-
Notifications
You must be signed in to change notification settings - Fork 0
/
drm-i915-Boost-GPU-clocks-if-we-miss-the-pageflip-s-vblank.patch
200 lines (190 loc) · 6.19 KB
/
drm-i915-Boost-GPU-clocks-if-we-miss-the-pageflip-s-vblank.patch
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
From patchwork Thu Aug 17 12:37:06 2017
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Subject: drm/i915: Boost GPU clocks if we miss the pageflip's vblank
From: Chris Wilson <[email protected]>
X-Patchwork-Id: 172204
Message-Id: <[email protected]>
Cc: Daniel Vetter <[email protected]>
Date: Thu, 17 Aug 2017 13:37:06 +0100
If we miss the current vblank because the gpu was busy, that may cause a
jitter as the frame rate temporarily drops. We try to limit the impact
of this by then boosting the GPU clock to deliver the frame as quickly
as possible. Originally done in commit 6ad790c0f5ac ("drm/i915: Boost GPU
frequency if we detect outstanding pageflips") but was never forward
ported to atomic and finally dropped in commit fd3a40242e87 ("drm/i915:
Rip out legacy page_flip completion/irq handling").
References: https://bugs.freedesktop.org/show_bug.cgi?id=102199
Signed-off-by: Chris Wilson <[email protected]>
Cc: Maarten Lankhorst <[email protected]>
Cc: Ville Syrjälä <[email protected]>
Cc: Daniel Vetter <[email protected]>
Tested-by: Lyude Paul <[email protected]>
Reviewed-by: Radoslaw Szwichtenberg <[email protected]>
---
drivers/gpu/drm/i915/intel_display.c | 59 ++++++++++++++++++++++++++++++++++++
drivers/gpu/drm/i915/intel_drv.h | 1 -
drivers/gpu/drm/i915/intel_pm.c | 42 ++-----------------------
3 files changed, 62 insertions(+), 40 deletions(-)
diff --git a/drivers/gpu/drm/i915/intel_display.c b/drivers/gpu/drm/i915/intel_display.c
index 0e93ec201fe3..7d5b19553637 100644
--- a/drivers/gpu/drm/i915/intel_display.c
+++ b/drivers/gpu/drm/i915/intel_display.c
@@ -12636,6 +12636,55 @@ static const struct drm_crtc_funcs intel_crtc_funcs = {
.set_crc_source = intel_crtc_set_crc_source,
};
+struct wait_rps_boost {
+ struct wait_queue_entry wait;
+
+ struct drm_crtc *crtc;
+ struct drm_i915_gem_request *request;
+};
+
+static int do_rps_boost(struct wait_queue_entry *_wait,
+ unsigned mode, int sync, void *key)
+{
+ struct wait_rps_boost *wait = container_of(_wait, typeof(*wait), wait);
+ struct drm_i915_gem_request *rq = wait->request;
+
+ gen6_rps_boost(rq, NULL);
+ i915_gem_request_put(rq);
+
+ drm_crtc_vblank_put(wait->crtc);
+
+ list_del(&wait->wait.entry);
+ kfree(wait);
+ return 1;
+}
+
+static void add_rps_boost_after_vblank(struct drm_crtc *crtc,
+ struct dma_fence *fence)
+{
+ struct wait_rps_boost *wait;
+
+ if (!dma_fence_is_i915(fence))
+ return;
+
+ if (drm_crtc_vblank_get(crtc))
+ return;
+
+ wait = kmalloc(sizeof(*wait), GFP_KERNEL);
+ if (!wait) {
+ drm_crtc_vblank_put(crtc);
+ return;
+ }
+
+ wait->request = to_request(dma_fence_get(fence));
+ wait->crtc = crtc;
+
+ wait->wait.func = do_rps_boost;
+ wait->wait.flags = 0;
+
+ add_wait_queue(drm_crtc_vblank_waitqueue(crtc), &wait->wait);
+}
+
/**
* intel_prepare_plane_fb - Prepare fb for usage on plane
* @plane: drm plane to prepare for
@@ -12733,12 +12782,22 @@ intel_prepare_plane_fb(struct drm_plane *plane,
return ret;
if (!new_state->fence) { /* implicit fencing */
+ struct dma_fence *fence;
+
ret = i915_sw_fence_await_reservation(&intel_state->commit_ready,
obj->resv, NULL,
false, I915_FENCE_TIMEOUT,
GFP_KERNEL);
if (ret < 0)
return ret;
+
+ fence = reservation_object_get_excl_rcu(obj->resv);
+ if (fence) {
+ add_rps_boost_after_vblank(new_state->crtc, fence);
+ dma_fence_put(fence);
+ }
+ } else {
+ add_rps_boost_after_vblank(new_state->crtc, new_state->fence);
}
return 0;
diff --git a/drivers/gpu/drm/i915/intel_drv.h b/drivers/gpu/drm/i915/intel_drv.h
index fa47285918f4..e092354b4d63 100644
--- a/drivers/gpu/drm/i915/intel_drv.h
+++ b/drivers/gpu/drm/i915/intel_drv.h
@@ -1844,7 +1844,6 @@ void gen6_rps_reset_ei(struct drm_i915_private *dev_priv);
void gen6_rps_idle(struct drm_i915_private *dev_priv);
void gen6_rps_boost(struct drm_i915_gem_request *rq,
struct intel_rps_client *rps);
-void intel_queue_rps_boost_for_request(struct drm_i915_gem_request *req);
void g4x_wm_get_hw_state(struct drm_device *dev);
void vlv_wm_get_hw_state(struct drm_device *dev);
void ilk_wm_get_hw_state(struct drm_device *dev);
diff --git a/drivers/gpu/drm/i915/intel_pm.c b/drivers/gpu/drm/i915/intel_pm.c
index ed662937ec3c..c9fa2eb1903c 100644
--- a/drivers/gpu/drm/i915/intel_pm.c
+++ b/drivers/gpu/drm/i915/intel_pm.c
@@ -6169,6 +6169,7 @@ void gen6_rps_boost(struct drm_i915_gem_request *rq,
struct intel_rps_client *rps)
{
struct drm_i915_private *i915 = rq->i915;
+ unsigned long flags;
bool boost;
/* This is intentionally racy! We peek at the state here, then
@@ -6178,13 +6179,13 @@ void gen6_rps_boost(struct drm_i915_gem_request *rq,
return;
boost = false;
- spin_lock_irq(&rq->lock);
+ spin_lock_irqsave(&rq->lock, flags);
if (!rq->waitboost && !i915_gem_request_completed(rq)) {
atomic_inc(&i915->rps.num_waiters);
rq->waitboost = true;
boost = true;
}
- spin_unlock_irq(&rq->lock);
+ spin_unlock_irqrestore(&rq->lock, flags);
if (!boost)
return;
@@ -9132,43 +9133,6 @@ int intel_freq_opcode(struct drm_i915_private *dev_priv, int val)
return DIV_ROUND_CLOSEST(val, GT_FREQUENCY_MULTIPLIER);
}
-struct request_boost {
- struct work_struct work;
- struct drm_i915_gem_request *req;
-};
-
-static void __intel_rps_boost_work(struct work_struct *work)
-{
- struct request_boost *boost = container_of(work, struct request_boost, work);
- struct drm_i915_gem_request *req = boost->req;
-
- if (!i915_gem_request_completed(req))
- gen6_rps_boost(req, NULL);
-
- i915_gem_request_put(req);
- kfree(boost);
-}
-
-void intel_queue_rps_boost_for_request(struct drm_i915_gem_request *req)
-{
- struct request_boost *boost;
-
- if (req == NULL || INTEL_GEN(req->i915) < 6)
- return;
-
- if (i915_gem_request_completed(req))
- return;
-
- boost = kmalloc(sizeof(*boost), GFP_ATOMIC);
- if (boost == NULL)
- return;
-
- boost->req = i915_gem_request_get(req);
-
- INIT_WORK(&boost->work, __intel_rps_boost_work);
- queue_work(req->i915->wq, &boost->work);
-}
-
void intel_pm_setup(struct drm_i915_private *dev_priv)
{
mutex_init(&dev_priv->rps.hw_lock);