Hi Aaron, > My expectation ws that depending on the processor's capabilities, this > code would either succeed (on reasonably modern processors) or fail with > SIGILL (Illegal instruction). In general, i386 binaries should be able > to use modern processor features on suitable hardware; GCC just makes a > point of conservatively targeting the lowest common denominator unless > specifically directed otherwise. > > However, I was able to reproduce the segfault on a system of my own that > definitely supports these instructions, so I'm not sure what's up -- > particularly given that the x32 build ran into no such trouble, and that > -Wall yielded no warnings whatsoever. As such, perhaps you should just > give up on any-i386 after all.
OK, thanks for the detailed explanation. I have re-uploaded the package with restricted architectures. Best Sascha -- The Wellcome Trust Sanger Institute is operated by Genome Research Limited, a charity registered in England with number 1021457 and a company registered in England with number 2742969, whose registered office is 215 Euston Road, London, NW1 2BE.