summaryrefslogtreecommitdiff
path: root/libdrm/radeon/radeon_cs_gem.c
blob: f9c9fabbe6793d53927610c0d3beacfd8d2e09e0 (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
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
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
/* 
 * Copyright © 2008 Jérôme Glisse
 * All Rights Reserved.
 * 
 * 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, sub license, 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 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
 * NON-INFRINGEMENT. IN NO EVENT SHALL THE COPYRIGHT HOLDERS, AUTHORS
 * AND/OR ITS SUPPLIERS 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.
 *
 * The above copyright notice and this permission notice (including the
 * next paragraph) shall be included in all copies or substantial portions
 * of the Software.
 */
/*
 * Authors:
 *      Aapo Tahkola <aet@rasterburn.org>
 *      Nicolai Haehnle <prefect_@gmx.net>
 *      Jérôme Glisse <glisse@freedesktop.org>
 */
#include <errno.h>
#include <stdlib.h>
#include <sys/mman.h>
#include <sys/ioctl.h>
#include "radeon_cs.h"
#include "radeon_cs_gem.h"
#include "radeon_bo_gem.h"
#include "drm.h"
#include "xf86drm.h"
#include "radeon_drm.h"

#pragma pack(1)
struct cs_reloc_gem {
    uint32_t    handle;
    uint32_t    start_offset;
    uint32_t    end_offset;
    uint32_t    read_domain;
    uint32_t    write_domain;
    uint32_t    flags;
};
#pragma pack()

struct cs_gem {
    struct radeon_cs            base;
    struct drm_radeon_cs2       cs;
    struct drm_radeon_cs_chunk  chunks[2];
    unsigned                    nrelocs;
    uint32_t                    *relocs;
    struct radeon_bo            **relocs_bo;
};

static struct radeon_cs *cs_gem_create(struct radeon_cs_manager *csm,
                                       uint32_t ndw)
{
    struct cs_gem *csg;

    /* max cmd buffer size is 64Kb */
    if (ndw > (64 * 1024 / 4)) {
        return NULL;
    }
    csg = (struct cs_gem*)calloc(1, sizeof(struct cs_gem));
    if (csg == NULL) {
        return NULL;
    }
    csg->base.csm = csm;
    csg->base.ndw = 64 * 1024 / 4;
    csg->base.packets = (uint32_t*)calloc(1, 64 * 1024);
    if (csg->base.packets == NULL) {
        free(csg);
        return NULL;
    }
    csg->base.relocs_total_size = 0;
    csg->base.crelocs = 0;
    csg->nrelocs = 4096 / (4 * 4) ;
    csg->relocs_bo = (struct radeon_bo**)calloc(1,
                                                csg->nrelocs*sizeof(void*));
    if (csg->relocs_bo == NULL) {
        free(csg->base.packets);
        free(csg);
        return NULL;
    }
    csg->base.relocs = csg->relocs = (uint32_t*)calloc(1, 4096);
    if (csg->relocs == NULL) {
        free(csg->relocs_bo);
        free(csg->base.packets);
        free(csg);
        return NULL;
    }
    csg->chunks[0].chunk_id = RADEON_CHUNK_ID_IB;
    csg->chunks[0].length_dw = 0;
    csg->chunks[0].chunk_data = (uint64_t)(intptr_t)csg->base.packets;
    csg->chunks[1].chunk_id = RADEON_CHUNK_ID_RELOCS;
    csg->chunks[1].length_dw = 0;
    csg->chunks[1].chunk_data = (uint64_t)(intptr_t)csg->relocs;
    return (struct radeon_cs*)csg;
}

static int cs_gem_write_dword(struct radeon_cs *cs, uint32_t dword)
{
    struct cs_gem *csg = (struct cs_gem*)cs;
    if (cs->cdw >= cs->ndw) {
        uint32_t tmp, *ptr;
        tmp = (cs->cdw + 1 + 0x3FF) & (~0x3FF);
        ptr = (uint32_t*)realloc(cs->packets, 4 * tmp);
        if (ptr == NULL) {
            return -ENOMEM;
        }
        cs->packets = ptr;
        cs->ndw = tmp;
        csg->chunks[0].chunk_data = (uint64_t)(intptr_t)csg->base.packets;
    }
    cs->packets[cs->cdw++] = dword;
    csg->chunks[0].length_dw += 1;
    return 0;
}

static int cs_gem_write_reloc(struct radeon_cs *cs,
                              struct radeon_bo *bo,
                              uint32_t start_offset,
                              uint32_t end_offset,
                              uint32_t read_domain,
                              uint32_t write_domain,
                              uint32_t flags)
{
    struct cs_gem *csg = (struct cs_gem*)cs;
    struct cs_reloc_gem *reloc;
    uint32_t idx;
    unsigned i;

    /* check domains */
    if ((read_domain && write_domain) || (!read_domain && !write_domain)) {
        /* in one CS a bo can only be in read or write domain but not
         * in read & write domain at the same sime
         */
        return -EINVAL;
    }
    if (read_domain == RADEON_GEM_DOMAIN_CPU) {
        return -EINVAL;
    }
    if (write_domain == RADEON_GEM_DOMAIN_CPU) {
        return -EINVAL;
    }
    /* check reloc window */
    if (end_offset > bo->size) {
        return -EINVAL;
    }
    if (start_offset > end_offset) {
        return -EINVAL;
    }
    /* check if bo is already referenced */
    for(i = 0; i < cs->crelocs; i++) {
        idx = i * 6;
        reloc = (struct cs_reloc_gem*)&csg->relocs[idx];
        if (reloc->handle == bo->handle) {
            /* Check domains must be in read or write. As we check already
             * checked that in argument one of the read or write domain was
             * set we only need to check that if previous reloc as the read
             * domain set then the read_domain should also be set for this
             * new relocation.
             */
            if (reloc->read_domain && !read_domain) {
                return -EINVAL;
            }
            if (reloc->write_domain && !write_domain) {
                return -EINVAL;
            }
            reloc->read_domain |= read_domain;
            reloc->write_domain |= write_domain;
            /* update start and end offset */
            if (start_offset < reloc->start_offset) {
                reloc->start_offset = start_offset;
            }
            if (end_offset > reloc->end_offset) {
                reloc->end_offset = end_offset;
            }
            /* update flags */
            reloc->flags |= (flags & reloc->flags);
            /* write relocation packet */
            cs_gem_write_dword(cs, 0xc0001000);
            cs_gem_write_dword(cs, idx);
            return 0;
        }
    }
    /* new relocation */
    if (csg->base.crelocs >= csg->nrelocs) {
        /* allocate more memory (TODO: should use a slab allocatore maybe) */
        uint32_t *tmp, size;
        size = ((csg->nrelocs + 1) * sizeof(struct radeon_bo*));
        tmp = (uint32_t*)realloc(csg->relocs_bo, size);
        if (tmp == NULL) {
            return -ENOMEM;
        }
        csg->relocs_bo = (struct radeon_bo**)tmp;
        size = ((csg->nrelocs + 1) * 6 * 4);
        tmp = (uint32_t*)realloc(csg->relocs, size);
        if (tmp == NULL) {
            return -ENOMEM;
        }
        cs->relocs = csg->relocs = tmp;
        csg->nrelocs += 1;
        csg->chunks[1].chunk_data = (uint64_t)(intptr_t)csg->relocs;
    }
    csg->relocs_bo[csg->base.crelocs] = bo;
    idx = (csg->base.crelocs++) * 6;
    reloc = (struct cs_reloc_gem*)&csg->relocs[idx];
    reloc->handle = bo->handle;
    reloc->start_offset = start_offset;
    reloc->end_offset = end_offset;
    reloc->read_domain = read_domain;
    reloc->write_domain = write_domain;
    reloc->flags = flags;
    csg->chunks[1].length_dw += 6;
    radeon_bo_ref(bo);
    cs->relocs_total_size += bo->size;
    cs_gem_write_dword(cs, 0xc0001000);
    cs_gem_write_dword(cs, idx);
    return 0;
}

static int cs_gem_begin(struct radeon_cs *cs,
                        uint32_t ndw,
                        const char *file,
                        const char *func,
                        int line)
{
    return 0;
}

static int cs_gem_end(struct radeon_cs *cs,
                      const char *file,
                      const char *func,
                      int line)

{
    cs->section = 0;
    return 0;
}

static int cs_gem_emit(struct radeon_cs *cs)
{
    struct cs_gem *csg = (struct cs_gem*)cs;
    uint64_t chunk_array[2];
    int r;

    chunk_array[0] = (uint64_t)(intptr_t)&csg->chunks[0];
    chunk_array[1] = (uint64_t)(intptr_t)&csg->chunks[1];

    csg->cs.num_chunks = 2;
    csg->cs.chunks = (uint64_t)(intptr_t)chunk_array;

    r = drmCommandWriteRead(cs->csm->fd, DRM_RADEON_CS2,
                            &csg->cs, sizeof(struct drm_radeon_cs2));
    if (r) {
        return r;
    }
    return 0;
}

static int cs_gem_destroy(struct radeon_cs *cs)
{
    struct cs_gem *csg = (struct cs_gem*)cs;

    free(csg->relocs_bo);
    free(cs->relocs);
    free(cs->packets);
    free(cs);
    return 0;
}

static int cs_gem_erase(struct radeon_cs *cs)
{
    struct cs_gem *csg = (struct cs_gem*)cs;

    cs->relocs_total_size = 0;
    cs->cdw = 0;
    cs->section = 0;
    cs->crelocs = 0;
    csg->chunks[0].length_dw = 0;
    csg->chunks[1].length_dw = 0;
    return 0;
}

static int cs_gem_need_flush(struct radeon_cs *cs)
{
    return (cs->relocs_total_size > (16*1024*1024));
}

static struct radeon_cs_funcs radeon_cs_gem_funcs = {
    cs_gem_create,
    cs_gem_write_dword,
    cs_gem_write_reloc,
    cs_gem_begin,
    cs_gem_end,
    cs_gem_emit,
    cs_gem_destroy,
    cs_gem_erase,
    cs_gem_need_flush
};

struct radeon_cs_manager *radeon_cs_manager_gem(int fd)
{
    struct radeon_cs_manager *csm;

    csm = (struct radeon_cs_manager*)calloc(1,
                                            sizeof(struct radeon_cs_manager));
    if (csm == NULL) {
        return NULL;
    }
    csm->funcs = &radeon_cs_gem_funcs;
    csm->fd = fd;
    return csm;
}

void radeon_cs_manager_gem_shutdown(struct radeon_cs_manager *csm)
{
    free(csm);
}