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] Tree SRA and atomicity/volatility

From: kenner at vlsi1 dot ultra dot nyu dot edu (Richard Kenner)
To: richard dot guenther at gmail dot com
Cc: ebotcazou at adacore dot com, gcc-patches at gcc dot gnu dot org
Date: Sat, 06 Jan 2007 08:52:54 EST
Subject: Re: [PATCH] Tree SRA and atomicity/volatility
References: <200701061422.39157.ebotcazou@adacore.com> <84fc9c000701060531g4b2da007mf5434b3fce63ec24@mail.gmail.com>

> The problem is really that the middle-end doesn't have a good
> definition of volatile semantics on a structure.  At least not what
> you want from it.

As was discussed in another thread a while ago, the middle-end doesn't wreally
have a good definition of volatile semantics AT ALL: a lot of it is just "do
what we used to do".

My feeling is that the proper semantics of volatile are that the access
pattern in the generated code should match, as closely as the underlying
machine allows, the access pattern in the source.  Unfortunately, this is an
informal, and hence necessarily imprecise, definition, but I think it might
be the best we can do.  It's certainly consistent with past practice.

References:
- [PATCH] Tree SRA and atomicity/volatility
  - From: Eric Botcazou
- Re: [PATCH] Tree SRA and atomicity/volatility
  - From: Richard Guenther

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