aboutsummaryrefslogtreecommitdiffstats
path: root/libavutil/hwcontext_cuda.c
blob: 2c5980d619edb6799d1dd11a488926389bd021a6 (plain) (blame)
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
/*
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "buffer.h"
#include "common.h"
#include "hwcontext.h"
#include "hwcontext_internal.h"
#include "hwcontext_cuda.h"
#include "mem.h"
#include "pixdesc.h"
#include "pixfmt.h"

typedef struct CUDAFramesContext {
    int shift_width, shift_height;
} CUDAFramesContext;

static const enum AVPixelFormat supported_formats[] = {
    AV_PIX_FMT_NV12,
    AV_PIX_FMT_YUV420P,
    AV_PIX_FMT_YUV444P,
};

static void cuda_buffer_free(void *opaque, uint8_t *data)
{
    AVHWFramesContext *ctx = opaque;
    AVCUDADeviceContext *hwctx = ctx->device_ctx->hwctx;

    CUcontext dummy;

    cuCtxPushCurrent(hwctx->cuda_ctx);

    cuMemFree((CUdeviceptr)data);

    cuCtxPopCurrent(&dummy);
}

static AVBufferRef *cuda_pool_alloc(void *opaque, int size)
{
    AVHWFramesContext     *ctx = opaque;
    AVCUDADeviceContext *hwctx = ctx->device_ctx->hwctx;

    AVBufferRef *ret = NULL;
    CUcontext dummy = NULL;
    CUdeviceptr data;
    CUresult err;

    err = cuCtxPushCurrent(hwctx->cuda_ctx);
    if (err != CUDA_SUCCESS) {
        av_log(ctx, AV_LOG_ERROR, "Error setting current CUDA context\n");
        return NULL;
    }

    err = cuMemAlloc(&data, size);
    if (err != CUDA_SUCCESS)
        goto fail;

    ret = av_buffer_create((uint8_t*)data, size, cuda_buffer_free, ctx, 0);
    if (!ret) {
        cuMemFree(data);
        goto fail;
    }

fail:
    cuCtxPopCurrent(&dummy);
    return ret;
}

static int cuda_frames_init(AVHWFramesContext *ctx)
{
    CUDAFramesContext *priv = ctx->internal->priv;
    int i;

    for (i = 0; i < FF_ARRAY_ELEMS(supported_formats); i++) {
        if (ctx->sw_format == supported_formats[i])
            break;
    }
    if (i == FF_ARRAY_ELEMS(supported_formats)) {
        av_log(ctx, AV_LOG_ERROR, "Pixel format '%s' is not supported\n",
               av_get_pix_fmt_name(ctx->sw_format));
        return AVERROR(ENOSYS);
    }

    av_pix_fmt_get_chroma_sub_sample(ctx->sw_format, &priv->shift_width, &priv->shift_height);

    if (!ctx->pool) {
        int size;

        switch (ctx->sw_format) {
        case AV_PIX_FMT_NV12:
        case AV_PIX_FMT_YUV420P:
            size = ctx->width * ctx->height * 3 / 2;
            break;
        case AV_PIX_FMT_YUV444P:
            size = ctx->width * ctx->height * 3;
            break;
        }

        ctx->internal->pool_internal = av_buffer_pool_init2(size, ctx, cuda_pool_alloc, NULL);
        if (!ctx->internal->pool_internal)
            return AVERROR(ENOMEM);
    }

    return 0;
}

static int cuda_get_buffer(AVHWFramesContext *ctx, AVFrame *frame)
{
    frame->buf[0] = av_buffer_pool_get(ctx->pool);
    if (!frame->buf[0])
        return AVERROR(ENOMEM);

    switch (ctx->sw_format) {
    case AV_PIX_FMT_NV12:
        frame->data[0]     = frame->buf[0]->data;
        frame->data[1]     = frame->data[0] + ctx->width * ctx->height;
        frame->linesize[0] = ctx->width;
        frame->linesize[1] = ctx->width;
        break;
    case AV_PIX_FMT_YUV420P:
        frame->data[0]     = frame->buf[0]->data;
        frame->data[2]     = frame->data[0] + ctx->width * ctx->height;
        frame->data[1]     = frame->data[2] + ctx->width * ctx->height / 4;
        frame->linesize[0] = ctx->width;
        frame->linesize[1] = ctx->width / 2;
        frame->linesize[2] = ctx->width / 2;
        break;
    case AV_PIX_FMT_YUV444P:
        frame->data[0]     = frame->buf[0]->data;
        frame->data[1]     = frame->data[0] + ctx->width * ctx->height;
        frame->data[2]     = frame->data[1] + ctx->width * ctx->height;
        frame->linesize[0] = ctx->width;
        frame->linesize[1] = ctx->width;
        frame->linesize[2] = ctx->width;
        break;
    default:
        av_frame_unref(frame);
        return AVERROR_BUG;
    }

    frame->format = AV_PIX_FMT_CUDA;
    frame->width  = ctx->width;
    frame->height = ctx->height;

    return 0;
}

static int cuda_transfer_get_formats(AVHWFramesContext *ctx,
                                     enum AVHWFrameTransferDirection dir,
                                     enum AVPixelFormat **formats)
{
    enum AVPixelFormat *fmts;

    fmts = av_malloc_array(2, sizeof(*fmts));
    if (!fmts)
        return AVERROR(ENOMEM);

    fmts[0] = ctx->sw_format;
    fmts[1] = AV_PIX_FMT_NONE;

    *formats = fmts;

    return 0;
}

static int cuda_transfer_data_from(AVHWFramesContext *ctx, AVFrame *dst,
                                   const AVFrame *src)
{
    CUDAFramesContext           *priv = ctx->internal->priv;
    AVCUDADeviceContext *device_hwctx = ctx->device_ctx->hwctx;

    CUcontext dummy;
    CUresult err;
    int i;

    err = cuCtxPushCurrent(device_hwctx->cuda_ctx);
    if (err != CUDA_SUCCESS)
        return AVERROR_UNKNOWN;

    for (i = 0; i < FF_ARRAY_ELEMS(src->data) && src->data[i]; i++) {
        CUDA_MEMCPY2D cpy = {
            .srcMemoryType = CU_MEMORYTYPE_DEVICE,
            .dstMemoryType = CU_MEMORYTYPE_HOST,
            .srcDevice     = (CUdeviceptr)src->data[i],
            .dstHost       = dst->data[i],
            .srcPitch      = src->linesize[i],
            .dstPitch      = dst->linesize[i],
            .WidthInBytes  = FFMIN(src->linesize[i], dst->linesize[i]),
            .Height        = src->height >> (i ? priv->shift_height : 0),
        };

        err = cuMemcpy2D(&cpy);
        if (err != CUDA_SUCCESS) {
            av_log(ctx, AV_LOG_ERROR, "Error transferring the data from the CUDA frame\n");
            return AVERROR_UNKNOWN;
        }
    }

    cuCtxPopCurrent(&dummy);

    return 0;
}

static int cuda_transfer_data_to(AVHWFramesContext *ctx, AVFrame *dst,
                                 const AVFrame *src)
{
    CUDAFramesContext           *priv = ctx->internal->priv;
    AVCUDADeviceContext *device_hwctx = ctx->device_ctx->hwctx;

    CUcontext dummy;
    CUresult err;
    int i;

    err = cuCtxPushCurrent(device_hwctx->cuda_ctx);
    if (err != CUDA_SUCCESS)
        return AVERROR_UNKNOWN;

    for (i = 0; i < FF_ARRAY_ELEMS(src->data) && src->data[i]; i++) {
        CUDA_MEMCPY2D cpy = {
            .srcMemoryType = CU_MEMORYTYPE_HOST,
            .dstMemoryType = CU_MEMORYTYPE_DEVICE,
            .srcHost       = src->data[i],
            .dstDevice     = (CUdeviceptr)dst->data[i],
            .srcPitch      = src->linesize[i],
            .dstPitch      = dst->linesize[i],
            .WidthInBytes  = FFMIN(src->linesize[i], dst->linesize[i]),
            .Height        = src->height >> (i ? priv->shift_height : 0),
        };

        err = cuMemcpy2D(&cpy);
        if (err != CUDA_SUCCESS) {
            av_log(ctx, AV_LOG_ERROR, "Error transferring the data from the CUDA frame\n");
            return AVERROR_UNKNOWN;
        }
    }

    cuCtxPopCurrent(&dummy);

    return 0;
}

const HWContextType ff_hwcontext_type_cuda = {
    .type                 = AV_HWDEVICE_TYPE_CUDA,
    .name                 = "CUDA",

    .device_hwctx_size    = sizeof(AVCUDADeviceContext),
    .frames_priv_size     = sizeof(CUDAFramesContext),

    .frames_init          = cuda_frames_init,
    .frames_get_buffer    = cuda_get_buffer,
    .transfer_get_formats = cuda_transfer_get_formats,
    .transfer_data_to     = cuda_transfer_data_to,
    .transfer_data_from   = cuda_transfer_data_from,

    .pix_fmts             = (const enum AVPixelFormat[]){ AV_PIX_FMT_CUDA, AV_PIX_FMT_NONE },
};