This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH 3/5] combine: add regno field to LOG_LINKS
- From: Segher Boessenkool <segher at kernel dot crashing dot org>
- To: Jeff Law <law at redhat dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Wed, 26 Nov 2014 15:59:26 -0600
- Subject: Re: [PATCH 3/5] combine: add regno field to LOG_LINKS
- Authentication-results: sourceware.org; auth=none
- References: <cover dot 1415984897 dot git dot segher at kernel dot crashing dot org> <66755c54deabecd6a35f4b141a186910b98d965a dot 1415984897 dot git dot segher at kernel dot crashing dot org> <5474CE9C dot 4090002 at redhat dot com> <20141125214738 dot GC11223 at gate dot crashing dot org> <5476188C dot 4010802 at redhat dot com>
On Wed, Nov 26, 2014 at 11:14:36AM -0700, Jeff Law wrote:
> >Are you talking about create_log_links? There can be no duplicates there
> >(anymore), that would be multiple defs of the same reg in the same insn,
> >indeed.
> Yes, I was referring to the code in create_log_links. You dropped the
> check for duplicate links. It caught my eye when reading the changes,
> but then I realized the check may no longer be necessary.
>
> Hmm, what about an insn that has two destinations, which happen to be
> upper and lower SUBREGs of a pseudo. Would that create duplicate links
> after your change?
Yes it would. And I'm not so certain distribute_log_links handles that
situation gracefully. Rats.
IMNSHO such RTL should be invalid (it can always be written simpler as
one SET); but there seems to be no such rule.
I'll add the check back.
Wouldn't it be lovely if we could just use DF here...
Segher