fosslinux/live-bootstrap

Race condition in GCC 4.0.04 with parallel build on

Closed this issue · 4 comments

nanonyme commented

This was with parallel build 12

    echo "\"$f\", "; done >> tmp-gtyp.h
    echo timestamp > doc/cpp.1
    gawk -f ../../gcc/opt-functions.awk -f ../../gcc/opth-gen.awk \
           < optionlist > options.h
    gengtype-yacc.c:1115:60: warning: no newline at end of file
    gengtype-yacc.c:1114:1: error: unterminated #if
    gengtype-yacc.c: In function 'yyparse':
    gengtype-yacc.c:1112: error: syntax error at end of input
    gengtype-yacc.c:1112: warning: unused variable 'yyval'
    gengtype-yacc.c:1109: warning: unused variable 'yytoken'
    gengtype-yacc.c:1107: warning: unused variable 'yyresult'
    gengtype-yacc.c:1106: warning: unused variable 'yyn'
    gengtype-yacc.c:1104: warning: unused variable 'yystacksize'
    gengtype-yacc.c:1102: warning: unused variable 'yyvsp'
    gengtype-yacc.c:1101: warning: unused variable 'yyvs'
    gengtype-yacc.c:1100: warning: unused variable 'yyvsa'
    gengtype-yacc.c:1097: warning: unused variable 'yyssp'
    gengtype-yacc.c:1096: warning: unused variable 'yyss'
    gengtype-yacc.c:1095: warning: unused variable 'yyssa'
    gengtype-yacc.c:1085: warning: unused variable 'yyerrsta
tus'
    gengtype-yacc.c:1083: warning: unused variable 'yystate'
    make: *** [build/gengtype-yacc.o] Error 1
    make: *** Waiting for unfinished jobs....
    (pod2man --center="GNU" --release="gcc-4.0.4" --section=1 cpp.pod > doc/cpp.1.T$$ && \
    	mv -f doc/cpp.1.T$$ doc/cpp.1) || \
    	(rm -f doc/cpp.1.T$$ && exit 1)
    rm -f include/limits.h
    cp xlimits.h include/limits.h
    /bin/sh: line 1: pod2man: command not found
    chmod a+r include/limits.h

Did not reproduce with second attempt. Sounds like possibly root cause related to pod2man.

fosslinux commented

Hm, this is a bit strange. We have had issues with pod2man before. Do you happen to have a complete log from this run you could upload somewhere?

nanonyme commented

There was unfortunately a bug in our CI and we didn't seem to be collecting artifacts on failures so full logs were not available.

nanonyme commented

Looks like this is really really hard to reproduce.

nanonyme commented

I will re-open if we hit this again. There's nothing really actionable without logs.