possible egcs bug

[Michael Poole]

Robin Stephenson <robin.stephenson@blackwell.co.uk> writes:

> I just tried building mysql-3.21.33b (< http://www.tcx.se/ >) using
> egcs-2.91.57 19980901 (egcs-1.1), but it barfed and died with the
> following error:
> 
> cc1plus: memory violation at pc=1200293ec rp=120025da0 (bad address = 00000018)
> sql_yacc.yy: In function `int yyparse()':
> sql_yacc.yy: 945: Internal compiler error.
> sql_yacc.yy: 945: Please submit a full bug report to `egcs-bugs@cygnus.com'.
> make[2]: *** [sql_yacc.o] Error 1
> sh: Mmemory violation at pc=15555a4bcf4 rp=1200111e0 (bad address = 170a0224388)
> make[1]: *** [all-recursive] Segmentation fault
> make: *** [all-recursive-am] Error 2
> 
> Whoops.  Since the only reason I'm trying to build this is because the 
> distribution disappeared, I though first I'd ask if anyone happens to
> have the binary package handy.  Secondly, what's wrong with the build?
> Is it, as it seems to be, an egcs bug?  I'll Cc: egcs-bugs in any case.

    How much memory/swap do you have on that machine?  That file is a
really, really big one, and since it's C++, it sucks down a fair bit of
memory per line.  egcs-1.1 peaked at about 80 or 90 MB size when I
compiled sql_yacc.yy from earlier versions of MySQL (though that was on
an x86 box), and it could be getting bigger.  I'd advise trying to
increase swap (or RAM) and seeing if that fixes things; if not, then
perhaps it really is a bug in egcs.

-- Michael