Re: memory corruption bug?

[Norbert Nemec]

On Thu, Mar 08, 2001 at 10:12:04AM +0100, Erik Schnetter wrote:
> If the Sather compiler that you use contains a bug, then changing the
> compilation options might not cure the problem.  Remember, during
> installation of the Sather compiler, you compiled the compiler from C
> sources.  What C compiler and what options did you use there?  I am
> hinting here that the Sather compiler may contain a bug introduced by
> being miscompiled from a bad C compiler.
> 
> You might want to try using a known "good" C compiler, such as gcc
> 2.95.2, and no optimisation, for compiling the Sather compiler.  You
> will have to fiddle with the Makefile in the Boot directory for that,
> but it's straightforward.  Using the resulting Sather compiler will be
> slow, but it might be worth a try.
> 
> I remember seeing posts in this group that even gcc 2.95.2 (and not
> only Red Hat's 2.96) with -O2 miscompiles some code.

Correct, 2.95.2 has a nifty little bug which I could cleanly isolate and which 
is corrected in 2.95.3.

Anyway, I do not believe, a bad cc would normally cause such difficile 
problems connected so closely to a certain piece of Sather code. The old bug, 
for example, did simply show up when there were a large number of local 
variables eliminated by the optimizer (like those L11 = L9; L12 = L11; ...) 
That does not usually happen in handwritten code, but sacomp C-code just 
strained 
the C-optimizer extremely in that point.

Ciao,
Nobbi

-- 
-- ______________________________________________________
-- JESUS CHRIST IS LORD!
--          To Him, even that machine here has to obey...
--
-- _________________________________Norbert "Nobbi" Nemec
-- Hindenburgstr. 44  ...  D-91054 Erlangen  ...  Germany
-- eMail: <address@hidden>   Tel: +49-(0)-9131-204180