +1737,28 @@ retry:
 	/* reset hpd state */
 	radeon_hpd_init(rdev);
 
+	ttm_bo_unlock_delayed_workqueue(&rdev->mman.bdev, resched);
+	downgrade_write(&rdev->exclusive_lock);
+
 	drm_helper_resume_force_mode(rdev->ddev);
 
 	/* set the power state here in case we are a PX system or headless */
 	if ((rdev->pm.pm_method == PM_METHOD_DPM) && rdev->pm.dpm_enabled)
 		radeon_pm_compute_clocks(rdev);
 
-	ttm_bo_unlock_delayed_workqueue(&rdev->mman.bdev, resched);
-	if (r) {
+	if (!r) {
+		r = radeon_ib_ring_tests(rdev);
+		if (r && saved)
+			r = -EAGAIN;
+	} else {
 		/* bad news, how to tell it to userspace ? */
 		dev_info(rdev->dev, "GPU reset failed\n");
 	}
 
-	up_write(&rdev->exclusive_lock);
+	rdev->needs_reset = r == -EAGAIN;
+	rdev->in_reset = false;
+
+	up_read(&rdev->exclusive_lock);
 	return r;
 }
 
diff --git a/drivers/gpu/drm/radeon/radeon_display.c b/drivers/gpu/drm/radeon/radeon_display.c
index 3fdf873..bd0d687 100644
--- a/drivers/gpu/drm/radeon/radeon_display.c
+++ b/drivers/gpu/drm/radeon/radeon_display.c
@@ -405,7 +405,9 @@ static void radeon_flip_work_func(struct work_struct *__work)
 		r = radeon_fence_wait(work->fence, false);
 		if (r == -EDEADLK) {
 			up_read(&rdev->exclusive_lock);
-			r = radeon_gpu_reset(rdev);
+			do {
+				r = radeon_gpu_reset(rdev);
+			} while (r == -EAGAIN);
 			down_read(&rdev->exclusive_lock);
 		}
 		if (r)
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 6/9] drm/radeon: handle lockup in delayed work, v5

[Christian König]

From: Christian König <christian.koenig@amd.com>

v5 (chk): complete rework, start when the first fence is emitted,
          stop when the last fence is signalled, make it work
          correctly with GPU resets, cleanup radeon_fence_wait_seq

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/radeon.h       |   2 +
 drivers/gpu/drm/radeon/radeon_fence.c | 200 +++++++++++++++++++++-------------
 2 files changed, 124 insertions(+), 78 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index f528ae8..fce8b32 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -349,6 +349,7 @@ extern void evergreen_tiling_fields(unsigned tiling_flags, unsigned *bankw,
  * Fences.
  */
 struct radeon_fence_driver {
+	struct radeon_device		*rdev;
 	uint32_t			scratch_reg;
 	uint64_t			gpu_addr;
 	volatile uint32_t		*cpu_addr;
@@ -356,6 +357,7 @@ struct radeon_fence_driver {
 	uint64_t			sync_seq[RADEON_NUM_RINGS];
 	atomic64_t			last_seq;
 	bool				initialized;
+	struct delayed_work		lockup_work;
 };
 
 struct radeon_fence {
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index e8a28e7..ac15f34 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -98,6 +98,25 @@ static u32 radeon_fence_read(struct radeon_device *rdev, int ring)
 }
 
 /**
+ * radeon_fence_schedule_check - schedule lockup check
+ *
+ * @rdev: radeon_device pointer
+ * @ring: ring index we should work with
+ *
+ * Queues a delayed work item to check for lockups.
+ */
+static void radeon_fence_schedule_check(struct radeon_device *rdev, int ring)
+{
+	/*
+	 * Do not reset the timer here with mod_delayed_work,
+	 * this can livelock in an interaction with TTM delayed destroy.
+	 */
+	queue_delayed_work(system_power_efficient_wq,
+			   &rdev->fence_drv[ring].lockup_work,
+			   RADEON_FENCE_JIFFIES_TIMEOUT);
+}
+
+/**
  * radeon_fence_emit - emit a fence on the requested ring
  *
  * @rdev: radeon_device pointer
@@ -122,19 +141,21 @@ int radeon_fence_emit(struct radeon_device *rdev,
 	(*fence)->ring = ring;
 	radeon_fence_ring_emit(rdev, ring, *fence);
 	trace_radeon_fence_emit(rdev->ddev, ring, (*fence)->seq);
+	radeon_fence_schedule_check(rdev, ring);
 	return 0;
 }
 
 /**
- * radeon_fence_process - process a fence
+ * radeon_fence_activity - check for fence activity
  *
  * @rdev: radeon_device pointer
  * @ring: ring index the fence is associated with
  *
- * Checks the current fence value and wakes the fence queue
- * if the sequence number has increased (all asics).
+ * Checks the current fence value and calculates the last
+ * signalled fence value. Returns true if activity occured
+ * on the ring, and the fence_queue should be waken up.
  */
-void radeon_fence_process(struct radeon_device *rdev, int ring)
+static bool radeon_fence_activity(struct radeon_device *rdev, int ring)
 {
 	uint64_t seq, last_seq, last_emitted;
 	unsigned count_loop = 0;
@@ -190,7 +211,67 @@ void radeon_fence_process(struct radeon_device *rdev, int ring)
 		}
 	} while (atomic64_xchg(&rdev->fence_drv[ring].last_seq, seq) > seq);
 
-	if (wake)
+	if (seq < last_emitted)
+		radeon_fence_schedule_check(rdev, ring);
+
+	return wake;
+}
+
+/**
+ * radeon_fence_check_lockup - check for hardware lockup
+ *
+ * @work: delayed work item
+ *
+ * Checks for fence activity and if there is none probe
+ * the hardware if a lockup occured.
+ */
+static void radeon_fence_check_lockup(struct work_struct *work)
+{
+	struct radeon_fence_driver *fence_drv;
+	struct radeon_device *rdev;
+	int ring;
+
+	fence_drv = container_of(work, struct radeon_fence_driver,
+				 lockup_work.work);
+	rdev = fence_drv->rdev;
+	ring = fence_drv - &rdev->fence_drv[0];
+
+	if (!down_read_trylock(&rdev->exclusive_lock)) {
+		/* just reschedule the check if a reset is going on */
+		radeon_fence_schedule_check(rdev, ring);
+		return;
+	}
+
+	if (radeon_fence_activity(rdev, ring))
+		wake_up_all(&rdev->fence_queue);
+
+	else if (radeon_ring_is_lockup(rdev, ring, &rdev->ring[ring])) {
+
+		/* good news we believe it's a lockup */
+		dev_warn(rdev->dev, "GPU lockup (current fence id "
+			 "0x%016llx last fence id 0x%016llx on ring %d)\n",
+			 (uint64_t)atomic64_read(&fence_drv->last_seq),
+			 fence_drv->sync_seq[ring], ring);
+
+		/* remember that we need an reset */
+		rdev->needs_reset = true;
+		wake_up_all(&rdev->fence_queue);
+	}
+	up_read(&rdev->exclusive_lock);
+}
+
+/**
+ * radeon_fence_process - process a fence
+ *
+ * @rdev: radeon_device pointer
+ * @ring: ring index the fence is associated with
+ *
+ * Checks the current fence value and wakes the fence queue
+ * if the sequence number has increased (all asics).
+ */
+void radeon_fence_process(struct radeon_device *rdev, int ring)
+{
+	if (radeon_fence_activity(rdev, ring))
 		wake_up_all(&rdev->fence_queue);
 }
 
@@ -300,86 +381,43 @@ static bool radeon_fence_any_seq_signaled(struct radeon_device *rdev, u64 *seq)
 static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq,
 				 bool intr)
 {
-	uint64_t last_seq[RADEON_NUM_RINGS];
-	bool signaled;
-	int i, r;
-
-	while (!radeon_fence_any_seq_signaled(rdev, target_seq)) {
+	long r;
+	int i;
 
-		/* Save current sequence values, used to check for GPU lockups */
-		for (i = 0; i < RADEON_NUM_RINGS; ++i) {
-			if (!target_seq[i])
-				continue;
+	if (radeon_fence_any_seq_signaled(rdev, target_seq))
+		return 0;
 
-			last_seq[i] = atomic64_read(&rdev->fence_drv[i].last_seq);
-			trace_radeon_fence_wait_begin(rdev->ddev, i, target_seq[i]);
-			radeon_irq_kms_sw_irq_get(rdev, i);
-		}
+	/* enable IRQs and tracing */
+	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+		if (!target_seq[i])
+			continue;
 
-		if (intr) {
-			r = wait_event_interruptible_timeout(rdev->fence_queue, (
-				(signaled = radeon_fence_any_seq_signaled(rdev, target_seq))
-				 || rdev->needs_reset), RADEON_FENCE_JIFFIES_TIMEOUT);
-		} else {
-			r = wait_event_timeout(rdev->fence_queue, (
-				(signaled = radeon_fence_any_seq_signaled(rdev, target_seq))
-				 || rdev->needs_reset), RADEON_FENCE_JIFFIES_TIMEOUT);
-		}
+		trace_radeon_fence_wait_begin(rdev->ddev, i, target_seq[i]);
+		radeon_irq_kms_sw_irq_get(rdev, i);
+	}
 
-		for (i = 0; i < RADEON_NUM_RINGS; ++i) {
-			if (!target_seq[i])
-				continue;
+	if (intr) {
+		r = wait_event_interruptible_timeout(rdev->fence_queue, (
+			radeon_fence_any_seq_signaled(rdev, target_seq)
+			 || rdev->needs_reset), MAX_SCHEDULE_TIMEOUT);
+	} else {
+		r = wait_event_timeout(rdev->fence_queue, (
+			radeon_fence_any_seq_signaled(rdev, target_seq)
+			 || rdev->needs_reset), MAX_SCHEDULE_TIMEOUT);
+	}
 
-			radeon_irq_kms_sw_irq_put(rdev, i);
-			trace_radeon_fence_wait_end(rdev->ddev, i, target_seq[i]);
-		}
+	if (rdev->needs_reset)
+		r = -EDEADLK;
 
-		if (unlikely(r < 0))
-			return r;
+	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
+		if (!target_seq[i])
+			continue;
 
-		if (unlikely(!signaled)) {
-			if (rdev->needs_reset)
-				return -EDEADLK;
-
-			/* we were interrupted for some reason and fence
-			 * isn't signaled yet, resume waiting */
-			if (r)
-				continue;
-
-			for (i = 0; i < RADEON_NUM_RINGS; ++i) {
-				if (!target_seq[i])
-					continue;
-
-				if (last_seq[i] != atomic64_read(&rdev->fence_drv[i].last_seq))
-					break;
-			}
-
-			if (i != RADEON_NUM_RINGS)
-				continue;
-
-			for (i = 0; i < RADEON_NUM_RINGS; ++i) {
-				if (!target_seq[i])
-					continue;
-
-				if (radeon_ring_is_lockup(rdev, i, &rdev->ring[i]))
-					break;
-			}
-
-			if (i < RADEON_NUM_RINGS) {
-				/* good news we believe it's a lockup */
-				dev_warn(rdev->dev, "GPU lockup (waiting for "
-					 "0x%016llx last fence id 0x%016llx on"
-					 " ring %d)\n",
-					 target_seq[i], last_seq[i], i);
-
-				/* remember that we need an reset */
-				rdev->needs_reset = true;
-				wake_up_all(&rdev->fence_queue);
-				return -EDEADLK;
-			}
-		}
+		radeon_irq_kms_sw_irq_put(rdev, i);
+		trace_radeon_fence_wait_end(rdev->ddev, i, target_seq[i]);
 	}
-	return 0;
+
+	return r < 0 ? r : 0;
 }
 
 /**
@@ -711,6 +749,9 @@ static void radeon_fence_driver_init_ring(struct radeon_device *rdev, int ring)
 		rdev->fence_drv[ring].sync_seq[i] = 0;
 	atomic64_set(&rdev->fence_drv[ring].last_seq, 0);
 	rdev->fence_drv[ring].initialized = false;
+	INIT_DELAYED_WORK(&rdev->fence_drv[ring].lockup_work,
+			  radeon_fence_check_lockup);
+	rdev->fence_drv[ring].rdev = rdev;
 }
 
 /**
@@ -760,6 +801,7 @@ void radeon_fence_driver_fini(struct radeon_device *rdev)
 			/* no need to trigger GPU reset as we are unloading */
 			radeon_fence_driver_force_completion(rdev, ring);
 		}
+		cancel_delayed_work_sync(&rdev->fence_drv[ring].lockup_work);
 		wake_up_all(&rdev->fence_queue);
 		radeon_scratch_free(rdev, rdev->fence_drv[ring].scratch_reg);
 		rdev->fence_drv[ring].initialized = false;
@@ -778,8 +820,10 @@ void radeon_fence_driver_fini(struct radeon_device *rdev)
  */
 void radeon_fence_driver_force_completion(struct radeon_device *rdev, int ring)
 {
-	if (rdev->fence_drv[ring].initialized)
+	if (rdev->fence_drv[ring].initialized) {
 		radeon_fence_write(rdev, rdev->fence_drv[ring].sync_seq[ring], ring);
+		cancel_delayed_work_sync(&rdev->fence_drv[ring].lockup_work);
+	}
 }
 
 
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 7/9] drm/radeon: add timeout argument to, radeon_fence_wait_seq v2

[Christian König]

From: Maarten Lankhorst <maarten.lankhorst@canonical.com>

This makes it possible to wait for a specific amount of time,
rather than wait until infinity.

v2 (chk): rebased on other changes

Signed-off-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/radeon_fence.c | 48 ++++++++++++++++++++---------------
 1 file changed, 28 insertions(+), 20 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index ac15f34..a54bfd6 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -364,28 +364,31 @@ static bool radeon_fence_any_seq_signaled(struct radeon_device *rdev, u64 *seq)
 }
 
 /**
- * radeon_fence_wait_seq - wait for a specific sequence numbers
+ * radeon_fence_wait_seq_timeout - wait for a specific sequence numbers
  *
  * @rdev: radeon device pointer
  * @target_seq: sequence number(s) we want to wait for
  * @intr: use interruptable sleep
+ * @timeout: maximum time to wait, or MAX_SCHEDULE_TIMEOUT for infinite wait
  *
  * Wait for the requested sequence number(s) to be written by any ring
  * (all asics).  Sequnce number array is indexed by ring id.
  * @intr selects whether to use interruptable (true) or non-interruptable
  * (false) sleep when waiting for the sequence number.  Helper function
  * for radeon_fence_wait_*().
- * Returns 0 if the sequence number has passed, error for all other cases.
+ * Returns remaining time if the sequence number has passed, 0 when
+ * the wait timeout, or an error for all other cases.
  * -EDEADLK is returned when a GPU lockup has been detected.
  */
-static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq,
-				 bool intr)
+static long radeon_fence_wait_seq_timeout(struct radeon_device *rdev,
+					  u64 *target_seq, bool intr,
+					  long timeout)
 {
 	long r;
 	int i;
 
 	if (radeon_fence_any_seq_signaled(rdev, target_seq))
-		return 0;
+		return timeout;
 
 	/* enable IRQs and tracing */
 	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
@@ -399,11 +402,11 @@ static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq,
 	if (intr) {
 		r = wait_event_interruptible_timeout(rdev->fence_queue, (
 			radeon_fence_any_seq_signaled(rdev, target_seq)
-			 || rdev->needs_reset), MAX_SCHEDULE_TIMEOUT);
+			 || rdev->needs_reset), timeout);
 	} else {
 		r = wait_event_timeout(rdev->fence_queue, (
 			radeon_fence_any_seq_signaled(rdev, target_seq)
-			 || rdev->needs_reset), MAX_SCHEDULE_TIMEOUT);
+			 || rdev->needs_reset), timeout);
 	}
 
 	if (rdev->needs_reset)
@@ -417,14 +420,14 @@ static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq,
 		trace_radeon_fence_wait_end(rdev->ddev, i, target_seq[i]);
 	}
 
-	return r < 0 ? r : 0;
+	return r;
 }
 
 /**
  * radeon_fence_wait - wait for a fence to signal
  *
  * @fence: radeon fence object
- * @intr: use interruptable sleep
+ * @intr: use interruptible sleep
  *
  * Wait for the requested fence to signal (all asics).
  * @intr selects whether to use interruptable (true) or non-interruptable
@@ -434,7 +437,7 @@ static int radeon_fence_wait_seq(struct radeon_device *rdev, u64 *target_seq,
 int radeon_fence_wait(struct radeon_fence *fence, bool intr)
 {
 	uint64_t seq[RADEON_NUM_RINGS] = {};
-	int r;
+	long r;
 
 	if (fence == NULL) {
 		WARN(1, "Querying an invalid fence : %p !\n", fence);
@@ -445,9 +448,10 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)
 	if (seq[fence->ring] == RADEON_FENCE_SIGNALED_SEQ)
 		return 0;
 
-	r = radeon_fence_wait_seq(fence->rdev, seq, intr);
-	if (r)
+	r = radeon_fence_wait_seq_timeout(fence->rdev, seq, intr, MAX_SCHEDULE_TIMEOUT);
+	if (r < 0) {
 		return r;
+	}
 
 	fence->seq = RADEON_FENCE_SIGNALED_SEQ;
 	return 0;
@@ -472,7 +476,7 @@ int radeon_fence_wait_any(struct radeon_device *rdev,
 {
 	uint64_t seq[RADEON_NUM_RINGS];
 	unsigned i, num_rings = 0;
-	int r;
+	long r;
 
 	for (i = 0; i < RADEON_NUM_RINGS; ++i) {
 		seq[i] = 0;
@@ -493,8 +497,8 @@ int radeon_fence_wait_any(struct radeon_device *rdev,
 	if (num_rings == 0)
 		return -ENOENT;
 
-	r = radeon_fence_wait_seq(rdev, seq, intr);
-	if (r) {
+	r = radeon_fence_wait_seq_timeout(rdev, seq, intr, MAX_SCHEDULE_TIMEOUT);
+	if (r < 0) {
 		return r;
 	}
 	return 0;
@@ -513,6 +517,7 @@ int radeon_fence_wait_any(struct radeon_device *rdev,
 int radeon_fence_wait_next(struct radeon_device *rdev, int ring)
 {
 	uint64_t seq[RADEON_NUM_RINGS] = {};
+	long r;
 
 	seq[ring] = atomic64_read(&rdev->fence_drv[ring].last_seq) + 1ULL;
 	if (seq[ring] >= rdev->fence_drv[ring].sync_seq[ring]) {
@@ -520,7 +525,10 @@ int radeon_fence_wait_next(struct radeon_device *rdev, int ring)
 		   already the last emited fence */
 		return -ENOENT;
 	}
-	return radeon_fence_wait_seq(rdev, seq, false);
+	r = radeon_fence_wait_seq_timeout(rdev, seq, false, MAX_SCHEDULE_TIMEOUT);
+	if (r < 0)
+		return r;
+	return 0;
 }
 
 /**
@@ -536,18 +544,18 @@ int radeon_fence_wait_next(struct radeon_device *rdev, int ring)
 int radeon_fence_wait_empty(struct radeon_device *rdev, int ring)
 {
 	uint64_t seq[RADEON_NUM_RINGS] = {};
-	int r;
+	long r;
 
 	seq[ring] = rdev->fence_drv[ring].sync_seq[ring];
 	if (!seq[ring])
 		return 0;
 
-	r = radeon_fence_wait_seq(rdev, seq, false);
-	if (r) {
+	r = radeon_fence_wait_seq_timeout(rdev, seq, false, MAX_SCHEDULE_TIMEOUT);
+	if (r < 0) {
 		if (r == -EDEADLK)
 			return -EDEADLK;
 
-		dev_err(rdev->dev, "error waiting for ring[%d] to become idle (%d)\n",
+		dev_err(rdev->dev, "error waiting for ring[%d] to become idle (%ld)\n",
 			ring, r);
 	}
 	return 0;
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 8/9] drm/radeon: drop RADEON_FENCE_SIGNALED_SEQ v2

[Christian König]

From: Christian König <christian.koenig@amd.com>

It's causing issues with VMID handling and comparing the
fence value two times actually doesn't make handling faster.

v2: rebased on reset changes

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/radeon.h       |  3 ---
 drivers/gpu/drm/radeon/radeon_fence.c | 18 ++----------------
 2 files changed, 2 insertions(+), 19 deletions(-)

diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index fce8b32..ca74b72 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -119,9 +119,6 @@ extern int radeon_use_pflipirq;
 #define RADEONFB_CONN_LIMIT			4
 #define RADEON_BIOS_NUM_SCRATCH			8
 
-/* fence seq are set to this number when signaled */
-#define RADEON_FENCE_SIGNALED_SEQ		0LL
-
 /* internal ring indices */
 /* r1xx+ has gfx CP ring */
 #define RADEON_RING_TYPE_GFX_INDEX		0
diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index a54bfd6..ecdba3a 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -328,16 +328,10 @@ static bool radeon_fence_seq_signaled(struct radeon_device *rdev,
  */
 bool radeon_fence_signaled(struct radeon_fence *fence)
 {
-	if (!fence) {
-		return true;
-	}
-	if (fence->seq == RADEON_FENCE_SIGNALED_SEQ) {
+	if (!fence)
 		return true;
-	}
-	if (radeon_fence_seq_signaled(fence->rdev, fence->seq, fence->ring)) {
-		fence->seq = RADEON_FENCE_SIGNALED_SEQ;
+	if (radeon_fence_seq_signaled(fence->rdev, fence->seq, fence->ring))
 		return true;
-	}
 	return false;
 }
 
@@ -445,15 +439,11 @@ int radeon_fence_wait(struct radeon_fence *fence, bool intr)
 	}
 
 	seq[fence->ring] = fence->seq;
-	if (seq[fence->ring] == RADEON_FENCE_SIGNALED_SEQ)
-		return 0;
-
 	r = radeon_fence_wait_seq_timeout(fence->rdev, seq, intr, MAX_SCHEDULE_TIMEOUT);
 	if (r < 0) {
 		return r;
 	}
 
-	fence->seq = RADEON_FENCE_SIGNALED_SEQ;
 	return 0;
 }
 
@@ -487,10 +477,6 @@ int radeon_fence_wait_any(struct radeon_device *rdev,
 
 		seq[i] = fences[i]->seq;
 		++num_rings;
-
-		/* test if something was allready signaled */
-		if (seq[i] == RADEON_FENCE_SIGNALED_SEQ)
-			return 0;
 	}
 
 	/* nothing to wait for ? */
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

[PATCH 9/9] drm/radeon: drop doing resets in a work item

[Christian König]

From: Christian König <christian.koenig@amd.com>

Blocking completely innocent processes with a GPU reset is
a pretty bad idea. Just set needs_reset and let the next
command submission or fence wait do the job.

Signed-off-by: Christian König <christian.koenig@amd.com>
---
 drivers/gpu/drm/radeon/cik.c            |  6 ++++--
 drivers/gpu/drm/radeon/radeon.h         |  1 -
 drivers/gpu/drm/radeon/radeon_device.c  |  7 ++++---
 drivers/gpu/drm/radeon/radeon_irq_kms.c | 18 ------------------
 4 files changed, 8 insertions(+), 24 deletions(-)

diff --git a/drivers/gpu/drm/radeon/cik.c b/drivers/gpu/drm/radeon/cik.c
index db11446..c2eca22 100644
--- a/drivers/gpu/drm/radeon/cik.c
+++ b/drivers/gpu/drm/radeon/cik.c
@@ -8237,8 +8237,10 @@ restart_ih:
 	}
 	if (queue_hotplug)
 		schedule_work(&rdev->hotplug_work);
-	if (queue_reset)
-		schedule_work(&rdev->reset_work);
+	if (queue_reset) {
+		rdev->needs_reset = true;
+		wake_up_all(&rdev->fence_queue);
+	}
 	if (queue_thermal)
 		schedule_work(&rdev->pm.dpm.thermal.work);
 	rdev->ih.rptr = rptr;
diff --git a/drivers/gpu/drm/radeon/radeon.h b/drivers/gpu/drm/radeon/radeon.h
index ca74b72..fa66f65 100644
--- a/drivers/gpu/drm/radeon/radeon.h
+++ b/drivers/gpu/drm/radeon/radeon.h
@@ -2346,7 +2346,6 @@ struct radeon_device {
 	struct radeon_mec mec;
 	struct work_struct hotplug_work;
 	struct work_struct audio_work;
-	struct work_struct reset_work;
 	int num_crtc; /* number of crtcs */
 	struct mutex dc_hw_i2c_mutex; /* display controller hw i2c mutex */
 	bool has_uvd;
diff --git a/drivers/gpu/drm/radeon/radeon_device.c b/drivers/gpu/drm/radeon/radeon_device.c
index 9f66637..d30f1cc 100644
--- a/drivers/gpu/drm/radeon/radeon_device.c
+++ b/drivers/gpu/drm/radeon/radeon_device.c
@@ -1673,9 +1673,6 @@ int radeon_gpu_reset(struct radeon_device *rdev)
 		return 0;
 	}
 
-	rdev->in_reset = true;
-	rdev->needs_reset = false;
-
 	radeon_save_bios_scratch_regs(rdev);
 	/* block TTM */
 	resched = ttm_bo_lock_delayed_workqueue(&rdev->mman.bdev);
@@ -1738,6 +1735,10 @@ int radeon_gpu_reset(struct radeon_device *rdev)
 	radeon_hpd_init(rdev);
 
 	ttm_bo_unlock_delayed_workqueue(&rdev->mman.bdev, resched);
+
+	rdev->in_reset = true;
+	rdev->needs_reset = false;
+
 	downgrade_write(&rdev->exclusive_lock);
 
 	drm_helper_resume_force_mode(rdev->ddev);
diff --git a/drivers/gpu/drm/radeon/radeon_irq_kms.c b/drivers/gpu/drm/radeon/radeon_irq_kms.c
index 16807af..f0bff4b 100644
--- a/drivers/gpu/drm/radeon/radeon_irq_kms.c
+++ b/drivers/gpu/drm/radeon/radeon_irq_kms.c
@@ -88,23 +88,6 @@ static void radeon_hotplug_work_func(struct work_struct *work)
 }
 
 /**
- * radeon_irq_reset_work_func - execute gpu reset
- *
- * @work: work struct
- *
- * Execute scheduled gpu reset (cayman+).
- * This function is called when the irq handler
- * thinks we need a gpu reset.
- */
-static void radeon_irq_reset_work_func(struct work_struct *work)
-{
-	struct radeon_device *rdev = container_of(work, struct radeon_device,
-						  reset_work);
-
-	radeon_gpu_reset(rdev);
-}
-
-/**
  * radeon_driver_irq_preinstall_kms - drm irq preinstall callback
  *
  * @dev: drm dev pointer
@@ -284,7 +267,6 @@ int radeon_irq_kms_init(struct radeon_device *rdev)
 
 	INIT_WORK(&rdev->hotplug_work, radeon_hotplug_work_func);
 	INIT_WORK(&rdev->audio_work, r600_audio_update_hdmi);
-	INIT_WORK(&rdev->reset_work, radeon_irq_reset_work_func);
 
 	rdev->irq.installed = true;
 	r = drm_irq_install(rdev->ddev, rdev->ddev->pdev->irq);
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

Re: [PATCH 1/9] drm/radeon: wake up all fences on manual reset

[Michel Dänzer]

I noticed a bunch of typos, other than that the series looks good to me.


Patch 1:

> Wake up all fences when we manually trigger an reset.

'trigger a reset.'


Patch 2:

> pathes.

'paths.'


Patch 7:

> drm/radeon: add timeout argument to, radeon_fence_wait_seq

'to radeon_fence_wait_seq'


-- 
Earthling Michel Dänzer            |                  http://www.amd.com
Libre software enthusiast          |                Mesa and X developer
_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel

drm/radeon: reset patches and common fence preparation

[Christian König]

Hey Alex,

please pull the following patches into drm-next-3.18. Compared to the last version I only fixed the typos Michel noted.

@Maarten: Can you rebase your branch on top of those? I want to take a look at the remaining changes as well and also test that branch a bit.

Thanks,
Christian.

[PATCH 1/9] drm/radeon: wake up all fences on manual reset

[Christian König]

From: Christian König <christian.koenig@amd.com>

Wake up all fences when we manually trigger a reset.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Maarten Lankhorst <maarten.lankhorst@canonical.com>
---
 drivers/gpu/drm/radeon/radeon_fence.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/gpu/drm/radeon/radeon_fence.c b/drivers/gpu/drm/radeon/radeon_fence.c
index 9137870..5bd837a 100644
--- a/drivers/gpu/drm/radeon/radeon_fence.c
+++ b/drivers/gpu/drm/radeon/radeon_fence.c
@@ -833,6 +833,7 @@ static int radeon_debugfs_gpu_reset(struct seq_file *m, void *data)
 	down_read(&rdev->exclusive_lock);
 	seq_printf(m, "%d\n", rdev->needs_reset);
 	rdev->needs_reset = true;
+	wake_up_all(&rdev->fence_queue);
 	up_read(&rdev->exclusive_lock);
 
 	return 0;
-- 
1.9.1

_______________________________________________
dri-devel mailing list
dri-devel@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/dri-devel