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: pr11135: make PIC register a pseudo

From: Mike Stump <mrs at apple dot com>
To: Rask Ingemann Lambertsen <rask at sygehus dot dk>
Cc: Richard Henderson <rth at redhat dot com>, Ian Lance Taylor <ian at airs dot com>, Stuart Hastings <stuart at apple dot com>, gcc at gcc dot gnu dot org
Date: Mon, 14 May 2007 11:31:52 -0700
Subject: Re: pr11135: make PIC register a pseudo
References: <4061F155-7B01-47B4-B4C5-56A82C2E9824@apple.com> <m3y81uxsju.fsf@gossamer.airs.com> <20060106224830.GA6328@redhat.com> <20070513103221.GL5690@sygehus.dk>

On May 13, 2007, at 3:32 AM, Rask Ingemann Lambertsen wrote:

On Fri, Jan 06, 2006 at 02:48:30PM -0800, Richard Henderson wrote:
What they're looking for is, for functions that don't use
the pic register, to not reserve the pic register so that
it's available for computation.  This is harder.  In theory,
ppc has some scheme for this, where they eliminate the pic
register to itself.  I've never really understood how that's
supposed to work.
Suppose you tell reload that %rpic can be eliminated as %rpic + 0. Reload will start out by assuming that the elimination is possible and %rpic will be available for reloads. If no insns mention %rpic at all, it should just work, and you will have recovered a register.

Hum... Could this work on the i386 port (32-bit) to get another register? If so, we'd love to have the extra register when possible. :-)

References:
- Re: pr11135: make PIC register a pseudo
  - From: Rask Ingemann Lambertsen

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