From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jim Wilson Date: Wed, 16 Jun 2004 21:57:00 +0000 Subject: Re: [PATCH] new ".serialize" gas directive Message-Id: <1087423020.1178.36.camel@leaf.tuliptree.org> List-Id: References: <16537.51724.854691.934006@napali.hpl.hp.com> In-Reply-To: <16537.51724.854691.934006@napali.hpl.hp.com> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: linux-ia64@vger.kernel.org On Tue, 2004-06-15 at 18:54, Chris Wedgwood wrote: > I was just looking over the nested fault handler this thread came to > mind. > There is exactly that situation with '.pred.rel "mutex", p6, p7' where > gas should be able to infer this from the 'cmp.eq p6,p7 = 5,r17' > preceding this and there are no intermediate labels. I assume you are talking about the first one in arch/ia64/kernel/ivt.c. Current gas seems to get this right without the .pred.rel. I see that the LOAD_PHYSICAL macro expansion includes a tag, which is kind of like a label. Perhaps there was a problem with tags in earlier gas versions, or perhaps earlier kernel versions used an actual label instead of a tag here. Use of a label here may have been necessary before the assembler had tag support for instance, in which case the .pred.rel would have been necessary. -- Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com