From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759483AbcLUJWY (ORCPT ); Wed, 21 Dec 2016 04:22:24 -0500 Received: from pandora.armlinux.org.uk ([78.32.30.218]:47104 "EHLO pandora.armlinux.org.uk" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757829AbcLUJWU (ORCPT ); Wed, 21 Dec 2016 04:22:20 -0500 Date: Wed, 21 Dec 2016 09:21:59 +0000 From: Russell King - ARM Linux To: AMIT NAGAL Cc: Maninder Singh , "catalin.marinas@arm.com" , "will.deacon@arm.com" , "akpm@linux-foundation.org" , "kirill.shutemov@linux.intel.com" , "mark.rutland@arm.com" , "jack@suse.cz" , "linux-kernel@vger.kernel.org" , Ajeet Kumar Yadav , "shijie.huang@arm.com" , "james.morse@arm.com" , "sandeepa.s.prabhu@gmail.com" , "labbott@redhat.com" , "linux-arm-kernel@lists.infradead.org" , PANKAJ MISHRA , Vaneet Narang Subject: Re: Re: [PATCH 1/1] mm: call force_sig_info before prints Message-ID: <20161221092158.GT14217@n2100.armlinux.org.uk> References: <1482143832-11158-1-git-send-email-maninder1.s@samsung.com> <20161221055911epcms5p1146cee9212cc17fa8d5b18c49a085def@epcms5p1> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20161221055911epcms5p1146cee9212cc17fa8d5b18c49a085def@epcms5p1> User-Agent: Mutt/1.5.23 (2014-03-12) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Transfer-Encoding: 8bit X-MIME-Autoconverted: from quoted-printable to 8bit by mail.home.local id uBL9MWes014827 This is unreadable. Sorry. On Wed, Dec 21, 2016 at 05:59:11AM +0000, AMIT NAGAL wrote: > > > > > > > > > > > >

>

Hello,

On Mon, Dec 19, 2016 at 04:07:12PM +0530, Maninder Singh wrote:
> >> prints can delay queuing of signal, so better to print
> >>after force_sig_info.
> >> 
> >> Let's say process generated SIGSEGV , and some other thread sends
> >> SIGKILL to crashing process and it gets queued before SIGSEGV becuase
> >> of little delay due to prints so in this case coredump might not generate.
> 
> > In any case, that's going to be a race - you can't predict exactly when
> > the "other thread" will send the SIGKILL in relation to the SIGSEGV.
> 
> > So, I don't see the point of this change.
Actually SIGSEGV queueing gets delayed in __do_user_fault as logs are printed first(show_pte , show_regs).
meanwhile if SIGKILL gets queued , SIGSEGV wont be serviced first and coredump will be skipped . 
so our intention is to queue the SIGSEGV as early as possible . so that the probability of race between
SIGKILL and SIGSEGV can be minimized .
Amit
 

 


-- RMK's Patch system: http://www.armlinux.org.uk/developer/patches/ FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up according to speedtest.net.