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
|
/*
* Copyright © 2014-2015 Broadcom
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
* to deal in the Software without restriction, including without limitation
* the rights to use, copy, modify, merge, publish, distribute, sublicense,
* and/or sell copies of the Software, and to permit persons to whom the
* Software is furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice (including the next
* paragraph) shall be included in all copies or substantial portions of the
* Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
* THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
* FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
* IN THE SOFTWARE.
*/
#ifndef _UAPI_VC4_DRM_H_
#define _UAPI_VC4_DRM_H_
#include <drm.h>
#define DRM_VC4_SUBMIT_CL 0x00
#define DRM_VC4_WAIT_SEQNO 0x01
#define DRM_VC4_WAIT_BO 0x02
#define DRM_VC4_CREATE_BO 0x03
#define DRM_VC4_MMAP_BO 0x04
#define DRM_IOCTL_VC4_SUBMIT_CL DRM_IOWR( DRM_COMMAND_BASE + DRM_VC4_SUBMIT_CL, struct drm_vc4_submit_cl)
#define DRM_IOCTL_VC4_WAIT_SEQNO DRM_IOWR( DRM_COMMAND_BASE + DRM_VC4_WAIT_SEQNO, struct drm_vc4_wait_seqno)
#define DRM_IOCTL_VC4_WAIT_BO DRM_IOWR( DRM_COMMAND_BASE + DRM_VC4_WAIT_BO, struct drm_vc4_wait_bo)
#define DRM_IOCTL_VC4_CREATE_BO DRM_IOWR( DRM_COMMAND_BASE + DRM_VC4_CREATE_BO, struct drm_vc4_create_bo)
#define DRM_IOCTL_VC4_MMAP_BO DRM_IOWR( DRM_COMMAND_BASE + DRM_VC4_MMAP_BO, struct drm_vc4_mmap_bo)
/**
* struct drm_vc4_submit_cl - ioctl argument for submitting commands to the 3D
* engine.
*
* Drivers typically use GPU BOs to store batchbuffers / command lists and
* their associated state. However, because the VC4 lacks an MMU, we have to
* do validation of memory accesses by the GPU commands. If we were to store
* our commands in BOs, we'd need to do uncached readback from them to do the
* validation process, which is too expensive. Instead, userspace accumulates
* commands and associated state in plain memory, then the kernel copies the
* data to its own address space, and then validates and stores it in a GPU
* BO.
*/
struct drm_vc4_submit_cl {
/* Pointer to the binner command list.
*
* This is the first set of commands executed, which runs the
* coordinate shader to determine where primitives land on the screen,
* then writes out the state updates and draw calls necessary per tile
* to the tile allocation BO.
*/
uint64_t bin_cl;
/* Pointer to the render command list.
*
* The render command list contains a set of packets to load the
* current tile's state (reading from memory, or just clearing it)
* into the GPU, then call into the tile allocation BO to run the
* stored rendering for that tile, then store the tile's state back to
* memory.
*/
uint64_t render_cl;
/* Pointer to the shader records.
*
* Shader records are the structures read by the hardware that contain
* pointers to uniforms, shaders, and vertex attributes. The
* reference to the shader record has enough information to determine
* how many pointers are necessary (fixed number for shaders/uniforms,
* and an attribute count), so those BO indices into bo_handles are
* just stored as uint32_ts before each shader record passed in.
*/
uint64_t shader_rec;
/* Pointer to uniform data and texture handles for the textures
* referenced by the shader.
*
* For each shader state record, there is a set of uniform data in the
* order referenced by the record (FS, VS, then CS). Each set of
* uniform data has a uint32_t index into bo_handles per texture
* sample operation, in the order the QPU_W_TMUn_S writes appear in
* the program. Following the texture BO handle indices is the actual
* uniform data.
*
* The individual uniform state blocks don't have sizes passed in,
* because the kernel has to determine the sizes anyway during shader
* code validation.
*/
uint64_t uniforms;
uint64_t bo_handles;
/* Size in bytes of the binner command list. */
uint32_t bin_cl_size;
/* Size in bytes of the render command list */
uint32_t render_cl_size;
/* Size in bytes of the set of shader records. */
uint32_t shader_rec_size;
/* Number of shader records.
*
* This could just be computed from the contents of shader_records and
* the address bits of references to them from the bin CL, but it
* keeps the kernel from having to resize some allocations it makes.
*/
uint32_t shader_rec_count;
/* Size in bytes of the uniform state. */
uint32_t uniforms_size;
/* Number of BO handles passed in (size is that times 4). */
uint32_t bo_handle_count;
uint32_t flags;
uint32_t pad;
/* Returned value of the seqno of this render job (for the
* wait ioctl).
*/
uint64_t seqno;
};
/**
* struct drm_vc4_wait_seqno - ioctl argument for waiting for
* DRM_VC4_SUBMIT_CL completion using its returned seqno.
*
* timeout_ns is the timeout in nanoseconds, where "0" means "don't
* block, just return the status."
*/
struct drm_vc4_wait_seqno {
uint64_t seqno;
uint64_t timeout_ns;
};
/**
* struct drm_vc4_wait_bo - ioctl argument for waiting for
* completion of the last DRM_VC4_SUBMIT_CL on a BO.
*
* This is useful for cases where multiple processes might be
* rendering to a BO and you want to wait for all rendering to be
* completed.
*/
struct drm_vc4_wait_bo {
uint32_t handle;
uint32_t pad;
uint64_t timeout_ns;
};
/**
* struct drm_vc4_create_bo - ioctl argument for creating VC4 BOs.
*
* There are currently no values for the flags argument, but it may be
* used in a future extension.
*/
struct drm_vc4_create_bo {
uint32_t size;
uint32_t flags;
/** Returned GEM handle for the BO. */
uint32_t handle;
uint32_t pad;
};
/**
* struct drm_vc4_mmap_bo - ioctl argument for mapping VC4 BOs.
*
* This doesn't actually perform an mmap. Instead, it returns the
* offset you need to use in an mmap on the DRM device node. This
* means that tools like valgrind end up knowing about the mapped
* memory.
*
* There are currently no values for the flags argument, but it may be
* used in a future extension.
*/
struct drm_vc4_mmap_bo {
/** Handle for the object being mapped. */
uint32_t handle;
uint32_t flags;
/** offset into the drm node to use for subsequent mmap call. */
uint64_t offset;
};
#endif /* _UAPI_VC4_DRM_H_ */
|
