This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PR 6212
- From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
- To: rth at redhat dot com
- Cc: gcc at gcc dot gnu dot org, mark at codesourcery dot com
- Date: Sun, 5 May 02 23:03:05 EDT
- Subject: Re: PR 6212
The mips thing has been fixed.
Good.
But the alignment issue is still a serious performance regression.
That is why the PR is still open and high priority.
Well, not quite as high as something being broken ...
Fixing it without major work is tricky. The problem is that we are
removing a divide/multiply combination, so we lose the info that
there's an aligned value here.
My idea for GCC 3.2 is to add a NOP_WITH_ASSERT tree node that can be used
to record the alignment information.
The first few times I looked at this, that's the only solution I saw.
But now I just realized that one way of doing this may be saying that
a POINTER_TYPE variable has the alignment of the underlying type.
That's an easy change, but one that has the potential of causing
regressions. What's the thought here?