October 2013 - MPlayer-dev-eng - lists.mplayerhq.hu

Homepage skin download issue
by Ingo Brückl 09 Sep '15

09 Sep '15

Yesterday, I updated hwswskin to 1.3, but it still isn't downloadable. Is the generation of the skin download tarballs working? Ingo

3 8

Homepage DOCS needs update
by Ingo Brückl 24 Apr '14

24 Apr '14

If it's already on the to-do list ignore, but the documentation on the homepage doesn't get updated. Ingo

3 12

[PATCH] xvmc: handle shm xvimage allocation error
by Marcin Slusarz 06 Jan '14

06 Jan '14

- handle xvimage/shm allocation failure and fallback to non-shm xv - assert xvimage->data_size!=0 because 0 size will lead to memory corruptions (I hit it because of bug in Nouveau's early xvmc implementation) Index: libvo/vo_xvmc.c =================================================================== --- libvo/vo_xvmc.c (wersja 34098) +++ libvo/vo_xvmc.c (kopia robocza) @@ -181,6 +181,15 @@ return cur_render->state || (cur_render->mpi && cur_render->mpi->usage_count); } +static void allocate_xvimage_non_shm(int xvimage_width, int xvimage_height, int xv_format) +{ + xvimage = (XvImage *) XvCreateImage(mDisplay, xv_port, xv_format, NULL, + xvimage_width, xvimage_height); + assert(xvimage->data_size); + xvimage->data = malloc(xvimage->data_size); + XSync(mDisplay,False); +} + static void allocate_xvimage(int xvimage_width,int xvimage_height,int xv_format) { /* @@ -188,7 +197,8 @@ * mit-shm this will bomb... trzing to fix ::atmos */ #ifdef HAVE_SHM - if ( mLocalDisplay && XShmQueryExtension( mDisplay ) ) Shmem_Flag = 1; + if ( mLocalDisplay && XShmQueryExtension( mDisplay ) ) + Shmem_Flag = 0; else { Shmem_Flag = 0; @@ -198,25 +208,41 @@ { xvimage = (XvImage *) XvShmCreateImage(mDisplay, xv_port, xv_format, NULL, xvimage_width, xvimage_height, &Shminfo); + if (!xvimage) + goto noshmimage; + assert(xvimage->data_size); Shminfo.shmid = shmget(IPC_PRIVATE, xvimage->data_size, IPC_CREAT | 0777); + if (Shminfo.shmid == -1) + goto shmgetfail; + Shminfo.shmaddr = (char *) shmat(Shminfo.shmid, 0, 0); + if ((long)Shminfo.shmaddr == -1) + goto shmatfail; + Shminfo.readOnly = False; xvimage->data = Shminfo.shmaddr; - XShmAttach(mDisplay, &Shminfo); + if (!XShmAttach(mDisplay, &Shminfo)) + goto shmattachfail; + XSync(mDisplay, False); shmctl(Shminfo.shmid, IPC_RMID, 0); } else #endif - { - xvimage = (XvImage *) XvCreateImage(mDisplay, xv_port, xv_format, NULL, xvimage_width, xvimage_height); - xvimage->data = malloc(xvimage->data_size); - XSync(mDisplay,False); - } -// memset(xvimage->data,128,xvimage->data_size); + allocate_xvimage_non_shm(xvimage_width, xvimage_height, xv_format); return; + +shmattachfail: + shmdt(Shminfo.shmaddr); +shmatfail: + shmctl(Shminfo.shmid, IPC_RMID, 0); +shmgetfail: + XFree(xvimage); +noshmimage: + Shmem_Flag = 0; + allocate_xvimage_non_shm(xvimage_width, xvimage_height, xv_format); } static void deallocate_xvimage(void)

4 6

patch: Make stdout unbuffered in slave mode
by Conrad Wood 04 Nov '13

04 Nov '13

attached Patch sets the standard output to non-buffered, if mplayer is in 'slave' mode. This has the benefit that a reader connected to mplayer via pipe is a lot more responsive (feels faster for the user) since the mplayer output is passed into the pipe (and thus the UI) immediately. I assumed that this is the case almost always when mplayer is invoked with '-slave' mode. The Patch adds these lines to mplayers main() function: if (slave_mode) { if ( setvbuf(stdout,NULL,_IONBF,0) ==0) { printf("Buffering disabled.\n"); } else { printf("Attempt to disable buffering failed: %s \n",strerror(errno)); } } I consider it safe on linux systems, but not sure about windows/mac/other systems. I would appreciate if it is applied to current svn as it would make my usecase much, much better ;) Thank you, Conrad

3 6

sharedbuffer vo?
by Kevin DeKorte 30 Oct '13

30 Oct '13

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Hi, I was noticing that in the corevideo code there is a shared buffer option. That made me wonder if a generic shared buffer vo would be useful. The reason I am asking this is that as display servers move to wayland we lose the -wid option in mplayer (or at least we appear to be at the moment). So how will applications like gnome-mplayer or smplayer work on wayland without X? If we were to use something like the shared buffer perhaps that would work on newer display servers without reverting to the X11 layer. Thanks, Kevin - -- Get my public GnuPG key from http://pgp.mit.edu:11371/pks/lookup?op=get&search=0x7D0BD5D1 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.15 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iEYEARECAAYFAlJxWZUACgkQ6w2kMH0L1dFL7wCbBf0P1TRyEZQVhVxHm5AArzW2 ZAIAnihSy0574LBIZDPZ2qsX1x085zYH =JUgG -----END PGP SIGNATURE-----

4 4

[PATCH] Fix build error on OSX 10.9
by Johan Andersson 30 Oct '13

30 Oct '13

Hello With latest OSX 10.9 Mavericks mplayer fail to compile becuase of: libvo/vo_corevideo.m:472:11: error: use of undeclared identifier 'GL_BLEND' glEnable(GL_BLEND); ^ libvo/vo_corevideo.m:473:2: warning: implicit declaration of function 'glDisable' is invalid in C99 [-Wimplicit-function-declaration] glDisable(GL_DEPTH_TEST); ^ libvo/vo_corevideo.m:473:12: error: use of undeclared identifier 'GL_DEPTH_TEST' glDisable(GL_DEPTH_TEST); Here is a patch to fix it cheers -- Johan Andersson

2 3

[PATCH] Add NEON optimizations to some critical audio functions.
by Reimar Döffinger 26 Oct '13

26 Oct '13

One big issue is that lrintf is ridiculously slow on most (all?) ARM Linux distributions, which makes the format conversion take more time than the audio decoding. It uses intrinsics because I was too lazy to learn the inline asm syntax and for these trivial cases gcc doesn't seem to be able to mess it up. In theory it also should allow the compiler to do loop unrolling when it makes sense. --- configure | 17 +++++++++++++++++ libaf/af_format.c | 21 ++++++++++++++++++++- libmpcodecs/ad_ffmpeg.c | 33 ++++++++++++++++++++++++++++++++- 3 files changed, 69 insertions(+), 2 deletions(-) diff --git a/configure b/configure index b464adf..b1b5e12 100755 --- a/configure +++ b/configure @@ -578,6 +578,7 @@ Advanced options: --enable-armvfp enable ARM VFP (ARM) [autodetect] --enable-vfpv3 enable ARM VFPV3 (ARM) [autodetect] --enable-neon enable NEON (ARM) [autodetect] + --enable-neon-intrin enable NEON intrinsics (ARM) [autodetect] --enable-thumb enable THUMB (ARM) [autodetect] --enable-iwmmxt enable iWMMXt (ARM) [autodetect] --disable-fastmemcpy disable 3DNow!/SSE/MMX optimized memcpy [enable] @@ -636,6 +637,7 @@ _armv6t2=auto _armvfp=auto vfpv3=auto neon=auto +neon_intrin=auto armthumb=auto _iwmmxt=auto _mtrr=auto @@ -1439,6 +1441,8 @@ for ac_option do --disable-vfpv3) vfpv3=no ;; --enable-neon) neon=yes ;; --disable-neon) neon=no ;; + --enable-neon-intrin) neon_intrin=yes ;; + --disable-neon-intrin) neon_intrin=no ;; --enable-thumb) armthumb=yes ;; --disable-thumb) armthumb=no ;; --enable-iwmmxt) _iwmmxt=yes ;; @@ -3105,6 +3109,18 @@ if arm ; then fi echores "$neon" + echocheck "ARM NEON intrinsics" + if test $neon_intrin = "auto" ; then + neon_intrin=no + statement_check 'arm_neon.h' 'float in[4] = {0}; float32x4_t tmpf = vld1q_f32(in);' && neon_intrin=yes + fi + echores "$neon_intrin" + if test $neon_intrin = "yes" ; then + def_neon_intrin='#define NEON_INTRIN 1' + else + def_neon_intrin='#define NEON_INTRIN 0' + fi + echocheck "ARM THUMB" if test $armthumb = "auto" ; then armthumb=no @@ -8716,6 +8732,7 @@ $def_altivec_h $def_malloc_h $def_mman_h $def_mman_has_map_failed +$def_neon_intrin $def_soundcard_h $def_sys_soundcard_h $def_sys_sysinfo_h diff --git a/libaf/af_format.c b/libaf/af_format.c index 797b011..38f1642 100644 --- a/libaf/af_format.c +++ b/libaf/af_format.c @@ -20,14 +20,18 @@ * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. */ +#include "config.h" + #include <stdio.h> #include <stdlib.h> #include <string.h> #include <inttypes.h> #include <limits.h> #include <math.h> +#if NEON_INTRIN +#include <arm_neon.h> +#endif -#include "config.h" #include "af.h" #include "mp_msg.h" #include "mpbswap.h" @@ -494,8 +498,23 @@ static void float2int(const float* in, void* out, int len, int bps) ((int8_t *)out)[i] = av_clip_int8(lrintf(128.0f * in[i])); break; case(2): +#if NEON_INTRIN + for (i = 0; i + 4 <= len; i += 4) { + float32x4_t tmpf = vld1q_f32(in + i); + int32x4_t tmp32 = vcvtq_n_s32_f32(tmpf, 16); + int16x4_t tmp16 = vqmovn_s32(tmp32); + vst1_s16((int16_t *)out + i, tmp16); + } + for (; i < len; i++) { + float32x4_t tmpf = vld1q_dup_f32(in + i); + int32x4_t tmp32 = vcvtq_n_s32_f32(tmpf, 16); + int16x4_t tmp16 = vqmovn_s32(tmp32); + vst1_lane_s16((int16_t *)out + i, tmp16, 0); + } +#else for(i=0;i<len;i++) ((int16_t*)out)[i] = av_clip_int16(lrintf(32768.0f * in[i])); +#endif break; case(3): for(i=0;i<len;i++){ diff --git a/libmpcodecs/ad_ffmpeg.c b/libmpcodecs/ad_ffmpeg.c index 11c502a..dcf9661 100644 --- a/libmpcodecs/ad_ffmpeg.c +++ b/libmpcodecs/ad_ffmpeg.c @@ -17,11 +17,15 @@ */ #define AVCODEC_MAX_AUDIO_FRAME_SIZE 192000 +#include "config.h" + #include <stdio.h> #include <stdlib.h> #include <unistd.h> +#if NEON_INTRIN +#include <arm_neon.h> +#endif -#include "config.h" #include "mp_msg.h" #include "help_mp.h" @@ -220,6 +224,33 @@ static av_always_inline void copy_samples_planar(size_t bps, { size_t s, c, o = 0; +#if NEON_INTRIN + if (nb_channels == 2 && bps == 4) { + for (s = 0; s + 4 <= nb_samples; s += 4) { + const uint32_t *src0 = (const uint32_t *)(src[0] + o); + const uint32_t *src1 = (const uint32_t *)(src[1] + o); + uint32x4x2_t tmp; + tmp.val[0] = vld1q_u32(src0); + tmp.val[1] = vld1q_u32(src1); + vst2q_u32((uint32_t *)dst, tmp); + o += 4*bps; + dst += 4*bps*nb_channels; + } + nb_samples -= s; + } else if (nb_channels == 2 && bps == 2) { + for (s = 0; s + 8 <= nb_samples; s += 8) { + const uint16_t *src0 = (const uint16_t *)(src[0] + o); + const uint16_t *src1 = (const uint16_t *)(src[1] + o); + uint16x8x2_t tmp; + tmp.val[0] = vld1q_u16(src0); + tmp.val[1] = vld1q_u16(src1); + vst2q_u16((uint16_t *)dst, tmp); + o += 8*bps; + dst += 8*bps*nb_channels; + } + nb_samples -= s; + } +#endif for (s = 0; s < nb_samples; s++) { for (c = 0; c < nb_channels; c++) { memcpy(dst, src[c] + o, bps); -- 1.8.4.rc3

2 4

minor build errors
by Ingo Brückl 23 Oct '13

23 Oct '13

* Forward from area mplayer-user * Originally on: Mon, 21 Oct 2013 03:09:26 -0400 * Originally by: Robert Henney <robh(a)rut.org> * Originally to: mplayer-users(a)mplayerhq.hu * Original subj: [MPlayer-users] minor build errors noticed some shell errors fly by in the build today. (have no idea if they're new) didn't seem to stop the build from completing though. clipped out section from output is below. (ie. "/bin/sh: 1: -wN: not found") CC libswscale/utils.o YASM libswscale/x86/input.o STRIP libswscale/x86/input.o /bin/sh: 1: -wN: not found [...] * End of Forward Should we define a Makefile variable STRIP? (It's actually rather a FFmpeg bug though. See the user list there.) Ingo

2 4

New redmine bugtracker - feedbacks ?
by Nicolas Coiffier 22 Oct '13

22 Oct '13

Hi everyone !! Almost a month ago, there was a thread about moving the actual mplayer bugtracker to a new location. ( http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/2013-July/071853.html) Since then, I've contact Moritz, and I've tried to migrate the actual bugtracker to a new one. I choose redmine (http://www.redmine.org/) because it was easier to install and configure than trac, but with the same kind of features. An old version (from the backup of August 7th) is available here, on my personnal server: http://redmine.ncoif.net I managed to import everything, and I did my best for the configuration. But during the import, I had to reset all the passwords, so you will have to click on the link "lost password" to create a new one. But the usernames are still the same. And since it's my personnal server, I don't have any ssl certificate, so I generated a self-signed one, which can create a warning in your browser. One last thing: it is possible to link bugs to the source code, usng SCM subversion, but I didn't had time to configure it yet, I plan to do it as soon as possible. So if you like it, and that it doesn't need too much load, then I can host the new bugtracker on my server. Any feedback is wellcome !! Nicolas

6 22

Blurry artefacts during playblack
by Ingo Brückl 21 Oct '13

21 Oct '13

After updating FFmpeg today, I'm experiencing a huge amount of blurry artefacts during playback with scenes where there is some motion in it (as if the difference information from the last frame was corrupt). Before I'm going to the trouble of bisecting it down, has anyone an idea what might cause this (since the FFmpeg approx. one month ago)? Ingo

3 3