Re: gcc 2.95 : internal compiler error with qt 1.44

[Alexander Feigl <Alexander dot Feigl at gmx dot de>]

Am Don, 05 Aug 1999 schrieben Sie:
> In message <99080508385900.01652@PowerBox.MysticWorld.de>you write:
>   > Any suggestions? Ask the Trolls for permission to send a .ii file?
> That is a good option.  Explain what you are trying to accomplish and I
> suspect they'll accommodate you.
> 

I sent a mail to them. Will take some time until they'll answer

>   > While I wrote this mail and doing some testing I just found out that even
>   > the building directory has effects. The building directory affects the
>   > -I directory path which is passed to the compiler:
> [ ... ]
>   > I diffed the .ii outputs of both variants and there were only differences
>   > in the path of the included QT headers 
> This sounds like a memory management/memory corruption problem.  There is a
> good chance that we won't even be able to reproduce it if you send us the
> exact sources since those kinds of bugs are *extremely* fragile.
> 
> However, there is also the *possibility* that a recent bugfix we made to
> gcc/tree.c will fix this problem.  Then again, it may not fix the problem.
> You can pick up the patch via cvsweb.  Look for a change to gcc/tree.c which
> mentions obstacks.
> 

Fetched it but it didn't fix my problem. :(

> If that doesn't fix the problem, the best thing to do will be for you to
> debug it yourself.  Given the same compiler, run it in two different
> windows, one with the input that works, the other with the input that does
> not work and debug the compiler.
> 

How can I call the cc1plus compiler directy without frontend. I need following
compile options:

gcc -O2 -fPIC qmenubar.ii

When I call :

cc1plus -O2 -fPIC qmenubar.ii

I get very verbose output but the compiler error doesn't occur. Any Options I
missed or is this bug only caused when cc1plus is called by gcc.

I tracked down where the internal compiler error gets called. It is a
segmentation fault which is catched by gcc/cp/decl.c. If I get cc1plus to crash
without gcc frontend I would be able to send a backtrace. I'm not very
experienced with debugging, so debugging the whole cc1plus would be hard. :(


> Another thing you can try is using a debugging malloc or purify to try and
> track this down.  They may help, then again, they may not.

I'm just trying dmalloc. Running with level "high" doesn't give any results so
I retried with level "all". gcc is running for an hour now and I suppose it
will run some more time. When I link the dmalloc library the SEGV doesn't occur
but I hope I'll get something in the log files.

Greetings

Alexander