July 2021 - MPlayer-dev-eng - lists.mplayerhq.hu

[PATCH] add initial RISC-V support
by Brad Smith 27 Feb '22

27 Feb '22

OpenBSD only supports riscv64 but this is an attempt at adding some of the initial bits for RISC-V support. Index: configure =================================================================== --- configure (revision 38313) +++ configure (working copy) @@ -289,6 +289,13 @@ esac } +riscv() { + case "$host_arch" in + riscv*) return 0;; + *) return 1;; + esac +} + # Use this before starting a check echocheck() { echo "============ Checking for $@ ============" >> "$TMPLOG" @@ -1991,7 +1998,10 @@ if arm ; then cc_check && host_arch=aarch64 || host_arch=arm fi +if riscv ; then + cc_check && host_arch=riscv64 || host_arch=riscv32 fi +fi echo "Detected operating system: $system_name" echo "Detected host architecture: $host_arch" @@ -2836,6 +2846,15 @@ iproc='arc' ;; + riscv*) + arch='riscv' + iproc='riscv' + + if test "$host_arch" = "riscv64" ; then + def_fast_64bit='#define HAVE_FAST_64BIT 1' + fi + ;; + *) echo "The architecture of your CPU ($host_arch) is not supported by this configure script" echo "It seems nobody has ported MPlayer to your OS or CPU type yet."

3 2

[PATCH] configure: Do not try to detect GnuTLS if OpenSSL was enabled
by Alexander Strasser 24 Feb '22

24 Feb '22

Enabling GnuTLS via auto detection could have us end up with both libraries enabled. This won't build anymore. IIRC at some point it did build, though it was not easy to figure out which implementation would be used in the end. This change is done in the light of having no auto detection for OpenSSL and probably never will, because of the OpenSSL license. --- Also Cc to mplayer-users. I hoped the original reporter of the build problem would test it, but it seems it didn't happen. So now I am posting it for review. This needs some more testing, I had this in my tree for a while, but forgot about testing it further, so I will only push after a bit more testing. Probably in a week or so. As explained in the commit message this change is assuming quite a bit. I think the risk of OpenSSL auto detection being added in the future is very low, so I feel it is worth it, because of the lower complexity and smaller patch. Comments welcome! configure | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/configure b/configure index d21df0ff7..91b6f9534 100755 --- a/configure +++ b/configure @@ -4531,7 +4531,7 @@ fi echocheck "GnuTLS" -if test "$_gnutls" = auto ; then +if test "$_gnutls" = auto && test "$_openssl" != "yes" ; then _gnutls=no if $_pkg_config --exists gnutls ; then statement_check gnutls/gnutls.h 'gnutls_global_init()' $($_pkg_config --libs --cflags gnutls) && --

1 2

[PATCH 1/4] libmpdemux/mf: Replace sprintf by mp_asprintf
by Alexander Strasser 24 Feb '22

24 Feb '22

It's not a good idea to use sprintf. Especially in this case where the format string is supplied by the user. Signed-off-by: Alexander Strasser <eclipse7(a)gmx.net> --- libmpdemux/mf.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/libmpdemux/mf.c b/libmpdemux/mf.c index 7cdd533d7..d4f5be98b 100644 --- a/libmpdemux/mf.c +++ b/libmpdemux/mf.c @@ -37,6 +37,7 @@ #include "mp_msg.h" #include "help_mp.h" #include "stream/stream.h" +#include "mp_strings.h" #include "mf.h" @@ -110,10 +111,10 @@ mf_t* open_mf(char * filename){ goto exit_mf; } - fname=malloc( strlen( filename ) + 32 ); - if ( !strchr( filename,'%' ) ) { + fname=malloc( strlen( filename ) + 32 ); + strcpy( fname,filename ); if ( !strchr( filename,'*' ) ) strcat( fname,"*" ); @@ -142,19 +143,23 @@ mf_t* open_mf(char * filename){ while ( error_count < 5 ) { - sprintf( fname,filename,count++ ); + fname = mp_asprintf( filename,count++ ); + if ( stat( fname,&fs ) ) { error_count++; mp_msg( MSGT_STREAM,MSGL_V,"[mf] file not found: '%s'\n",fname ); + free(fname); } else { mf->names=realloc( mf->names,( mf->nr_of_files + 1 ) * sizeof( char* ) ); - mf->names[mf->nr_of_files]=strdup( fname ); + mf->names[mf->nr_of_files]=fname; // mp_msg( MSGT_STREAM,MSGL_V,"[mf] added file %d.: %s\n",mf->nr_of_files,mf->names[mf->nr_of_files] ); mf->nr_of_files++; } + + fname = NULL; } mp_msg( MSGT_STREAM,MSGL_INFO,"[mf] number of files: %d\n",mf->nr_of_files ); --

2 6

default faac object type
by Gleb Smirnoff 12 Jul '21

12 Jul '21

Hi, since 2017 libfaac no longer supports any object type other than LOW: #diff-41dda09fa8245ed46fc2dffa26a2f397d293adc5269b6c7dcd8dc53ba6a5e3e5 This basically makes it mandatory to specify '-faacopts object=2' for any invocation of '-oac faac'. Suggested minimal patch defaults the value to LOW. However, maybe makes sense to remove object subagument at all? -- Gleb Smirnoff

1 0