[tree-ssa] Miscompilation (SIGSEGV) with -ffast-math (!?)

[Richard Guenther]

My favorite testcase is miscompiled by g++-ssa (GCC) 3.5-tree-ssa 
20040220 (merged 20040211) causing it to segfault when using -O2 
-ffast-math as opposed to -O2 which is fine (cmdline params "-n 10").

The testcase can be fetched from
http://www.tat.physik.uni-tuebingen.de/~rguenth/gcc/tramp3d-v2.cpp.gz

Trying to hunt it down with available compilers now, and check todays CVS.

Any ideas though?

Richard.