From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 20981 invoked by alias); 25 Nov 2003 22:46:07 -0000 Mailing-List: contact gdb-patches-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sources.redhat.com Received: (qmail 20974 invoked from network); 25 Nov 2003 22:46:06 -0000 Received: from unknown (HELO localhost.redhat.com) (207.219.125.105) by sources.redhat.com with SMTP; 25 Nov 2003 22:46:06 -0000 Received: from redhat.com (localhost [127.0.0.1]) by localhost.redhat.com (Postfix) with ESMTP id 2921F2B8F; Tue, 25 Nov 2003 17:46:02 -0500 (EST) Message-ID: <3FC3DBAA.2070605@redhat.com> Date: Tue, 25 Nov 2003 22:46:00 -0000 From: Andrew Cagney User-Agent: Mozilla/5.0 (X11; U; NetBSD macppc; en-US; rv:1.0.2) Gecko/20030820 X-Accept-Language: en-us, en MIME-Version: 1.0 To: Roland McGrath Cc: gdb-patches@sources.redhat.com Subject: Re: [rfa/threads] Convert thread event descriptors to code addrs References: <200311252227.hAPMRcmn010069@magilla.sf.frob.com> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-SW-Source: 2003-11/txt/msg00582.txt.bz2 Does NPTL do anything with that address internally? Or just pass it through? > The alternative would be to modify libthread_db so that it knew that >> PPC64 symbol were special but I suspect that it doesn't want to know >> about such underlying details. > > > It seems to me that this is really what ought to change. ps_pglobal_lookup > and similar symbols have always had an architecture-neutral interface (with > the exception of our invention, ps_get_thread_area). There is no reason > the implementation of this function by a user of libthread_db should have > to do anything machine-specific. gdb already has tons of machine-specific > knowledge, so it's not really an issue there; but if there are ever simpler > users of libthread_db, they shouldn't have to worry about this. It is easy > enough to change libthread_db to request the symbols it really wants for > the addresses it needs. > > Depending how convert_from_func_ptr_addr works, it may well be reasonable > for gdb to use it as your patch does, to work around the existing buggy > PPC64 libthread_db implementations. That is, if that transformation will > always be a no-op for a real code symbol (i.e. a ".foo" on PPC64), then you > might as well make gdb seamlessly handle either case. I'll leave that up > to you, but do let me know whether you decide to take or leave it, so I > know how urgent the glibc fix is for gdb users. > >