From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 18986 invoked by alias); 31 May 2011 20:41:24 -0000 Mailing-List: contact archer-help@sourceware.org; run by ezmlm Sender: Precedence: bulk List-Post: List-Help: List-Subscribe: List-Id: Received: (qmail 18977 invoked by uid 22791); 31 May 2011 20:41:23 -0000 X-SWARE-Spam-Status: No, hits=-2.0 required=5.0 tests=AWL,BAYES_00,DKIM_SIGNED,DKIM_VALID,DKIM_VALID_AU,SPF_HELO_PASS,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org MIME-Version: 1.0 In-Reply-To: References: <20110525153649.GB3149@redhat.com> <20110531162428.GB3464@redhat.com> From: Paul Pluzhnikov Date: Tue, 31 May 2011 20:41:00 -0000 Message-ID: Subject: Re: Improved linker-debugger interface To: Tom Tromey Cc: archer@sourceware.org Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable X-System-Of-Record: true X-SW-Source: 2011-q2/txt/msg00009.txt.bz2 On Tue, May 31, 2011 at 12:46 PM, Tom Tromey wrote: > Gary> Is it possible for in-process debuggers to locate SystemTap probes? > > The probes are described in an non-allocated section, so IIUC they > aren't mapped by ld.so. =A0I guess a program could examine the executables > and libraries it uses and map them itself. Yes, the in-process debugger usually will try to open the ELF file (e.g. to get to .debug_*, etc.), so it looks like locating the SystemTap probes would work fine. On Tue, May 31, 2011 at 9:24 AM, Gary Benson wrote: > Is making _dl_debug_state_extended an indirect call the only non-hacky > way to allow in-process debuggers to get these notifications, or are > there other possibilities? Once the probe is located, the in-process debugger still has a problem: not enough space to place a patch on (in-process debugger can not put a breakpoint on the one instruction that is available). Making the probes wider would be ideal. > By the way, although _dl_debug_state compiles to a single ret, I think > function entries are aligned so you might get a few more bytes to work > with on some platforms. I believe I've seen some *86 builds of glibc which did not align functions. --=20 Paul Pluzhnikov