It crashes in do_yyparse.
| 5741 do_yyparse(); /* Parse da program */ | (gdb) n | Program received signal SIGSEGV, Segmentation fault. | 0x082e9574 in ?? ()
Defining COMPILER_DEBUG and PROGRAM_BUILD_DEBUG gives
[...]
| added new definition #6 | finishing program 14 (pass=2) | th(16384) 0x82c38e4 end_first_pass(1): threads_disabled:3, compilation_depth:2 | defining constant (pass=1): function( : object(is 14)) "Backend" = Need a new master object... | Cannot load master object yet! | /home/mirar/pike/src/object.c:616: Fatal error: | Couldn't load master object. | No stack - no backtrace. | make[3]: *** [compile1] Aborted (core dumped)
???
/ Mirar
Previous text:
2003-01-05 08:29: Subject: Re: Problem compiling Pike
It crashes in do_yyparse.
| 5741 do_yyparse(); /* Parse da program */ | (gdb) n | Program received signal SIGSEGV, Segmentation fault. | 0x082e9574 in ?? ()
/ Mirar
It runs with --without-threads and/or with --without-machine-code. I need a faster machine to rebuild pike. It takes minutes...
/ Mirar
Previous text:
2003-01-05 08:43: Subject: Re: Problem compiling Pike
Defining COMPILER_DEBUG and PROGRAM_BUILD_DEBUG gives
[...]
| added new definition #6 | finishing program 14 (pass=2) | th(16384) 0x82c38e4 end_first_pass(1): threads_disabled:3, compilation_depth:2 | defining constant (pass=1): function( : object(is 14)) "Backend" = Need a new master object... | Cannot load master object yet! | /home/mirar/pike/src/object.c:616: Fatal error: | Couldn't load master object. | No stack - no backtrace. | make[3]: *** [compile1] Aborted (core dumped)
???
/ Mirar
Now you get a Pike_fatal and previously you got a raw SIGSEGV. That's odd since neither COMPILER_DEBUG nor PROGRAM_BUILD_DEBUG adds more debug checks. It could perhaps be that there has become a circularity with PROGRAM_BUILD_DEBUG since it calls describe_svalue which can call functions in the master, something that wouldn't work if the debug printouts are made before the master is loaded.
Are you using --with-rtldebug?
/ Martin Stjernholm, Roxen IS
Previous text:
2003-01-05 08:43: Subject: Re: Problem compiling Pike
Defining COMPILER_DEBUG and PROGRAM_BUILD_DEBUG gives
[...]
| added new definition #6 | finishing program 14 (pass=2) | th(16384) 0x82c38e4 end_first_pass(1): threads_disabled:3, compilation_depth:2 | defining constant (pass=1): function( : object(is 14)) "Backend" = Need a new master object... | Cannot load master object yet! | /home/mirar/pike/src/object.c:616: Fatal error: | Couldn't load master object. | No stack - no backtrace. | make[3]: *** [compile1] Aborted (core dumped)
???
/ Mirar
Hm, yes, that could be it. I couldn't get any pike to work with PROGRAM_BUILD_DEBUG. When I removed it all my various configured Pikes started to work so I gave up...
No, I didn't use --with-rtldebug. --with-double-precision, though, but that shouldn't matter here (should it?).
/ Mirar
Previous text:
2003-01-05 23:12: Subject: Re: Problem compiling Pike
Now you get a Pike_fatal and previously you got a raw SIGSEGV. That's odd since neither COMPILER_DEBUG nor PROGRAM_BUILD_DEBUG adds more debug checks. It could perhaps be that there has become a circularity with PROGRAM_BUILD_DEBUG since it calls describe_svalue which can call functions in the master, something that wouldn't work if the debug printouts are made before the master is loaded.
Are you using --with-rtldebug?
/ Martin Stjernholm, Roxen IS
pike-devel@lists.lysator.liu.se
-
Martin Stjernholm, Roxen IS ＠ Pike developers forum -
Mirar ＠ Pike developers forum