------- Comment #3 from law at redhat dot com 2010-01-08 16:45 ------- Subject: Re: vectorizer created unaligned vector insns
On 01/08/10 04:58, rguenth at gcc dot gnu dot org wrote: > ------- Comment #2 from rguenth at gcc dot gnu dot org 2010-01-08 11:58 > ------- > There is some hacks with contains_packed_reference but IIRC I was complaining > about these at some point - they are not conservatively correct. > > > I was thinking about this some more and ISTM that if we can't prove suitable alignment, then we have to fall back to a runtime test. So improving the code to detect packed structures would be useful, but is not sufficient to resolve this class of problems. ie, it probably wouldn't be too hard to transform that testcase into one which passed in a pointer to an unaligned palette->elms at which point we're forced to use the runtime test. Thoughts? jeff -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42652
