This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: [tree-ssa] Question: going in and out of SSA

From: Devang Patel <dpatel at apple dot com>
To: law at redhat dot com
Cc: gcc mailing list <gcc at gcc dot gnu dot org>
Date: Tue, 6 Jan 2004 12:22:41 -0800
Subject: Re: [tree-ssa] Question: going in and out of SSA
References: <200401060607.i0667oWv011800@speedy.slc.redhat.com>

On Jan 5, 2004, at 10:07 PM, law@redhat.com wrote:


TER creates non-gimple code which is the most likely reason why
you got the abort.

I did not know about this. Thank you.

You'd probably be better off only rewriting those variables which are affected by your changes -- assuming of course that you can compute that set of variables without an amazing amount of difficulty. That's what the dominator optimizer does when it threads jumps -- threading jumps changes the dominator tree and requires a subset of the program's variables to be taken out of SSA form, then rewritten back into SSA form.

Yes, that's what I am going to do. If it does not work well then I'll try disabling TER.

--
Devang

References:
- Re: [tree-ssa] Question: going in and out of SSA
  - From: law

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]