Help tracing an internal compiler error

[Eric Botcazou]

> Well - if I up the min-expand and min-heapsize values by a factor of ten,
> everything is good... I just did:
>
>    g++ -c -g -O2 --param ggc-min-expand=1000 --param
> ggc-min-heapsize=1310720 -v tmp1.ii
>
> and it compiled fine...  However, it feels to me like this is really just
> a band-aid --- we haven't really fixed the problem, we've just hidden it
> for the moment...

Of course, but at least you have a workaround and we know that the GC is very 
likely to be blamed for the problem... which is actually only half 
interesting since I'm not a specialist of the GC at all.

I'd suggest to file a formal bug report with Bugzilla. Then I'll ask our GC 
specialist to take a look when he has some spare time.

-- 
Eric Botcazou