Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1755138AbXFUIZR (ORCPT ); Thu, 21 Jun 2007 04:25:17 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1751326AbXFUIZG (ORCPT ); Thu, 21 Jun 2007 04:25:06 -0400 Received: from mail.screens.ru ([213.234.233.54]:47864 "EHLO mail.screens.ru" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751586AbXFUIZF (ORCPT ); Thu, 21 Jun 2007 04:25:05 -0400 Date: Thu, 21 Jun 2007 12:25:09 +0400 From: Oleg Nesterov To: Linus Torvalds Cc: Benjamin Herrenschmidt , Davide Libenzi , Nicholas Miell , Linux Kernel Mailing List Subject: Re: Fix signalfd interaction with thread-private signals Message-ID: <20070621082509.GA88@tv-sign.ru> References: <1182125303.3794.8.camel@entropy> <1182127391.26853.207.camel@localhost.localdomain> <20070619091452.GA94@tv-sign.ru> <1182254988.26853.334.camel@localhost.localdomain> <20070619140646.GB27343@tv-sign.ru> <1182295473.26853.386.camel@localhost.localdomain> <20070620111415.GA91@tv-sign.ru> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.11 Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1283 Lines: 38 On 06/20, Linus Torvalds wrote: > > On Wed, 20 Jun 2007, Oleg Nesterov wrote: > > > > Also, suppose that some thread does > > > > for (;;) > > signal(SIGSEGV, SIG_IGN); > > > > Now we have the same situation. do_sigaction() can steal SIGSEGV from > > another thread. > > Actually, that shouldn't be possible. > > See "force_sig_info()". It does not allow blocking or ignoring forced > signals. We will reset such a signal handler to SIG_DFL, and unlock it. > > So if you get a SIGSEGV while SIGSEGV's are blocked or ignored, the kernel > *will* kill you. No questions asked. Yes, and no. Yes, force_sig() unblocks and un-ignores the signal. However, unlike group-wide signals, thread-specific signals do not convert themselves to SIGKILL on delivery. The target thread should dequeue SIGSEGV and then it calls do_group_exit(). Before it does so, another thread doing signal(SIGSEGV, SIG_IGN) can steal the signal. Of course, this is unlikely, and the target thread will take page fault again. The same for signalfd. 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/