From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 15896 invoked by alias); 10 Feb 2011 21:03:49 -0000 Mailing-List: contact archer-help@sourceware.org; run by ezmlm Sender: Precedence: bulk List-Post: List-Help: List-Subscribe: List-Id: Received: (qmail 15884 invoked by uid 22791); 10 Feb 2011 21:03:48 -0000 X-SWARE-Spam-Status: No, hits=-6.3 required=5.0 tests=AWL,BAYES_00,RCVD_IN_DNSWL_HI,SPF_HELO_PASS,T_RP_MATCHES_RCVD X-Spam-Check-By: sourceware.org Date: Thu, 10 Feb 2011 21:03:00 -0000 From: Oleg Nesterov To: Roland McGrath Cc: Jan Kratochvil , Project Archer Subject: Re: hw_breakpoint userland interface Message-ID: <20110210205543.GA4590@redhat.com> References: <20110203223905.D0C77180081@magilla.sf.frob.com> <20110207211129.GA23277@host1.dyn.jankratochvil.net> <20110208015844.B994A1814A4@magilla.sf.frob.com> <20110208205953.GA15932@redhat.com> <20110208231846.EC8BA1814AA@magilla.sf.frob.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20110208231846.EC8BA1814AA@magilla.sf.frob.com> User-Agent: Mutt/1.5.18 (2008-05-17) X-SW-Source: 2011-q1/txt/msg00041.txt.bz2 On 02/08, Roland McGrath wrote: > > There are indeed many things to be figured out about a userland interface > for hw_breakpoint. It's certainly quite possible that something entirely > separate from ptrace is the right approach for that. I haven't given it > any thought, really. The same. And yes, perhaps it should be more perf_event oriented. Afaics, it is already possible to create the bp via sys_perf_open(PERF_TYPE_BREAKPOINT), but I didn't verify this. > It should be worked out with Jan and other GDB folks Yes. Firstly we need to know what would be more convenient for those who will actually use the new interface. But there is one thing which is even less clear to me. From you previous email, > > Those can be defined with > > the inheritance and process-wide sharing issues in mind, (to remind, you were talking about watchpoints/PTRACE_O_INHERIT problems). Do you have any plans to introduce something like per-process breakpoints, or (apart from inheriting) everything continues to be per-thread? Oleg.