This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: Fix PR43077, debug regression with SSA expand
- From: Michael Matz <matz at suse dot de>
- To: gcc-patches at gcc dot gnu dot org
- Cc: Jakub Jelinek <jakub at redhat dot com>
- Date: Fri, 19 Feb 2010 16:40:12 +0100 (CET)
- Subject: Re: Fix PR43077, debug regression with SSA expand
- References: <Pine.LNX.4.64.1002191604380.18785@wotan.suse.de>
Hi,
On Fri, 19 Feb 2010, Michael Matz wrote:
> When TER decides to forward SSA names also debug instructions need to
> expand them in place. The testcase comes from Jakub (thanks) and works
> with the patch. If approved I'd commit both, patch and testcase.
> Regstrapping on x86_64-linux in progress. Okay for trunk if that works?
Due to some discussion I retract this one for now. It would do the wrong
thing for such situation:
a_1 = b_2 + c_3 // assume a_1 is TERed
... use(a_1) // only real use of a_1
b_3 = ...
#DEBUG something => f(a_1)
When forwarding the RHS "b_2 + c_3" into the debug use we might have
clobbered b_2 already, namely when b_3 and b_2 can be coalesced. I'm now
planning to use debug temps for this situation.
Ciao,
Michael.