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: [PATCH] Fix small structure passing on x86-64

From: Jan Hubicka <jh at suse dot cz>
To: Eric Botcazou <ebotcazou at adacore dot com>
Cc: Jan Hubicka <hubicka at ucw dot cz>, gcc-patches at gcc dot gnu dot org
Date: Thu, 13 Nov 2008 16:23:29 +0100
Subject: Re: [PATCH] Fix small structure passing on x86-64
References: <200810311146.58286.ebotcazou@adacore.com> <20081112163503.GM27401@atrey.karlin.mff.cuni.cz> <200811131558.16459.ebotcazou@adacore.com>

> > These seems to be all wrong in all GCC versions on x86-64.  I've updated
> > move_block_to_reg to tage size in bytes, not in the number of words.  It
> > uses partial move at the end of block if there is appropriate mode or
> > bitfield extraction otherwise.
> >
> > I hope I did not introduced any endianity issues, but I am not sure if
> > there is big endian machine that cares.
> >
> > Seems sane?
> 
> Are you sure that's really the way to go?  Why not generate PARALLELs like the 
> other ports?

Well, x86-64 does generate parallels, but the generic code is using too
long moves to load them...
If I have structure of size 3 bytes to be passed in RAX, how should the
parallel look like?

Honza
> 
> -- 
> Eric Botcazou

References:
- Re: [PATCH] Fix small structure passing on x86-64
  - From: Jan Hubicka
- Re: [PATCH] Fix small structure passing on x86-64
  - From: Eric Botcazou

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