From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 22061 invoked by alias); 20 Mar 2003 09:45:02 -0000 Mailing-List: contact libc-hacker-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-hacker-owner@sources.redhat.com Received: (qmail 22027 invoked from network); 20 Mar 2003 09:45:00 -0000 Received: from unknown (HELO gateway.sf.frob.com) (64.160.55.131) by sources.redhat.com with SMTP; 20 Mar 2003 09:45:00 -0000 Received: from magilla.sf.frob.com (magilla.sf.frob.com [198.49.250.228]) by gateway.sf.frob.com (Postfix) with ESMTP id D798F354C; Thu, 20 Mar 2003 01:44:59 -0800 (PST) Received: (from roland@localhost) by magilla.sf.frob.com (8.11.6/8.11.6) id h2K9ixk07463; Thu, 20 Mar 2003 01:44:59 -0800 Date: Fri, 21 Mar 2003 03:09:00 -0000 Message-Id: <200303200944.h2K9ixk07463@magilla.sf.frob.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit From: Roland McGrath To: sjmunroe@vnet.ibm.com Cc: libc-hacker@sources.redhat.com Subject: Re: Re:check-abi In-Reply-To: Steven Munroe's message of Wednesday, 19 March 2003 15:42:21 -0600 <3E78E43D.CD005008@us.ibm.com> X-Antipastobozoticataclysm: When George Bush projectile vomits antipasto on the Japanese. X-SW-Source: 2003-03/txt/msg00054.txt.bz2 > > After you have a good build, do: > > > > make -s update-abi update-abi-config='CPU-.*-linux.*' > > I gave this a shot for powerpc64 and the update-abi seems to work but the > check-abi fails: In fact you ran this literal command, rather than replacing "CPU" with "powerpc64" or whatever the appropriate bit of regexp is, as was clearly explained in my message. Since your platform's configuration tuple does not match the regexp "CPU-.*-linux.*", check-abi's extraction from the .abilist files yields nothing. Hence the 100k diffs that you helpfully attached uncompressed. There was no problem with the . symbols. I am guessing you didn't look at either the .abilist files or the diffs, since a glance at either shows that both the ".foo F" and "foo O" lines are all there. It seems likely that check-abi will work fine once you have actually done an update-abi run that does anything useful. Please try it. However, seeing all those . symbols makes me think perhaps we should do something to avoid them doubling the size of the .abilist files. Is it always the case that for each "foo O" there must be a ".foo F" and vice versa (among exported global symbols)? I am thinking to make abilist.awk notice when there are either . symbols or O-type symbols and in that case do a check that they all match, and emit just "foo F" in the end (so as to match other platforms). Thanks, Roland