From: Mark C. <mp...@st...> - 2003-05-26 20:41:01
|
Hi Thibaut, Tim, all, Looking more closely, I put together a little test case, and tried compiling it with just 'gcc t.c', and it worked fine. I tried with the optimisation flags in current cvs, and it segfaulted gcc. Removing the -m128bit-long-double avoids the segfault in gcc, but I'm not sure of the implications for xine. Failed case: $ gcc -O3 -march=pentium3 -mfpmath=sse,387 -mmmx -msse -mno-3dnow -mno-sse2 -funroll-loops -fomit-frame-pointer -foptimize-sibling-calls -minline-all-stringops -malign-double -m128bit-long-double t.c t.c: In function `foo': t.c:15: internal error: Segmentation fault Please submit a full bug report, with preprocessed source if appropriate. See <URL:http://bugzilla.redhat.com/bugzilla/> for instructions. Successful case: $ gcc -O3 -march=pentium3 -mfpmath=sse,387 -mmmx -msse -mno-3dnow -mno-sse2 -funroll-loops -fomit-frame-pointer -foptimize-sibling-calls -minline-all-stringops -malign-double t.c /usr/lib/gcc-lib/i386-redhat-linux/3.2.2/../../../crt1.o(.text+0x18): In function `_start': ../sysdeps/i386/elf/start.S:77: undefined reference to `main' collect2: ld returned 1 exit status I went back and re-checked earlier (compilable) CVS, it seems the optimisation flags used do not include that -m128bit-long-double, and are a quite different set. -O2 -g -pipe -march=i686 -Wall -D_REENTRANT -D_FILE_OFFSET_BITS=64 -DXINE_COMPILE -O3 -pipe -fomit-frame-pointer -falign-functions=4 -falign-loops=4 -falign-jumps=4 -mpreferred-stack-boundary=2 -fexpensive-optimizations -fschedule-insns2 -fno-strict-aliasing -ffast-math -funroll-loops -finline-functions -mcpu=athlon Sorry for the wasted bandwidth pointing to that delta to audio_out.c, when it would appear to be the change to xine-lib.spec.in is actually the culprit. This bit of the delta from 1.67 to 1.68: %build -export CFLAGS="%{optflags}" +export CFLAGS="-O3 -march=pentium3 -mfpmath=sse,387 -mmmx -msse -mno-3dnow -mno-sse2 -funroll-loops -fomit-frame-pointer -foptimize-sibling-calls -minline-all-stringops -malign-double -m128bit-long-double" Reverting this fixed my compile issue. Mark On Mon, 2003-05-26 at 20:42, Tim Champagne wrote: > Thibaut, > > The problem with that code was MSVC could not type cast from a > int64_t to a double. I can assign a double to an int64 value with a > warning about "possible loss of data". > > P.S. Please look at the patch that I just sent out. > > -Tim > > > -----Original Message----- > > From: Thibaut Mattern [mailto:tma...@no...] > > Sent: Monday, May 26, 2003 02:27 PM > > To: Mark Cooke; tc...@au... > > Cc: xin...@li... > > Subject: Re: [xine-user] RH9 compile failure with xine-lib cvs (1.128 > > ofaudio_out.c) > > > > Moving this to xine-devel, > > > > The following code looks strange (audio_out line line 843): > > ------------------------- > > double factor; /* defined line 762 */ > > > > #ifdef _MSC_VER > > factor = (int64_t)gap_diff / (int64_t)info->window_duration + > > info->last_factor; > > #else > > factor = (double)gap_diff / (double)info->window_duration + > > info->last_factor; > > #endif /* _MSC_VER */ > > ------------------------- > > > > int64_t is an integer type > > double is a floating point type > > > > why is there an integer div in the windows port ? > > > > cheers, > > thibaut > > > > > > On Mon, 2003-05-26 at 00:38, Mark Cooke wrote: > > > Hi, > > > > > > On a bit more investigation, it seems to be the use of int64_t. > > > > > > The 25-May-2003 checkout (1.127) compiles fine. The current one > does > > > not. > > > > > > See: cvs diff -u -r 1.127 -r 1.128 > xine-lib/src/xine-engine/audio_out.c > > > > > > Mark > > > > > > On Sun, 2003-05-25 at 22:51, Mark Cooke wrote: > > > > Though I just tried to compile latest CVS on two different RH9 > boxes, > > > > and it bombed out with a compiler error in audio_out.c both times: > > > > > > > > audio_out.c: In function `resample_rate_adjust': > > > > audio_out.c:843: internal error: Segmentation fault > > > > > > > > So it's possible it is tickling a real compiler bug. > > > > > > > > gcc-3.2.2-5 on my fully up-to-date RH9 box. > > > > > > > > Mark > > > > > > > > On Sun, 2003-05-25 at 22:17, Bartłomiej Muryn wrote: > > > > > Yes try if there is no libpng-devel package > > > > > B. > > > > > On Sunday 25 of May 2003 22:47, chris ferguson wrote: > > > > > > Thanks for the suggestion... > > > > > > > > > > > > On trying to install the files from CD, the package manager > tells > > me > > > > > > they are already installed... > > > > > > > > > > > > any other ideas I could try? > > > > > > > > > > > > Regards, > > > > > > > > > > > > Chris > > > > > > > > > > > > On Sun, 2003-05-25 at 21:15, Bartłomiej Muryn wrote: > > > > > > > No No = in Your case - critical is only the libpng library > which > > You have > > > > > > > to install from You rh90 installation disk > > > > > > > > > > > > > > B. > > > > > > > > > > > > > > On Sunday 25 of May 2003 22:00, chris ferguson wrote: > > > > > > > > Greetings to you all! > > > > > > > > > > > > > > > > I've just moved to RedHat 9.0 from Mandrake 9.0. On > Mandrake, > > I had > > > > > > > > few problems compiling and installing Xine. However, on > > RedHat 9, this > > > > > > > > process has so far been a minefield. > > > > > > > > > > > > > > > > Having compiled and "successfully" installed xine-lib-1- > > beta12, I then > > > > > > > > had to change the /etc/ld.so.conf to add the line > > "/usr/local/lib". So > > > > > > > > far so good. > > > > > > > > > > > > > > > > I've now unpacked the file xine-ui-0.9.21 and am trying to > run > > the > > > > > > > > ./configure script. > > > > > > > > > > > > > > > > I am getting the following error-message output: > > > > > > > > > > > > > > > > ...(Full output cut here)... > > > > > > > > > > > > > > > > checking for CURL-LIB version >= 7.10.2... no > > > > > > > > *** The curl-config script installed by CURL could not be > > found > > > > > > > > *** If CURL was installed in PREFIX, make sure PREFIX/bin > is > > in > > > > > > > > *** your path, or set the CURL_CONFIG environment variable > to > > the > > > > > > > > *** full path to curl-config. > > > > > > > > *** All of curl dependent parts will be disabled *** > > > > > > > > checking for socket in -lsocket... no > > > > > > > > checking for gethostbyname in -lnsl... yes > > > > > > > > checking for aalib-config... no > > > > > > > > checking for aainfo... no > > > > > > > > checking for AALIB version >= 1.2.0... no > > > > > > > > *** The aalib-config|aainfo program installed by AALIB > could > > not be > > > > > > > > found > > > > > > > > *** If AALIB was installed in PREFIX, make sure PREFIX/bin > is > > in > > > > > > > > *** your path, or use --with-aalib-prefix to set the > prefix > > > > > > > > *** where AALIB is installed. > > > > > > > > checking for pkg-config... /usr/bin/pkg-config > > > > > > > > checking for png_read_info in -lpng... no > > > > > > > > configure: error: *** libpng is needed (PNG library not > found) > > - try to > > > > > > > > use --with-png-prefix option *** > > > > > > > > [root@localhost xine-ui-0.9.21]# > > > > > > > > > > > > > > > > As a relatively new user of Linux, and having installed > the > > latest > > > > > > > > release of curl, I am somewhat puzzled as to what I am to > do > > next. > > > > > > > > > > > > > > > > I had thought that AALIB was installed as part of the > > libraries file... > > > > > > > > is this not the case? Has anyone else had similar > problems > > with Redhat > > > > > > > > 9.0? > > > > > > > > > > > > > > > > In short, what do I need to do to get xine working in > Redhat > > 9.0?? > > > > > > > > > > > > > > > > If it helps, the core of my system consists of: > > > > > > > > > > > > > > > > - AMD Duron 1.1 Ghz > > > > > > > > - 256Mb Ram > > > > > > > > - ATI Mach 64 (Utah?) 8Mb AGP Graphics > > > > > > > > > > > > > > > > > > > > > > > > I'd appreciate any help that anyone can offer me on this. > > > > > > > > > > > > > > > > Regards, > > > > > > > > > > > > > > > > > > > > > > > > chris ferguson <ch...@so...> > > > > > > > > Solstace > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------- > > > > > > > > This SF.net email is sponsored by: ObjectStore. > > > > > > > > If flattening out C++ or Java code to make your > application > > fit in a > > > > > > > > relational database is painful, don't do it! Check out > > ObjectStore. > > > > > > > > Now part of Progress Software. > > http://www.objectstore.net/sourceforge > > > > > > > > _______________________________________________ > > > > > > > > xine-user mailing list > > > > > > > > xin...@li... > > > > > > > > https://lists.sourceforge.net/lists/listinfo/xine-user > > > > > > > > > > > > > > ------------------------------------------------------- > > > > > > > This SF.net email is sponsored by: ObjectStore. > > > > > > > If flattening out C++ or Java code to make your application > fit > > in a > > > > > > > relational database is painful, don't do it! Check out > > ObjectStore. > > > > > > > Now part of Progress Software. > > http://www.objectstore.net/sourceforge > > > > > > > _______________________________________________ > > > > > > > xine-user mailing list > > > > > > > xin...@li... > > > > > > > https://lists.sourceforge.net/lists/listinfo/xine-user > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------- > > > > > This SF.net email is sponsored by: ObjectStore. > > > > > If flattening out C++ or Java code to make your application fit > in a > > > > > relational database is painful, don't do it! Check out > ObjectStore. > > > > > Now part of Progress Software. > > http://www.objectstore.net/sourceforge > > > > > _______________________________________________ > > > > > xine-user mailing list > > > > > xin...@li... > > > > > https://lists.sourceforge.net/lists/listinfo/xine-user -- Mark Cooke <mp...@st...> |