This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Dec alias patch causing regressions
- To: law at redhat dot com
- Subject: Re: Dec alias patch causing regressions
- From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- Date: Mon, 12 Mar 01 10:44:53 EST
- Cc: gcc at gcc dot gnu dot org
No, that's not a complete solution. Consider that if the memory
locations do not conflict, then the scheduler may also reorder those
instructions and we can get incorrect code.
Oh, that's right. We are not supposed to reuse memory locations
unless they conflict.
We really need to avoid multiple references to the same memory locations
with different alias sets. That's bad. It's very bad.
I'll have to look at why the code that's there to prevent that didn't work
in this case.
Sorry about the confusion.
It'll be a few more hours until the testing for the previous change finishes,
then I'll look at this issue.