annotate libvo/vo_dxr3.c @ 4513:2e3800da1ceb

Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e (high quality divx movies that before ran very poor now plays perfectly). Also includes some minor fixes to the osd support. Since libmp1e has issues with non-mmx system I think this move is a smart one...
author mswitch
date Sun, 03 Feb 2002 14:55:27 +0000
parents 0919d2ec5e22
children 4e21ffb56104
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
1 /*
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
2 * vo_dxr3.c - DXR3/H+ video out
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
3 *
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
4 * Copyright (C) 2002 David Holm <dholm@iname.com>
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
5 *
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
6 */
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
7
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
8 /* ChangeLog added 2002-01-10
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
9 * 2002-02-03:
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
10 * Removal of libmp1e, libavcodec has finally become faster (and it's code is helluva lot cleaner)
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
11 * We will call this one the "gentoo" release in favour of the gentoo linux distribution (www.gentoo.org), which
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
12 * aims at providing maximum performance
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
13 *
4480
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
14 * 2002-02-02:
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
15 * Cleaned out some old code which might have slowed down writes
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
16 *
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
17 * 2002-01-17:
4219
70888f501b4a Fixed seeking
mswitch
parents: 4218
diff changeset
18 * Testrelease of new sync engine (using previously undocumented feature of em8300).
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
19 *
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
20 * 2002-01-15:
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
21 * Preliminary subpic support with -vc mpegpes and dvd's
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
22 * Device interfaces tries the new naming scheme by default (even though most users probably still use the old one)
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
23 *
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
24 * 2002-01-10:
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
25 * I rehauled the entire codebase. I have now changed to
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
26 * Kernighan & Ritchie codingstyle, please mail me if you
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
27 * find any inconcistencies.
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
28 */
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
29
2921
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
30 #include <linux/em8300.h>
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
31 #include <sys/ioctl.h>
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
32 #include <sys/stat.h>
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
33 #include <sys/types.h>
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
34 #include <unistd.h>
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
35 #include <stdio.h>
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
36 #include <stdlib.h>
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
37 #include <string.h>
2921
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
38 #include <fcntl.h>
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
39 #include <stdio.h>
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
40 #include <time.h>
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
41
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
42 #include "config.h"
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
43 #include "fastmemcpy.h"
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
44
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
45 #include "video_out.h"
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
46 #include "video_out_internal.h"
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
47 #include "../postproc/rgb2rgb.h"
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
48
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
49 #ifdef USE_LIBAVCODEC
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
50 #ifdef USE_LIBAVCODEC_SO
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
51 #include <libffmpeg/avcodec.h>
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
52 #else
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
53 #include "libavcodec/avcodec.h"
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
54 #endif
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
55 /* for video encoder */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
56 static AVCodec *avc_codec = NULL;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
57 static AVCodecContext *avc_context = NULL;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
58 static AVPicture avc_picture;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
59 char *picture_buf = NULL;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
60 char *avc_outbuf = NULL;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
61 int avc_outbuf_size = 100000;
3432
5cb7caab5536 MP1E configure patch by D. Holm.
atmos4
parents: 3387
diff changeset
62 #endif
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
63
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
64 #ifdef HAVE_MMX
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
65 #include "mmx.h"
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
66 #endif
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
67
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
68 LIBVO_EXTERN (dxr3)
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
69
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
70 /* Resolutions and positions */
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
71 static int v_width, v_height;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
72 static int s_width, s_height;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
73 static int s_pos_x, s_pos_y;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
74 static int d_pos_x, d_pos_y;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
75 static int osd_w, osd_h;
3329
529a001496c2 Switched from rte which caused alot of problems for some users to libmp1e.
mswitch
parents: 3232
diff changeset
76
2770
b9732eaa82d9 DXR3 beta2 patch by David Holm.
atmos4
parents: 2732
diff changeset
77 static int img_format = 0;
3329
529a001496c2 Switched from rte which caused alot of problems for some users to libmp1e.
mswitch
parents: 3232
diff changeset
78
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
79 /* File descriptors */
2921
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
80 static int fd_control = -1;
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
81 static int fd_video = -1;
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
82 static int fd_spu = -1;
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
83
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
84 /* Static variable used in ioctl's */
2921
ee37c7fc3022 dxr3-patch4 by David Holm
arpi
parents: 2866
diff changeset
85 static int ioval = 0;
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
86
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
87 static vo_info_t vo_info =
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
88 {
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
89 "DXR3/H+ video out",
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
90 "dxr3",
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
91 "David Holm <dholm@iname.com>",
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
92 ""
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
93 };
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
94
4433
df8e0f71cc3c new info for tuning
nick
parents: 4352
diff changeset
95 static uint32_t config(uint32_t scr_width, uint32_t scr_height, uint32_t width, uint32_t height, uint32_t fullscreen, char *title, uint32_t format,const vo_tune_info_t *info)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
96 {
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
97 int tmp1, tmp2;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
98 em8300_register_t reg;
3639
64ee21b3bd09 Modified the sync code once again, commented out hardware pts sync (I'll likely burn in hell before understanding how to get this bastard to sync well)
mswitch
parents: 3630
diff changeset
99
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
100 /* This activates the subpicture processor, you can safely disable this and still send */
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
101 /* broken subpics to the em8300, if it's enabled and you send broken subpics you will end */
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
102 /* up in a lockup */
4160
ac62f8534409 hw spu support for dxr3 - patch by David Holm
arpi
parents: 4109
diff changeset
103 ioval = EM8300_SPUMODE_ON;
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
104 if (ioctl(fd_control, EM8300_IOCTL_SET_SPUMODE, &ioval) < 0) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
105 printf("VO: [dxr3] Unable to set subpicture mode!\n");
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
106 uninit();
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
107 return -1;
3201
039a973b3dda Newest DXR3 patch (improved 1.0rc2) by David Holm.
atmos4
parents: 3186
diff changeset
108 }
039a973b3dda Newest DXR3 patch (improved 1.0rc2) by David Holm.
atmos4
parents: 3186
diff changeset
109
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
110 /* Set the playmode to play (just in case another app has set it to something else) */
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
111 ioval = EM8300_PLAYMODE_PLAY;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
112 if (ioctl(fd_control, EM8300_IOCTL_SET_PLAYMODE, &ioval) < 0) {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
113 printf("VO: [dxr3] Unable to set playmode!\n");
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
114 }
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
115
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
116 /* Start em8300 prebuffering and sync engine */
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
117 reg.microcode_register = 1;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
118 reg.reg = 0;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
119 reg.val = MVCOMMAND_SYNC;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
120 ioctl(fd_control, EM8300_IOCTL_WRITEREG, &reg);
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
121
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
122 /* Set the default playback speed to 0x900 */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
123 ioval = 0x900;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
124 ioctl(fd_control, EM8300_IOCTL_SCR_SETSPEED, &ioval);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
125
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
126 /* Flush the buffer and make sure it is clean by syncing it */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
127 ioval = EM8300_SUBDEVICE_VIDEO;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
128 ioctl(fd_control, EM8300_IOCTL_FLUSH, &ioval);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
129 fsync(fd_video);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
130 /* We'll be nice and flush the audio buffer as well */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
131 ioval = EM8300_SUBDEVICE_AUDIO;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
132 ioctl(fd_control, EM8300_IOCTL_FLUSH, &ioval);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
133
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
134 /* Store some variables statically that we need later in another scope */
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
135 img_format = format;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
136 v_width = scr_width;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
137 v_height = scr_height;
2866
4f6190ab52e7 Added a temporary fix to the DXR3 win32 codec playback, win32 codecs might prove to play back an unscaled image!
mswitch
parents: 2770
diff changeset
138
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
139 /* libmp1e requires a width and height that is x|16 */
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
140 s_width = (v_width + 15) / 16;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
141 s_width *= 16;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
142 s_height = (v_height + 15) / 16;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
143 s_height *= 16;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
144
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
145 /* Try to figure out whether to use widescreen output or not */
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
146 /* Anamorphic widescreen modes makes this a pain in the ass */
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
147 tmp1 = abs(height - ((width / 4) * 3));
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
148 tmp2 = abs(height - (int) (width / 2.35));
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
149 if (tmp1 < tmp2) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
150 ioval = EM8300_ASPECTRATIO_4_3;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
151 printf("VO: [dxr3] Setting aspect ratio to 4:3\n");
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
152 } else {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
153 ioval = EM8300_ASPECTRATIO_16_9;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
154 printf("VO: [dxr3] Setting aspect ratio to 16:9\n");
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
155 }
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
156 ioctl(fd_control, EM8300_IOCTL_SET_ASPECTRATIO, &ioval);
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
157
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
158 if (format != IMGFMT_MPEGPES) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
159 #ifdef USE_LIBAVCODEC
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
160 int size;
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
161
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
162 avc_codec = avcodec_find_encoder(CODEC_ID_MPEG1VIDEO);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
163 if (!avc_codec) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
164 printf("VO: [dxr3] Unable to find mpeg1video codec\n");
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
165 uninit();
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
166 return -1;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
167 }
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
168 avc_context = malloc(sizeof(AVCodecContext));
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
169 memset(avc_context, 0, sizeof(avc_context));
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
170 avc_context->width = v_width;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
171 avc_context->height = v_height;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
172 avc_context->frame_rate = 30 * FRAME_RATE_BASE;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
173 avc_context->gop_size = 0;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
174 avc_context->bit_rate = 8e6;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
175 avc_context->flags = CODEC_FLAG_HQ;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
176 avc_context->pix_fmt = PIX_FMT_YUV420P;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
177 if (avcodec_open(avc_context, avc_codec) < 0) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
178 printf("VO: [dxr3] Unable to open codec\n");
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
179 uninit();
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
180 return -1;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
181 }
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
182
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
183 /* This stuff calculations the relative position of video and osd on screen */
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
184 /* Old stuff taken from the dvb driver, should be removed when introducing spuenc */
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
185 osd_w = s_width;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
186 d_pos_x = (s_width - v_width) / 2;
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
187 if (d_pos_x < 0) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
188 s_pos_x = -d_pos_x;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
189 d_pos_x = 0;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
190 osd_w = s_width;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
191 } else {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
192 s_pos_x = 0;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
193 }
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
194 osd_h = s_height;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
195 d_pos_y = (s_height - v_height) / 2;
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
196 if (d_pos_y < 0) {
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
197 s_pos_y = -d_pos_y;
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
198 d_pos_y = 0;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
199 osd_h = s_height;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
200 } else {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
201 s_pos_y = 0;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
202 }
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
203
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
204 /* Create a pixel buffer and set up pointers for color components */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
205 memset(&avc_picture, 0, sizeof(avc_picture));
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
206 avc_picture.linesize[0] = v_width;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
207 avc_picture.linesize[1] = v_width / 2;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
208 avc_picture.linesize[2] = v_width / 2;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
209 avc_outbuf = malloc(avc_outbuf_size);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
210
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
211 size = s_width * s_height;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
212 picture_buf = malloc((size * 3) / 2);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
213 avc_picture.data[0] = picture_buf;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
214 avc_picture.data[1] = avc_picture.data[0] + size;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
215 avc_picture.data[2] = avc_picture.data[1] + size / 4;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
216 if (format == IMGFMT_BGR24) {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
217 yuv2rgb_init(24, MODE_BGR);
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
218 } else {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
219 yuv2rgb_init(24, MODE_RGB);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
220 }
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
221 return 0;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
222 #endif
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
223 return -1;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
224 } else if (format == IMGFMT_MPEGPES) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
225 printf("VO: [dxr3] Format: MPEG-PES (no conversion needed)\n");
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
226 return 0;
3201
039a973b3dda Newest DXR3 patch (improved 1.0rc2) by David Holm.
atmos4
parents: 3186
diff changeset
227 }
039a973b3dda Newest DXR3 patch (improved 1.0rc2) by David Holm.
atmos4
parents: 3186
diff changeset
228
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
229 printf("VO: [dxr3] Format: Unsupported\n");
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
230 uninit();
3432
5cb7caab5536 MP1E configure patch by D. Holm.
atmos4
parents: 3387
diff changeset
231 return -1;
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
232 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
233
2866
4f6190ab52e7 Added a temporary fix to the DXR3 win32 codec playback, win32 codecs might prove to play back an unscaled image!
mswitch
parents: 2770
diff changeset
234 static const vo_info_t* get_info(void)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
235 {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
236 return &vo_info;
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
237 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
238
2866
4f6190ab52e7 Added a temporary fix to the DXR3 win32 codec playback, win32 codecs might prove to play back an unscaled image!
mswitch
parents: 2770
diff changeset
239 static void draw_alpha(int x0, int y0, int w, int h, unsigned char* src, unsigned char *srca, int srcstride)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
240 {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
241 #ifdef USE_LIBAVCODEC
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
242 /* This function draws the osd and subtitles etc. It will change to use spuenc soon */
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
243 switch (img_format) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
244 case IMGFMT_BGR24:
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
245 case IMGFMT_YV12:
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
246 vo_draw_alpha_yv12(w, h, src, srca, srcstride,
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
247 avc_picture.data[0] + (x0 + d_pos_x) + (y0 + d_pos_y) * avc_picture.linesize[0], avc_picture.linesize[0]);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
248 break;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
249 case IMGFMT_YUY2:
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
250 vo_draw_alpha_yuy2(w, h, src, srca, srcstride,
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
251 avc_picture.data[0] + (x0 + d_pos_x) * 2 + (y0 + d_pos_y) * avc_picture.linesize[0], avc_picture.linesize[0]);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
252 break;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
253 }
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
254 #endif
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
255 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
256
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
257 static void draw_osd(void)
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
258 {
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
259 if (img_format != IMGFMT_MPEGPES) {
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
260 vo_draw_text(osd_w, osd_h, draw_alpha);
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
261 }
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
262 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
263
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
264 static uint32_t draw_frame(uint8_t * src[])
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
265 {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
266 if (img_format == IMGFMT_MPEGPES) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
267 vo_mpegpes_t *p = (vo_mpegpes_t *) src[0];
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
268
4160
ac62f8534409 hw spu support for dxr3 - patch by David Holm
arpi
parents: 4109
diff changeset
269 if (p->id == 0x20) {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
270 ioval = vo_pts / 2;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
271 ioctl(fd_spu, EM8300_IOCTL_SPU_SETPTS, &ioval);
4480
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
272 write(fd_spu, p->data, p->size);
4160
ac62f8534409 hw spu support for dxr3 - patch by David Holm
arpi
parents: 4109
diff changeset
273 } else {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
274 ioval = vo_pts / 2;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
275 ioctl(fd_video, EM8300_IOCTL_VIDEO_SETPTS, &ioval);
4480
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
276 write(fd_video, p->data, p->size);
4160
ac62f8534409 hw spu support for dxr3 - patch by David Holm
arpi
parents: 4109
diff changeset
277 }
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
278 return 0;
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
279 #ifdef USE_LIBAVCODEC
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
280 } else {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
281 int size = s_width * s_height;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
282 if (img_format == IMGFMT_YUY2) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
283 yuy2toyv12(src[0], avc_picture.data[0], avc_picture.data[1], avc_picture.data[2],
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
284 v_width, v_height, avc_picture.linesize[0], avc_picture.linesize[1], v_width*2);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
285 } else if (img_format == IMGFMT_BGR24) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
286 rgb24toyv12(src[0], avc_picture.data[0], avc_picture.data[1], avc_picture.data[2],
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
287 v_width, v_height, avc_picture.linesize[0], avc_picture.linesize[1], v_width*3);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
288 }
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
289 size = avcodec_encode_video(avc_context, avc_outbuf, avc_outbuf_size, &avc_picture);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
290 write(fd_video, avc_outbuf, size);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
291 return 0;
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
292 #endif
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
293 }
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
294 return -1;
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
295 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
296
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
297 static void flip_page(void)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
298 {
4219
70888f501b4a Fixed seeking
mswitch
parents: 4218
diff changeset
299 /* Flush the device if a seek occured */
4480
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
300 if (!vo_pts) {
4269
68c78efeea2b Reverted to using software sync again, some day I'll die of a heartattack if I can't get this
mswitch
parents: 4267
diff changeset
301 /* Flush video */
4219
70888f501b4a Fixed seeking
mswitch
parents: 4218
diff changeset
302 ioval = EM8300_SUBDEVICE_VIDEO;
70888f501b4a Fixed seeking
mswitch
parents: 4218
diff changeset
303 ioctl(fd_control, EM8300_IOCTL_FLUSH, &ioval);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
304 }
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
305 #ifdef USE_LIBAVCODEC
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
306 if (img_format == IMGFMT_YV12) {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
307 int out_size = avcodec_encode_video(avc_context, avc_outbuf, avc_outbuf_size, &avc_picture);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
308 write(fd_video, avc_outbuf, out_size);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
309 }
3432
5cb7caab5536 MP1E configure patch by D. Holm.
atmos4
parents: 3387
diff changeset
310 #endif
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
311 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
312
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
313 static uint32_t draw_slice(uint8_t *srcimg[], int stride[], int w, int h, int x0, int y0)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
314 {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
315 #ifdef USE_LIBAVCODEC
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
316 if (img_format == IMGFMT_YV12) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
317 int y;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
318 unsigned char *s, *s1;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
319 unsigned char *d, *d1;
3232
d037e1201721 Added support for codecs that supports BGR24 (some opensource codecs and vivo)
mswitch
parents: 3218
diff changeset
320
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
321 x0 += d_pos_x;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
322 y0 += d_pos_y;
3218
8ba06b63f873 Fix green borders -> black borders, patch from D. Holm, also small fix to dxr3 ao.
atmos4
parents: 3208
diff changeset
323
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
324 if ((x0 + w) > avc_picture.linesize[0]) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
325 w = avc_picture.linesize[0] - x0;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
326 }
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
327 if ((y0 + h) > s_height) {
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
328 h = s_height - y0;
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
329 }
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
330
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
331 s = srcimg[0] + s_pos_x + s_pos_y * stride[0];
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
332 d = avc_picture.data[0] + x0 + y0 * avc_picture.linesize[0];
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
333 for(y = 0; y < h; y++) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
334 memcpy(d, s, w);
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
335 s += stride[0];
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
336 d += avc_picture.linesize[0];
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
337 }
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
338
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
339 w /= 2;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
340 h /= 2;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
341 x0 /= 2;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
342 y0 /= 2;
3218
8ba06b63f873 Fix green borders -> black borders, patch from D. Holm, also small fix to dxr3 ao.
atmos4
parents: 3208
diff changeset
343
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
344 s = srcimg[1] + s_pos_x + (s_pos_y * stride[1]);
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
345 d = avc_picture.data[1] + x0 + (y0 * avc_picture.linesize[1]);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
346 s1 = srcimg[2] + s_pos_x + (s_pos_y * stride[2]);
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
347 d1 = avc_picture.data[2] + x0 + (y0 * avc_picture.linesize[2]);
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
348 for(y = 0; y < h; y++) {
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
349 memcpy(d, s, w);
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
350 memcpy(d1, s1, w);
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
351 s += stride[1];
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
352 s1 += stride[2];
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
353 d += avc_picture.linesize[1];
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
354 d1 += avc_picture.linesize[2];
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
355 }
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
356 return 0;
3232
d037e1201721 Added support for codecs that supports BGR24 (some opensource codecs and vivo)
mswitch
parents: 3218
diff changeset
357 }
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
358 #endif
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
359 return -1;
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
360 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
361
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
362 static uint32_t query_format(uint32_t format)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
363 {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
364 uint32_t flag = 0;
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
365
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
366 if (format == IMGFMT_MPEGPES) {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
367 /* Hardware accelerated | Hardware supports subpics */
4269
68c78efeea2b Reverted to using software sync again, some day I'll die of a heartattack if I can't get this
mswitch
parents: 4267
diff changeset
368 flag = 0x2 | 0x8;
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
369 #ifdef USE_LIBAVCODEC
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
370 } else if (format == IMGFMT_YV12) {
4267
90562e9c36d6 Only use hardware sync and prebuffering with mpegs!
mswitch
parents: 4265
diff changeset
371 /* Conversion needed | OSD Supported */
90562e9c36d6 Only use hardware sync and prebuffering with mpegs!
mswitch
parents: 4265
diff changeset
372 flag = 0x1 | 0x4;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
373 } else if (format == IMGFMT_YUY2) {
4267
90562e9c36d6 Only use hardware sync and prebuffering with mpegs!
mswitch
parents: 4265
diff changeset
374 /* Conversion needed | OSD Supported */
90562e9c36d6 Only use hardware sync and prebuffering with mpegs!
mswitch
parents: 4265
diff changeset
375 flag = 0x1 | 0x4;
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
376 } else if (format == IMGFMT_RGB24) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
377 /* Conversion needed | OSD Supported */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
378 flag = 0x1 | 0x4;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
379 } else if (format == IMGFMT_BGR24) {
4267
90562e9c36d6 Only use hardware sync and prebuffering with mpegs!
mswitch
parents: 4265
diff changeset
380 /* Conversion needed | OSD Supported */
90562e9c36d6 Only use hardware sync and prebuffering with mpegs!
mswitch
parents: 4265
diff changeset
381 flag = 0x1 | 0x4;
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
382 } else {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
383 printf("VO: [dxr3] Format unsupported, mail dholm@iname.com\n");
3432
5cb7caab5536 MP1E configure patch by D. Holm.
atmos4
parents: 3387
diff changeset
384 #else
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
385 } else {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
386 printf("VO: [dxr3] You have enable libavcodec support (Read DOCS/codecs.html)!\n");
3432
5cb7caab5536 MP1E configure patch by D. Holm.
atmos4
parents: 3387
diff changeset
387 #endif
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
388 }
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
389 return flag;
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
390 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
391
3208
98a587aaf505 Support for YV12 with DXR3, patch by D. Holm.
atmos4
parents: 3201
diff changeset
392 static void uninit(void)
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
393 {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
394 printf("VO: [dxr3] Uninitializing\n");
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
395 #ifdef USE_LIBAVCODEC
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
396 if (avc_context) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
397 avcodec_close(avc_context);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
398 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
399 if (picture_buf) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
400 free(picture_buf);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
401 }
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
402 #endif
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
403 if (fd_video) {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
404 close(fd_video);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
405 }
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
406 if (fd_spu) {
4095
a3f3858b8c15 useless cosmetics changes... a.k.a. cvslog killer - patch by David Holm
arpi
parents: 4063
diff changeset
407 close(fd_spu);
4218
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
408 }
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
409 if (fd_control) {
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
410 close(fd_control);
3931c41f740a Added new syncengine thanks to a new previously undocumented feature of the em8300, this might fix playback on both slow and fast machines (more testing needed). This also requires users to get the em8300 driver from cvs until the next version is released (will probably happen this weekend)
mswitch
parents: 4166
diff changeset
411 }
2645
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
412 }
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
413
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
414 static void check_events(void)
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
415 {
9d8d69fc517c ao_dxr3 by David Holm <dholm@telia.com>
arpi
parents:
diff changeset
416 }
4352
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
417
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
418 static uint32_t preinit(const char *arg)
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
419 {
4513
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
420 char devname[80];
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
421 int fdflags = O_WRONLY | O_NONBLOCK;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
422
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
423 /* Open the control interface */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
424 if (vo_subdevice) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
425 sprintf(devname, "/dev/em8300-%s", vo_subdevice);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
426 } else {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
427 /* Try new naming scheme by default */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
428 sprintf(devname, "/dev/em8300-0");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
429 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
430 fd_control = open(devname, fdflags);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
431 if (fd_control < 1) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
432 /* Fall back to old naming scheme */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
433 printf("VO: [dxr3] Error opening %s for writing, trying /dev/em8300 instead\n", devname);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
434 sprintf(devname, "/dev/em8300");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
435 fd_control = open(devname, fdflags);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
436 if (fd_control < 1) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
437 printf("VO: [dxr3] Error opening /dev/em8300 for writing as well!\nBailing\n");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
438 return -1;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
439 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
440 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
441
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
442 /* Open the video interface */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
443 if (vo_subdevice) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
444 sprintf(devname, "/dev/em8300_mv-%s", vo_subdevice);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
445 } else {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
446 /* Try new naming scheme by default */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
447 sprintf(devname, "/dev/em8300_mv-0");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
448 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
449 fd_video = open(devname, fdflags);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
450 if (fd_video < 0) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
451 /* Fall back to old naming scheme */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
452 printf("VO: [dxr3] Error opening %s for writing, trying /dev/em8300_mv instead\n", devname);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
453 sprintf(devname, "/dev/em8300_mv");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
454 fd_video = open(devname, fdflags);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
455 if (fd_video < 0) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
456 printf("VO: [dxr3] Error opening /dev/em8300_mv for writing as well!\nBailing\n");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
457 uninit();
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
458 return -1;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
459 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
460 } else {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
461 printf("VO: [dxr3] Opened %s\n", devname);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
462 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
463
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
464 /* Open the subpicture interface */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
465 if (vo_subdevice) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
466 sprintf(devname, "/dev/em8300_sp-%s", vo_subdevice);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
467 } else {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
468 /* Try new naming scheme by default */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
469 sprintf(devname, "/dev/em8300_sp-0");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
470 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
471 fd_spu = open(devname, fdflags);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
472 if (fd_spu < 0) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
473 /* Fall back to old naming scheme */
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
474 printf("VO: [dxr3] Error opening %s for writing, trying /dev/em8300_sp instead\n", devname);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
475 sprintf(devname, "/dev/em8300_sp");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
476 fd_spu = open(devname, fdflags);
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
477 if (fd_spu < 0) {
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
478 printf("VO: [dxr3] Error opening /dev/em8300_sp for writing as well!\nBailing\n");
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
479 uninit();
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
480 return -1;
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
481 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
482 }
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
483
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
484 #ifdef USE_LIBAVCODEC
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
485 avcodec_init();
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
486 avcodec_register_all();
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
487 #endif
2e3800da1ceb Switched from libmp1e to libavcodec, at least for me it runs helluva lot faster than libmp1e
mswitch
parents: 4480
diff changeset
488
4480
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
489 return 0;
4352
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
490 }
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
491
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
492 static void query_vaa(vo_vaa_t *vaa)
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
493 {
4480
0919d2ec5e22 Cleaned up some code
mswitch
parents: 4433
diff changeset
494 memset(vaa, 0, sizeof(vo_vaa_t));
4352
ed5b85b713a3 Extensions for video accelerated architecture
nick
parents: 4269
diff changeset
495 }