This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Very early 3.1 bootstrap failure on sparc-sun-solaris28
- To: Brad Lucier <lucier at math dot purdue dot edu>
- Subject: Re: Very early 3.1 bootstrap failure on sparc-sun-solaris28
- From: Daniel Berlin <dan at cgsoftware dot com>
- Date: Mon, 27 Aug 2001 16:15:41 -0400
- Cc: gcc-patches at gcc dot gnu dot org, gcc at gcc dot gnu dot org,vmakarov at touchme dot toronto dot redhat dot com
- References: <200108272001.f7RK1Ka19956@banach.math.purdue.edu>
Brad Lucier <lucier@math.purdue.edu> writes:
>>
>> The bootstrap failures on sparc-sun-solaris28 are getting earlier and earlier.
>>
>> The build command is:
>>
>> ( ../configure --prefix=/pkgs/gcc-2.96 --enable-languages=c ) ; ( make -j 12 bootstrap BOOT_CFLAGS='-O2 -g -pg' > & build.log ) &
>>
>> It fails before finishing building stage1 with
>>
>> ./genattrtab ../../gcc/config/sparc/sparc.md > tmp-attrtab.c
>> All other genattrtab stuff...done
>> Forming and outputing automata tables...Segmentation Fault
>> make[2]: *** [s-attrtab] Error 139
>> make[2]: Leaving directory `/export/home/lucier/gcc/gcc-3.1/objdir/gcc'
>> make[1]: *** [stage1_build] Error 2
>> make[1]: Leaving directory `/export/home/lucier/gcc/gcc-3.1/objdir/gcc'
>> make: *** [bootstrap] Error 2
>>
>
> A bit more info from gdb genattrtab on sparc:
>
This also happens on rs6000.
I got around it by just defining a simple bs automaton, like the
example in the docs.
> done
> Forming and outputing automata tables...
> Program received signal SIGSEGV, Segmentation fault.
> 0x3463c in output_dfa_max_issue_rate () at ../../gcc/genautomata.c:6123
> 6123 for (automaton = description->first_automaton;
> (gdb) info locals
> automaton = 0x0
> (gdb) where
> #0 0x3463c in output_dfa_max_issue_rate () at ../../gcc/genautomata.c:6123
> #1 0x3e610 in write_automata () at ../../gcc/genautomata.c:8686
> #2 0x24c60 in main (argc=2, argv=0xffbef4ec) at ../../gcc/genattrtab.c:6285
> (gdb) list
> 6118 automaton_t automaton;
> 6119
> 6120 if (UNDEFINED_LONGEST_PATH_LENGTH == ON_THE_PATH || ON_THE_PATH >= 0)
> 6121 abort ();
> 6122 max_dfa_issue_rate = 0;
> 6123 for (automaton = description->first_automaton;
> 6124 automaton != NULL;
> 6125 automaton = automaton->next_automaton)
> 6126 pass_states (automaton, process_state_longest_path_length);
> 6127 fprintf (output_file, "\nint %s = %d;\n",
>
> and, indeed, automaton is not initialized in output_dfa_max_issue_rate.
>
> This is from today's checkin:
>
> http://gcc.gnu.org/ml/gcc-cvs/2001-08/msg00671.html
>
> Brad
--
"I saw a subliminal advertising executive, but only for a second.
"-Steven Wright