From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 20141 invoked by alias); 23 Mar 2002 16:52:52 -0000 Mailing-List: contact gdb-help@sources.redhat.com; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-owner@sources.redhat.com Received: (qmail 20134 invoked from network); 23 Mar 2002 16:52:51 -0000 Received: from unknown (HELO nevyn.them.org) (128.2.145.6) by 209.249.29.67 with SMTP; 23 Mar 2002 16:52:51 -0000 Received: from drow by nevyn.them.org with local (Exim 3.35 #1 (Debian)) id 16ookx-0007p0-00; Sat, 23 Mar 2002 11:52:39 -0500 Date: Sat, 23 Mar 2002 08:52:00 -0000 From: Daniel Jacobowitz To: "H . J . Lu" Cc: GDB , Kevin Buettner Subject: Re: gdb doesn't work well with DSO on Linux/mips Message-ID: <20020323115239.A29855@nevyn.them.org> Mail-Followup-To: "H . J . Lu" , GDB , Kevin Buettner References: <20020322234648.A24693@lucon.org> <20020323000803.A25045@lucon.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20020323000803.A25045@lucon.org> User-Agent: Mutt/1.3.23i X-SW-Source: 2002-03/txt/msg00209.txt.bz2 On Sat, Mar 23, 2002 at 12:08:03AM -0800, H . J . Lu wrote: > On Fri, Mar 22, 2002 at 11:46:48PM -0800, H . J . Lu wrote: > > # gdb /lib/ld-2.2.4.so > > (gdb) b _dl_start > > Breakpoint 1 at 0xfe4: file rtld.c, line 181. > > (gdb) r > > Starting program: /lib/ld-2.2.4.so > > Breakpoint 1 at 0x55550fe4: file rtld.c, line 181. > > warning: shared library handler failed to enable breakpoint > > > > The problem is the entry point for SVR4/MIPS is __start. But bkpt_names > > in solib-svr4.c only checks _start. Should I add __start to bkpt_names > > or define SOLIB_BKPT_NAME to __start for Linux/MIPS? > > > > > > This seems to work for me. > > > H.J. > --- > 2002-03-22 H.J. Lu (hjl@gnu.org) > > * config/mips/tm-linux.h (SOLIB_BKPT_NAME): New. Actually, I'd prefer the former (add __start to bkpt_names, after _start). One less thing to worry about when multi-arching. Also, it may be that we should check for the dynamic linker breakpoint symbols in the currently loaded BFD, if there is no .interp section. That would explicitly improve debugging of the dynamic linker. Kevin, does that sound reasonable? -- Daniel Jacobowitz Carnegie Mellon University MontaVista Software Debian GNU/Linux Developer