Pooma regression in the 3.1 branch

Janis Johnson janis187@us.ibm.com
Wed May 1 09:11:00 GMT 2002


On Wed, May 01, 2002 at 01:47:08PM +0200, Paolo Carlini wrote:
> Hi all,
> 
> today, after two or three days since the last time, I run again the 
> Pooma testsuite on my i686-pc-linux-gnu using a freshly built 3.1.
> 
> Unfortunately I have to report that something is seriously broken in the 
> code generation. Testing stops in Bounce:
> 
> make[1]: Entering directory 
> `/home/paolo/pooma-gcc/examples/Particles/Bounce'
> CXXToSuite... See LINUXgcc/Bounce.o_APP.info
> LinkToSuite... See LINUXgcc/Bounce_APP.info
> Done linking /home/paolo/pooma-gcc/examples/Particles/Bounce/LINUXgcc/Bounce
> make[1]: Leaving directory `/home/paolo/pooma-gcc/examples/Particles/Bounce'
> ./Particles/Bounce/LINUXgcc/Bounce < Particles/Bounce/Bounce.in > 
> Bounce.output
> diff Bounce.output Particles/Bounce/Bounce.out > Bounce.output.diff
> make: *** [Particles/Bounce/Bounce.output] Error 1
> Compiling and executing the Pooma 2.3.0 examples failed.  Please submit 
> tester.log. at ./tester.pl line 196.
> 
> There is a comparison error between Bounce.output and the expected 
> Bounce.out, I'm attaching to this message the former.
> 
> Is anyone else seeing this??
> 
> Ciao, Paolo.

This is also failing on ia64-unknown-linux-gnu with 3.1 built from
sources from about ten hours ago.  It worked a few days ago, but I
haven't tried it since then.

Janis



More information about the Gcc-bugs mailing list