This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: excessive memory consumption
- To: jason at cygnus dot com, jbuck at synopsys dot com, egcs at cygnus dot com
- Subject: Re: excessive memory consumption
- From: Manfred Hollstein <manfred at s-direktnet dot de>
- Date: Sun, 5 Oct 1997 20:06:00 +0200
- References: <199709300033.RAA17703.cygnus.egcs@atrus.synopsys.com><u9oh56zn5s.fsf@yorick.cygnus.com><199710042053.WAA00236@saturn.s-direktnet.de>
- Reply-To: Manfred Hollstein <manfred at s-direktnet dot de>
On Sat, 4 October 1997, 22:53:05, manfred@s-direktnet.de wrote:
On , 3 October 1997, 12:01:19, jason@cygnus.com wrote:
Found it! In situations of recursive instantiation, there can be a lot of
stuff compiled in the middle of compiling a given function, running up the
label count. save_for_inline_copying creates new labels for everything
between min_labelno and max_labelno, which can add up to a *lot* of space.
Well, sounds great, but unfortunately, it doesn't work completely; I'm
getting unresolved symbols from libstdc++.so (vtables for __case_table
or so - sorry, removed the logfile too early) on i586-linux-gnulibc1.
Without your patch to toplev.c everything is ok. Looks like we have to
look further :-(
The patch just posted by Alexandre Oliva (Missing virtual tables for
type_info classes in tinfo.cc) cured this problem.
Manfred.