This is the mail archive of the gcc-patches@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: [RFC][PATCH] Preferred rename register in regrename pass

From: Bernd Schmidt <bschmidt at redhat dot com>
To: Robert Suchanek <Robert dot Suchanek at imgtec dot com>, James Greenhalgh <james dot greenhalgh at arm dot com>, Christophe Lyon <christophe dot lyon at linaro dot org>
Cc: "ebotcazou at adacore dot com" <ebotcazou at adacore dot com>, "gcc-patches at gcc dot gnu dot org" <gcc-patches at gcc dot gnu dot org>
Date: Tue, 10 Nov 2015 23:57:25 +0100
Subject: Re: [RFC][PATCH] Preferred rename register in regrename pass
Authentication-results: sourceware.org; auth=none
References: <B5E67142681B53468FAF6B7C31356562441B7622 at hhmail02 dot hh dot imgtec dot org> <55FC2B8D dot 6000106 at redhat dot com> <B5E67142681B53468FAF6B7C31356562441BCDA8 at hhmail02 dot hh dot imgtec dot org> <56179FC6 dot 3040503 at redhat dot com> <B5E67142681B53468FAF6B7C31356562442BE781 at hhmail02 dot hh dot imgtec dot org> <5640CA1E dot 3070205 at redhat dot com> <B5E67142681B53468FAF6B7C31356562442C19BE at hhmail02 dot hh dot imgtec dot org> <CAKdteObiHcfF3ZUzYdAdRMvqGywbQ8kTg5ZGZ06vKSTFNk5pRg at mail dot gmail dot com> <B5E67142681B53468FAF6B7C31356562442C3E5D at hhmail02 dot hh dot imgtec dot org> <CAKdteOaNpXDFuEPqUi9qr2zp=8tMM6_P57sJ_JuxsyK39GUAiQ at mail dot gmail dot com> <20151110174252 dot GA33923 at arm dot com> <B5E67142681B53468FAF6B7C31356562442C40DB at hhmail02 dot hh dot imgtec dot org>

On 11/10/2015 11:33 PM, Robert Suchanek wrote:


The reason behind the failures is that the terminated_this_insn had
a different number of consecutive registers (and mode) to the input
operand in a move currently being considered for tying. In the fix,
I allow tying only if there is matching number of NREGS.

Bernd, do you think that this check would be sufficient and safe?
I'm not sure what would be better: check the mode, nregs plus perhaps
consider tying only if nregs == 1.

Hmm, but shouldn't the regno still be the same? Or is this a case wherewe have a multi-word chain like ax/dx and then something like a "set bx,dx" involving only a part of it, but the entire chain dies?

I guess this is ok to stop the failures for now, but you may want tomove the check to the point where we set terminated_this_insn. Also, asI pointed out earlier, clearing terminated_this_insn should probablyhappen earlier.



Bernd

Follow-Ups:
- RE: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Robert Suchanek
- RE: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Robert Suchanek

References:
- RE: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Robert Suchanek
- Re: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Bernd Schmidt
- RE: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Robert Suchanek
- Re: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Christophe Lyon
- RE: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Robert Suchanek
- Re: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Christophe Lyon
- Re: [RFC][PATCH] Preferred rename register in regrename pass
  - From: James Greenhalgh
- RE: [RFC][PATCH] Preferred rename register in regrename pass
  - From: Robert Suchanek

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