gcc-4.3 optimizer bugs?

[Benny Amorsen]

Neal Becker <ndbecker2 at gmail.com> writes:

> We are trying to track down a numerical problem with unuran that occurs on
> gcc-4.3.  I know that gcc now has a very extensive test suite, and bugs are
> not common.  I'm just wondering, are others finding similar problems with
> gcc-4.3?
>
> We are working on preparing a bug report.  The symptoms are:
> 1) Problem disappears if -O0
> 2) Problem disappears if a printf is added
> 3) valgrind is silent

The traditional way to find such problems is to:

a) Compile the whole tree with 4.2 (or some other known-good compiler)
b) rm someobject.o
c) Compile someobject.o with 4.3
d) See if the problem appears
e) If not, try some other file

Once you have narrowed it down to a file, you can start looking at the
generated assembly. Or split the file if it's too big to analyse.

It is not a very entertaining process.


/Benny