Huge memory usage with current g++ :-( -> endless (?) recursion in walk_tree <expand_call_inline>

[Andreas Jaeger]

>>>>> Gabriel Dos Reis writes:

Gaby> Well, I don't know if I should say you're lucky or not :-)

Gaby> I reported a similar problem on the same inputs, yesterday or so.
Gaby> The first time, I got a segmentation fault and the tree-checker says
Gaby> something fishy was going on in expand_call_inline (a scope_stmt was
Gaby> being passed instead of expr_stmt, said the tree-checker).  The next
Gaby> time I tried to compile GCC, all my processes were killed and my
Gaby> machine went very bad. I tried again this morning and same result.
Sounds familiar :-(
Gaby> After some analysis, I was led to conclude that was due to a
Gaby> miscompilation o cc1plus caused by recent changes in the
Gaby> middle/back-end -- starting from 2000-07-21.

Gaby> But so far, nobody seems to experience the same problems. I'm glad to
Gaby> see that I'm not dreaming.
You're not.

Gaby> For that matter, my system is SuSE 6.4 running on a Pentium III at
Gaby> 550/700. 

Ok, so both of us are running SuSE 6.4 on a Pentium III (500 Mhz
here).  But I'm using the current binutils from CVS.  But I don't see
why using SuSE 6.4 should be relevant here.

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de