Re: c++/8511: (hopefully) reproducible cc1plus SIGSEGV.

[Volker Reichelt <reichelt at igpm dot rwth-aachen dot de>]

> The point is something different. I get a real SIGSEGV, NOT an internal 
> compiler error. Sadly, I have quite a lot of heavy template code which 
> triggers internal compiler errors (I reported one of them some time back 
> and it is not yet fixed), but this one really makes gcc SIGSEGV.

I suppose it's *not* a "Internal compiler error: Segmentation fault".
Did I get you rught?

> Also, the debugger shows the suspicious address 0xa5a5a5a5 which might 
> indicate some more serious bug inside the compiler than simply a missing 
> C++ language feature.

> So, please tell me if you can reproduce a SIGSEGV, and not an internal 
> compiler error.

I just tried gcc 3.2 on your sources, but I only get an ICE.
However, since you haven't provided a preprocessed source, I'm probably
compiling different code than you.

Can you generate a preprocessed source that also causes a segfault?
If yes, could you please send the preprocessed file?

If no, then something strange is happening :-(
My first wild guess would be that the compiler ran out of memory when
compiling your source. Can you make sure that this is not the reason
(i.e. by using a swap device that is large enough)?
You might have hardware problems: for example faulty memory (you could
try to swap your chips if you have multiple banks of memory). I remember
one PR where a faulty BIOS was responsible for strange gcc errors
(are BIOS upgrades available for your board).

Good luck,
Volker

PS: gcc-prs is a read-only list and therefore always bounces,
just don't send any mail to that address.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=8511