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
|
/*
* SPDX-License-Identifier: MIT
*
* Copyright © 2016 Intel Corporation
*/
#include <drm/drm_cache.h>
#include "display/intel_frontbuffer.h"
#include "i915_drv.h"
#include "i915_gem_clflush.h"
#include "i915_sw_fence_work.h"
#include "i915_trace.h"
struct clflush {
struct dma_fence_work base;
struct drm_i915_gem_object *obj;
};
static void __do_clflush(struct drm_i915_gem_object *obj)
{
GEM_BUG_ON(!i915_gem_object_has_pages(obj));
drm_clflush_sg(obj->mm.pages);
i915_gem_object_flush_frontbuffer(obj, ORIGIN_CPU);
}
static void clflush_work(struct dma_fence_work *base)
{
struct clflush *clflush = container_of(base, typeof(*clflush), base);
__do_clflush(clflush->obj);
}
static void clflush_release(struct dma_fence_work *base)
{
struct clflush *clflush = container_of(base, typeof(*clflush), base);
i915_gem_object_unpin_pages(clflush->obj);
i915_gem_object_put(clflush->obj);
}
static const struct dma_fence_work_ops clflush_ops = {
.name = "clflush",
.work = clflush_work,
.release = clflush_release,
};
static struct clflush *clflush_work_create(struct drm_i915_gem_object *obj)
{
struct clflush *clflush;
GEM_BUG_ON(!obj->cache_dirty);
clflush = kmalloc(sizeof(*clflush), GFP_KERNEL);
if (!clflush)
return NULL;
if (__i915_gem_object_get_pages(obj) < 0) {
kfree(clflush);
return NULL;
}
dma_fence_work_init(&clflush->base, &clflush_ops);
clflush->obj = i915_gem_object_get(obj); /* obj <-> clflush cycle */
return clflush;
}
bool i915_gem_clflush_object(struct drm_i915_gem_object *obj,
unsigned int flags)
{
struct drm_i915_private *i915 = to_i915(obj->base.dev);
struct clflush *clflush;
assert_object_held(obj);
if (IS_DGFX(i915)) {
WARN_ON_ONCE(obj->cache_dirty);
return false;
}
/*
* 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 = clflush_work_create(obj);
if (clflush) {
i915_sw_fence_await_reservation(&clflush->base.chain,
obj->base.resv, NULL, true,
i915_fence_timeout(i915),
I915_FENCE_GFP);
dma_resv_add_excl_fence(obj->base.resv, &clflush->base.dma);
dma_fence_work_commit(&clflush->base);
/*
* We must have successfully populated the pages(since we are
* holding a pin on the pages as per the flush worker) to reach
* this point, which must mean we have already done the required
* flush-on-acquire, hence resetting cache_dirty here should be
* safe.
*/
obj->cache_dirty = false;
} else if (obj->mm.pages) {
__do_clflush(obj);
obj->cache_dirty = false;
} else {
GEM_BUG_ON(obj->write_domain != I915_GEM_DOMAIN_CPU);
}
return true;
}
|
