Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1754009Ab2H2PsH (ORCPT ); Wed, 29 Aug 2012 11:48:07 -0400 Received: from mx1.redhat.com ([209.132.183.28]:45068 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753895Ab2H2PsF (ORCPT ); Wed, 29 Aug 2012 11:48:05 -0400 Date: Wed, 29 Aug 2012 17:49:52 +0200 From: Oleg Nesterov To: Sebastian Andrzej Siewior Cc: Peter Zijlstra , linux-kernel@vger.kernel.org, x86@kernel.org, Arnaldo Carvalho de Melo , Srikar Dronamraju , Ananth N Mavinakaynahalli , stan_shebs@mentor.com, gdb-patches@sourceware.org Subject: Re: [RFC 5/5 v2] uprobes: add global breakpoints Message-ID: <20120829154952.GA29101@redhat.com> References: <1344355952-2382-1-git-send-email-bigeasy@linutronix.de> <1344355952-2382-6-git-send-email-bigeasy@linutronix.de> <1344857686.31459.25.camel@twins> <20120821194200.GA32293@linutronix.de> <20120822134837.GA28878@redhat.com> <503BC2F1.5060003@linutronix.de> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <503BC2F1.5060003@linutronix.de> User-Agent: Mutt/1.5.18 (2008-05-17) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2597 Lines: 79 On 08/27, Sebastian Andrzej Siewior wrote: > > On 08/22/2012 03:48 PM, Oleg Nesterov wrote: >> On 08/21, Sebastian Andrzej Siewior wrote: >>> >>> - not putting the task in TASK_TRACED but simply halt. This would work >>> without a change to ptrace_attach() but the task continues on any >>> signal. So a signal friendly task would continue and not notice a >>> thing. >> >> TASK_KILLABLE > > That would help but would require a change in ptrace_attach() or > something in gdb/strace/… Well, I still think you should not touch ptrace_attach() at all. > One thing I just noticed: If I don't register a handler for SIGUSR1 and > send one to the application while it is in TASK_KILLABLE then the > signal gets delivered. Not really delivered... OK, it can be delivered (dequeued) before the task sees SIGKILL, but this can be changed. In short: in this case the task is correctly SIGKILL'ed. See sig_fatal() in complete_signal(). > If I register a signal handler for it than it > gets blocked and delivered once I resume the task. Sure, if you have a handler, the signal is not fatal. > Shouldn't it get blocked even if I don't register a handler for it? No. >> Am I understand correctly? >> >> If it was woken by PTRACE_ATTACH we set utask->skip_handler = 1 and >> re-execute the instruction (yes, SIGTRAP, but this doesn't matter). >> When the task hits this bp again we skip handler_chain() because it >> was already reported. >> >> Yes? If yes, I don't think this can work. Suppose that the task >> dequeues a signal before it returns to the usermode to re-execute >> and enters the signal handler which can hit another uprobe. > > ach, those signals make everything complicated. I though signals are > blocked until the single step is done Yes, see uprobe_deny_signal(). > but my test just showed my > something different. I guess you missed the UTASK_SSTEP_TRAPPED logic. But this doesn't matter. Surely we must not "block" signals _after_ the single step is done, and this is the problem. > Okay, what now? IMHO: don't do this ;) > Blocking signals isn't probably a good idea. This is bad and wrong idea, I think. And, once again. Whatever you do, you can race with uprobe_register(). I mean, you must never expect that the task will hit the same uprobe again, even if you are going to re-execute the same insn. Oleg. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/