summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/i915/gem/i915_gem_clflush.c
blob: 5295285d5843fc9cff676fd0c05a7fbba4e59174 (plain)
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
/*
 * SPDX-License-Identifier: MIT
 *
 * Copyright © 2016 Intel Corporation
 */

#include "display/intel_frontbuffer.h"

#include "i915_drv.h"
#include "i915_gem_clflush.h"

static DEFINE_SPINLOCK(clflush_lock);

struct clflush {
	struct dma_fence dma; /* Must be first for dma_fence_free() */
	struct i915_sw_fence wait;
	struct work_struct work;
	struct drm_i915_gem_object *obj;
};

static const char *i915_clflush_get_driver_name(struct dma_fence *fence)
{
	return DRIVER_NAME;
}

static const char *i915_clflush_get_timeline_name(struct dma_fence *fence)
{
	return "clflush";
}

static void i915_clflush_release(struct dma_fence *fence)
{
	struct clflush *clflush = container_of(fence, typeof(*clflush), dma);

	i915_sw_fence_fini(&clflush->wait);

	BUILD_BUG_ON(offsetof(typeof(*clflush), dma));
	dma_fence_free(&clflush->dma);
}

static const struct dma_fence_ops i915_clflush_ops = {
	.get_driver_name = i915_clflush_get_driver_name,
	.get_timeline_name = i915_clflush_get_timeline_name,
	.release = i915_clflush_release,
};

static void __i915_do_clflush(struct drm_i915_gem_object *obj)
{
	GEM_BUG_ON(!i915_gem_object_has_pages(obj));
	drm_clflush_sg(obj->mm.pages);
	intel_fb_obj_flush(obj, ORIGIN_CPU);
}

static void i915_clflush_work(struct work_struct *work)
{
	struct clflush *clflush = container_of(work, typeof(*clflush), work);
	struct drm_i915_gem_object *obj = clflush->obj;

	if (i915_gem_object_pin_pages(obj)) {
		DRM_ERROR("Failed to acquire obj->pages for clflushing\n");
		goto out;
	}

	__i915_do_clflush(obj);

	i915_gem_object_unpin_pages(obj);

out:
	i915_gem_object_put(obj);

	dma_fence_signal(&clflush->dma);
	dma_fence_put(&clflush->dma);
}

static int __i915_sw_fence_call
i915_clflush_notify(struct i915_sw_fence *fence,
		    enum i915_sw_fence_notify state)
{
	struct clflush *clflush = container_of(fence, typeof(*clflush), wait);

	switch (state) {
	case FENCE_COMPLETE:
		schedule_work(&clflush->work);
		break;

	case FENCE_FREE:
		dma_fence_put(&clflush->dma);
		break;
	}

	return NOTIFY_DONE;
}

bool i915_gem_clflush_object(struct drm_i915_gem_object *obj,
			     unsigned int flags)
{
	struct clflush *clflush;

	assert_object_held(obj);

	/*
	 * Stolen memory is always coherent with the GPU as it is explicitly
	 * marked as wc by the system, or the system is cache-coherent.
	 * Similarly, we only access struct pages through the CPU cache, so
	 * anything not backed by physical memory we consider to be always
	 * coherent and not need clflushing.
	 */
	if (!i915_gem_object_has_struct_page(obj)) {
		obj->cache_dirty = false;
		return false;
	}

	/* If the GPU is snooping the contents of the CPU cache,
	 * we do not need to manually clear the CPU cache lines.  However,
	 * the caches are only snooped when the render cache is
	 * flushed/invalidated.  As we always have to emit invalidations
	 * and flushes when moving into and out of the RENDER domain, correct
	 * snooping behaviour occurs naturally as the result of our domain
	 * tracking.
	 */
	if (!(flags & I915_CLFLUSH_FORCE) &&
	    obj->cache_coherent & I915_BO_CACHE_COHERENT_FOR_READ)
		return false;

	trace_i915_gem_object_clflush(obj);

	clflush = NULL;
	if (!(flags & I915_CLFLUSH_SYNC))
		clflush = kmalloc(sizeof(*clflush), GFP_KERNEL);
	if (clflush) {
		GEM_BUG_ON(!obj->cache_dirty);

		dma_fence_init(&clflush->dma,
			       &i915_clflush_ops,
			       &clflush_lock,
			       to_i915(obj->base.dev)->mm.unordered_timeline,
			       0);
		i915_sw_fence_init(&clflush->wait, i915_clflush_notify);

		clflush->obj = i915_gem_object_get(obj);
		INIT_WORK(&clflush->work, i915_clflush_work);

		dma_fence_get(&clflush->dma);

		i915_sw_fence_await_reservation(&clflush->wait,
						obj->base.resv, NULL,
						true, I915_FENCE_TIMEOUT,
						I915_FENCE_GFP);

		reservation_object_add_excl_fence(obj->base.resv,
						  &clflush->dma);

		i915_sw_fence_commit(&clflush->wait);
	} else if (obj->mm.pages) {
		__i915_do_clflush(obj);
	} else {
		GEM_BUG_ON(obj->write_domain != I915_GEM_DOMAIN_CPU);
	}

	obj->cache_dirty = false;
	return true;
}