------- Comment #5 from rguenther at suse dot de 2010-07-08 10:37 ------- Subject: Re: [4.6 regression] All 32-bit fortran execution tests SEGV on SPARC: unaligned access
On Thu, 8 Jul 2010, ebotcazou at gcc dot gnu dot org wrote: > ------- Comment #4 from ebotcazou at gcc dot gnu dot org 2010-07-08 10:32 > ------- > Looking into it. I would say that previously TARGET_MEM_REFs were handled quite conservatively in mem-attr setting while now with MEM_REFs we (as for indirect-refs) derive alignment from types (ugh, old latent problem). My plan is to add alignment information to pointers and propagate that and just get rid of the alignment extraction from types during expansion. Thus, in the meantime I had installed that use-the-original dereference type thing (which isn't technically correct, at least for non-zero offsets, as they could be propagated in ...). -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44843
