From: SourceForge.net <no...@so...> - 2004-11-27 17:52:22
|
Bugs item #1062987, was opened at 2004-11-09 11:11 Message generated for change (Comment added) made by valtri You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=109655&aid=1062987&group_id=9655 Category: build system problem Group: None Status: Open Resolution: None Priority: 5 Submitted By: Ivan Noris (vix) Assigned to: Nobody/Anonymous (nobody) Summary: can't build xine-lib 1rc-7 on Solaris 9/SPARC Initial Comment: Hi, I've tried to compile the newest 1-rc7 on Solaris9. I was able to build every version before this. No luck today: Making all in lib gmake[3]: Entering directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/lib' if /bin/bash ../libtool-nofpic --mode=compile --tag=CC /usr/local/bin/gcc -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I../include -I../include -I../src -I../src/xine-engine -I../src/xine-engine -I../src/xine-utils -I../src/input -I../src/input -I../lib -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -I/usr/local/include -I/usr/openwin/include -L/usr/openwin/lib -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -O3 -mcpu=ultrasparc -mtune=ultrasparc -funroll-loops -finline-functions -Wall -DNDEBUG -D_REENTRANT -D_FILE_OFFSET_BITS=64 -DXINE_COMPILE -Wpointer-arith -Wnested-externs -Wcast-align -Wchar-subscripts -Wmissing-declarations -Wmissing-prototypes -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -L/opt/BGSgnome-2.6.2/lib -L/usr/X11R6/lib -I/usr/local/include -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/usr/openwin/include -L/usr/openwin/lib -O3 -Wall -pipe -fomit-frame-pointer -mcpu=ultrasparc -MT hstrerror.lo -MD -MP -MF ".deps/hstrerror.Tpo" -c -o hstrerror.lo hstrerror.c; then mv -f ".deps/hstrerror.Tpo" ".deps/hstrerror.Plo"; else rm -f ".deps/hstrerror.Tpo"; exit 1; fi /usr/local/bin/gcc -DHAVE_CONFIG_H -I. -I. -I.. -I.. -I../include -I../include -I../src -I../src/xine-engine -I../src/xine-engine -I../src/xine-utils -I../src/input -I../src/input -I../lib -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -I/usr/local/include -I/usr/openwin/include -L/usr/openwin/lib -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -O3 -mcpu=ultrasparc -mtune=ultrasparc -funroll-loops -finline-functions -Wall -DNDEBUG -D_REENTRANT -D_FILE_OFFSET_BITS=64 -DXINE_COMPILE -Wpointer-arith -Wnested-externs -Wcast-align -Wchar-subscripts -Wmissing-declarations -Wmissing-prototypes -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -L/opt/BGSgnome-2.6.2/lib -L/usr/X11R6/lib -I/usr/local/include -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/usr/openwin/include -L/usr/openwin/lib -O3 -Wall -pipe -fomit-frame-pointer -mcpu=ultrasparc -MT hstrerror.lo -MD -MP -MF .deps/hstrerror.Tpo -c hstrerror.c -fPIC -DPIC -o .libs/hstrerror.o In file included from hstrerror.c:6: /usr/include/netdb.h:304: error: parse error before '(' token gmake[3]: *** [hstrerror.lo] Error 1 gmake[3]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/lib' gmake[2]: *** [all-recursive] Error 1 gmake[2]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake[1]: *** [all] Error 2 gmake[1]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake: *** [build-work/main.d/xine-lib-1-rc7/Makefile] Error 2 root@athena /net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib# I've tried to look for netdb.h problem further: config.log says: configure:41750: checking netdb.h usability configure:41762: /usr/local/bin/gcc -c -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -L/opt/BGSgnome-2.6.2/lib -L/usr/X11R6/lib -I/usr/local/include -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/usr/openwin/include -L/usr/openwin/lib -O3 -Wall -pipe -fomit-frame-pointer -mcpu=ultrasparc -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -I/usr/local/include -I/usr/openwin/include -L/usr/openwin/lib -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib conftest.c >&5 configure:41768: $? = 0 configure:41771: test -z || test ! -s conftest.err configure:41774: $? = 0 configure:41777: test -s conftest.o configure:41780: $? = 0 configure:41790: result: yes configure:41794: checking netdb.h presence configure:41804: /usr/local/bin/gcc -E -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -I/usr/local/include -I/usr/openwin/include -L/usr/openwin/lib -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib conftest.c configure:41810: $? = 0 configure:41830: result: yes configure:41865: checking for netdb.h configure:41872: result: yes Further more. I've tried to make simple "a.c" test "program" with: #include <netdb.h> int main (void) { exit(0); } This program compiles with gcc -Wall a.c. No warning. No error. I've even tried to take the last "gcc" command displayed in the xine-lib compilation and replace the source and output file with my /var/tmp/a.c and /var/tmp/a.o. Worked, no errorr displayed. As the hstrerror.c contains: #ifdef WIN32 #include <winsock.h> #else #include <netdb.h> #endif #include <errno.h> where #include <netdb.h> is on line 6, I really don't understand the problem or how to avoid it. Can you help? p.s. the line 304 of netdb.h says: const char *hstrerror(int); /* End BIND */ If you need more of context information, just drop a line and I will do what I can for this problem to dissapear. Thanks. ---------------------------------------------------------------------- >Comment By: Frantisek Dvorak (valtri) Date: 2004-11-27 18:52 Message: Logged In: YES user_id=543511 I made a fix for hstrerror problem. Can you please try lib_buildfix_herr.patch? You will need rerun autogen.sh. ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-15 15:12 Message: Logged In: YES user_id=566022 Confirmed the makefile problem. The exact error message follows: gmake[2]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/intl' Making install in po gmake[2]: Entering directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/po' Makefile:221: *** Recursive variable `mkinstalldirs' references itself (eventually). Stop. gmake[2]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/po' gmake[1]: *** [install-recursive] Error 1 gmake[1]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake: *** [install-work/main.d/xine-lib-1-rc7/Makefile] Error 2 ---------------------------------------------------------------------- Comment By: Dragan Milicic (milicic) Date: 2004-11-14 01:54 Message: Logged In: YES user_id=1150106 Adding LDFLAGS=-lresolv in make in my last message is unnecessary. Just first two modifications are needed. ---------------------------------------------------------------------- Comment By: Dragan Milicic (milicic) Date: 2004-11-13 23:36 Message: Logged In: YES user_id=1150106 I managed to get xine-lib-1-rc7 to compile on Solaris 9 (sparc) by adding #elif defined(sun) #define PRIiMAX PRIi64 #define PRIXMAX PRIX64 # include <inttypes.h> in lib/os_types.h; running configure, changing /* #undef HAVE_HSTRERROR */ into #define HAVE_HSTRERROR 1 in config.h and doing gnumake LDFLAGS=-lresolv Still, it seems that the problem could be more difficult. Since make install failed, I attempted to do install "by hand" what resulted in xine dumping core. ---------------------------------------------------------------------- Comment By: Dragan Milicic (milicic) Date: 2004-11-13 21:21 Message: Logged In: YES user_id=1150106 On Solaris 9 "fgrep -rl PRIXMAX /usr/include" finds nothing. sys/isa_defs.h defines similar macros as PRIxMAX but PRIXMAX and PRIiMAX are missing. ---------------------------------------------------------------------- Comment By: Michael Roitzsch (mroi) Date: 2004-11-13 16:44 Message: Logged In: YES user_id=552060 There was a change to the build system which moves a lot of platform-specific header inclusions, defines and function replacements to xine-lib/lib. Unfortunately, there seem to be some C99 printf macros missing (that's what the PRIXMAX and PRIX64 are). I assume they are defined in Solaris somewhere, so it would be great, if you could find the right header. (Try "fgrep -rl PRIXMAX /usr/include".) Then, we just need to modify xine-lib/lib/os_types.h to include that header. (The cast warnings should have been there for ages, it's just that no Solaris developer seemed to be disturbed enough to suggest a fix, I guess some alignment pragma at the right place should do.) Michael ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-13 11:53 Message: Logged In: YES user_id=566022 Well, the question is, what is PRIMAX and what is PRIX64 value? This should be configurable/autodetectable by the configure script... I still don't understand, what changes have been done to the source code, so that it's not buildable on Solaris anymore :( OK, anyway I will try to change this constant as well, and see. Maybe I will have luck and solve the Makefile problem. But not until the end of this weekend. ---------------------------------------------------------------------- Comment By: Dragan Milicic (milicic) Date: 2004-11-13 08:37 Message: Logged In: YES user_id=1150106 I changed PRIXMAX -> PRIX64 (I just tried in in analogy with PRIiMAX -> PRIi64 change) and changed more PRIiMAX -> PRIi64 and the library finally compiled. Then make install failed with .... gnumake[1]: Leaving directory `/home/dragan/Sources/xine-lib-1-rc7/intl' Making install in po gnumake[1]: Entering directory `/home/dragan/Sources/xine-lib-1-rc7/po' Makefile:221: *** Recursive variable `mkinstalldirs' references itself (eventually). Stop. gnumake[1]: Leaving directory `/home/dragan/Sources/xine-lib-1-rc7/po' gnumake: *** [install-recursive] Error 1 ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-12 18:41 Message: Logged In: YES user_id=566022 Well, there are more places with the same PRIiMAX -> PRIi64 thing. After changing them, I got: demux_film.c: In function `open_film_file': demux_film.c:202: warning: cast increases required alignment of target type demux_film.c:203: warning: cast increases required alignment of target type demux_film.c:314: error: parse error before "PRIXMAX" demux_film.c: In function `demux_film_send_chunk': demux_film.c:360: warning: cast increases required alignment of target type demux_film.c: In function `demux_film_send_headers': demux_film.c:661: warning: cast increases required alignment of target type demux_film.c: In function `demux_film_seek': demux_film.c:721: warning: cast increases required alignment of target type demux_film.c: In function `demux_film_dispose': demux_film.c:824: warning: cast increases required alignment of target type demux_film.c: In function `demux_film_get_status': demux_film.c:833: warning: cast increases required alignment of target type demux_film.c: In function `demux_film_get_stream_length': demux_film.c:839: warning: cast increases required alignment of target type gmake[4]: *** [demux_film.lo] Error 1 gmake[4]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/src/demuxers' gmake[3]: *** [all-recursive] Error 1 gmake[3]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/src' gmake[2]: *** [all-recursive] Error 1 gmake[2]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake[1]: *** [all] Error 2 gmake[1]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake: *** [build-work/main.d/xine-lib-1-rc7/Makefile] Error 2 I'm short of time currently, debugging Lotus Notes things and more funny toys, so I can't look into to the source code and search for solution. If you have another hint, please help, if you don't I may have some time in a few days again. Thanks. ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-12 18:19 Message: Logged In: YES user_id=566022 Thanks, this works, although MANY warning about casts & allignments are displayed. I will add another entry if there's another problem, or report success. P.S.: are there so MANY changes between rc6 and rc7?? I used to compile xine-lib for a long time without problems... on Solaris. ---------------------------------------------------------------------- Comment By: Michael Roitzsch (mroi) Date: 2004-11-12 12:34 Message: Logged In: YES user_id=552060 Try replacing all occurences of PRIiMAX with PRIi64 in src/xine-engine/input_rip.c Michael ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-11 18:09 Message: Logged In: YES user_id=566022 Able to go on with the LDFLAGS += -lresolv, although I stop at another point: /usr/local/bin/gcc -DHAVE_CONFIG_H -I. -I. -I../.. -I../.. -I../../include -I../../include -I../../src -I../../src/xine-engine -I../../src/xine-engine -I../../src/xine-utils -I../../src/input -I../../src/input -I../../lib -I/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/lib -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -I/usr/local/include -I/usr/openwin/include -L/usr/openwin/lib -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -I/usr/local/include -I/usr/openwin/include -L/usr/openwin/lib -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/opt/BGSgnome-2.6.2/include/freetype2 -I/opt/BGSgnome-2.6.2/include -O3 -mcpu=ultrasparc -mtune=ultrasparc -funroll-loops -finline-functions -Wall -DNDEBUG -D_REENTRANT -D_FILE_OFFSET_BITS=64 -DXINE_COMPILE -Wpointer-arith -Wnested-externs -Wcast-align -Wchar-subscripts -Wmissing-declarations -Wmissing-prototypes -I/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/lib -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -L/opt/BGSgnome-2.6.2/lib -L/usr/X11R6/lib -I/usr/local/include -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/usr/openwin/include -L/usr/openwin/lib -O3 -Wall -pipe -fomit-frame-pointer -mcpu=ultrasparc -I/opt/BGSgnome-2.6.2/include -I/usr/X11R6/include -L/opt/BGSgnome-2.6.2/lib -L/usr/X11R6/lib -I/usr/local/include -L/usr/local/lib -I/opt/sfw/cups/include -L/opt/sfw/cups/lib -I/usr/openwin/include -L/usr/openwin/lib -O3 -Wall -pipe -fomit-frame-pointer -mcpu=ultrasparc -MT input_rip.lo -MD -MP -MF .deps/input_rip.Tpo -c input_rip.c -fPIC -DPIC -o .libs/input_rip.o input_rip.c: In function `rip_plugin_read': input_rip.c:103: warning: cast increases required alignment of target type input_rip.c:160: error: parse error before "PRIiMAX" input_rip.c: In function `rip_plugin_open': input_rip.c:178: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_get_capabilities': input_rip.c:189: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_read_block': input_rip.c:209: warning: cast increases required alignment of target type input_rip.c:289: error: parse error before "PRIiMAX" input_rip.c: In function `rip_plugin_seek': input_rip.c:334: warning: cast increases required alignment of target type input_rip.c:395: error: parse error before "PRIiMAX" input_rip.c:395: error: `PRIiMAX' undeclared (first use in this function) input_rip.c:395: error: (Each undeclared identifier is reported only once input_rip.c:395: error: for each function it appears in.) input_rip.c: In function `rip_plugin_get_current_pos': input_rip.c:436: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_get_length': input_rip.c:450: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_get_blocksize': input_rip.c:461: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_get_mrl': input_rip.c:467: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_get_optional_data': input_rip.c:474: warning: cast increases required alignment of target type input_rip.c: In function `rip_plugin_dispose': input_rip.c:489: warning: cast increases required alignment of target type input_rip.c: In function `_x_rip_plugin_get_instance': input_rip.c:641: error: parse error before "PRIiMAX" gmake[4]: *** [input_rip.lo] Error 1 gmake[4]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/src/xine-engine' gmake[3]: *** [all-recursive] Error 1 gmake[3]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7/src' gmake[2]: *** [all-recursive] Error 1 gmake[2]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake[1]: *** [all] Error 2 gmake[1]: Leaving directory `/net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib/work/main.d/xine-lib-1-rc7' gmake: *** [build-work/main.d/xine-lib-1-rc7/Makefile] Error 2 root@athena /net/poseidon/export/BGSgnome/garnome-2.6.2_bgs/fifth-toe/xine-lib# ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-10 18:29 Message: Logged In: YES user_id=566022 I'm trying another solution. I've modified my Makefile to contain LDFLAGS += -lresolv Now I'm able to continue. Seems Solaris has hstrerror defined in libresolv and configure didn't know that, so hstrerror was not picked up by it. The xine-lib's version of hstrerror was used. I will give you more information after I build xine-lib successfully. ---------------------------------------------------------------------- Comment By: Michael Roitzsch (mroi) Date: 2004-11-10 18:19 Message: Logged In: YES user_id=552060 So if you put that define in your test program, compilation fails? In that case, could you attach your netdb.h to this bug report? Michael ---------------------------------------------------------------------- Comment By: Nobody/Anonymous (nobody) Date: 2004-11-10 17:43 Message: Logged In: NO I think this causes collision: in lib/os_internal.h: /* replacement of hstrerror */ #ifndef HAVE_HSTRERROR #define hstrerror(ERR) _xine_private_hstrerror((ERR)) const char *_xine_private_hstrerror(int err); #endif and in netdb.h included LATER there is hstrerror referenced. Maybe I should remove the #define hstrerror from os_internal.h ---------------------------------------------------------------------- Comment By: Michael Roitzsch (mroi) Date: 2004-11-10 16:59 Message: Logged In: YES user_id=552060 I would probably do a binary search in config.h. Just copy the upper half of config.h into your test program, replacing the #include "config.h" there. If it fails, the problem is in the upper half, if it succeeds, the problem is in the lower half. Then continue with the half you found to contain the problem and go on like that until you are down to the one line that causes it. Michael ---------------------------------------------------------------------- Comment By: Ivan Noris (vix) Date: 2004-11-10 09:17 Message: Logged In: YES user_id=566022 Confirmed. including "config.h" causes problem. #include "config.h" #include <netdb.h> #include <stdlib.h> int main (void) { exit(0); } no longer works. I'm looking into config.h hoping to find the problem on my own. If you have some hints or advices what to try, do not hesitate to ask me if you don't have Solaris machine. Thanks. ---------------------------------------------------------------------- Comment By: Michael Roitzsch (mroi) Date: 2004-11-09 16:08 Message: Logged In: YES user_id=552060 The only difference in hstrerror.c appears to be the #include "config.h" at the top. What happens when you include config.h in your test program? Michael ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=109655&aid=1062987&group_id=9655 |