From mboxrd@z Thu Jan 1 00:00:00 1970 From: ebiederm@xmission.com (Eric W. Biederman) Subject: Re: [PATCH] kernel/signal: Signal-based pre-coredump notification Date: Tue, 16 Oct 2018 10:09:34 -0500 Message-ID: <87bm7ukjwx.fsf@xmission.com> References: <20181015120521.GA10146@redhat.com> <20398328-4ee1-96b2-5723-4b7eed55f0a2@cisco.com> <20181016141405.GA22045@redhat.com> Mime-Version: 1.0 Content-Type: text/plain Return-path: In-Reply-To: <20181016141405.GA22045@redhat.com> (Oleg Nesterov's message of "Tue, 16 Oct 2018 16:14:06 +0200") Sender: linux-kernel-owner@vger.kernel.org To: Oleg Nesterov Cc: Enke Chen , Thomas Gleixner , Ingo Molnar , Borislav Petkov , "H. Peter Anvin" , x86@kernel.org, Peter Zijlstra , Arnd Bergmann , Khalid Aziz , Kate Stewart , Helge Deller , Greg Kroah-Hartman , Al Viro , Andrew Morton , Christian Brauner , Catalin Marinas , Will Deacon , Dave Martin , Mauro Carvalho Chehab , Michal Hocko , Rik van Riel List-Id: linux-arch.vger.kernel.org Oleg Nesterov writes: > On 10/15, Enke Chen wrote: >> >> > I don't understand why we need valid_predump_signal() at all. >> >> Most of the signals have well-defined semantics, and would not be appropriate >> for this purpose. > > you are going to change the rules anyway. I will just add that CLD_XXX is only valid with SIGCHLD as they are signal specific si_codes. In conjunction with another signal like SIGUSR it will have another meaning. I would really appreciate it if new code does not further complicate siginfo_layout. >> That is why it is limited to only SIGCHLD, SIGUSR1, SIGUSR2. > > Which do not queue. So the parent won't get the 2nd signal if 2 children > crash at the same time. We do best effort queueing but we don't guarantee anything. So yes this makes signals a very louzy interface for sending this kind of information. >> >> if (sig_kernel_coredump(signr)) { >> >> + /* >> >> + * Notify the parent prior to the coredump if the >> >> + * parent is interested in such a notificaiton. >> >> + */ >> >> + int p_sig = current->real_parent->predump_signal; >> >> + >> >> + if (valid_predump_signal(p_sig)) { >> >> + read_lock(&tasklist_lock); >> >> + do_notify_parent_predump(current); >> >> + read_unlock(&tasklist_lock); >> >> + cond_resched(); >> > >> > perhaps this should be called by do_coredump() after coredump_wait() kills >> > all the sub-threads? >> >> proc_coredump_connector(current) is located here, they should stay together. > > Why? > > Once again, other threads are still alive. So if the parent restarts the service > after it recieves -predump_signal, the new process can "race" with the old thread. Yes. It isn't until do_coredump calls coredump_wait that all of the threads are killed. Eric From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from out03.mta.xmission.com ([166.70.13.233]:37294 "EHLO out03.mta.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1726760AbeJPXBm (ORCPT ); Tue, 16 Oct 2018 19:01:42 -0400 From: ebiederm@xmission.com (Eric W. Biederman) References: <20181015120521.GA10146@redhat.com> <20398328-4ee1-96b2-5723-4b7eed55f0a2@cisco.com> <20181016141405.GA22045@redhat.com> Date: Tue, 16 Oct 2018 10:09:34 -0500 In-Reply-To: <20181016141405.GA22045@redhat.com> (Oleg Nesterov's message of "Tue, 16 Oct 2018 16:14:06 +0200") Message-ID: <87bm7ukjwx.fsf@xmission.com> MIME-Version: 1.0 Content-Type: text/plain Subject: Re: [PATCH] kernel/signal: Signal-based pre-coredump notification Sender: linux-arch-owner@vger.kernel.org List-ID: To: Oleg Nesterov Cc: Enke Chen , Thomas Gleixner , Ingo Molnar , Borislav Petkov , "H. Peter Anvin" , x86@kernel.org, Peter Zijlstra , Arnd Bergmann , Khalid Aziz , Kate Stewart , Helge Deller , Greg Kroah-Hartman , Al Viro , Andrew Morton , Christian Brauner , Catalin Marinas , Will Deacon , Dave Martin , Mauro Carvalho Chehab , Michal Hocko , Rik van Riel , "Kirill A. Shutemov" , Roman Gushchin , Marcos Paulo de Souza , Dominik Brodowski , Cyrill Gorcunov , Yang Shi , Jann Horn , Kees Cook , linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org, "Victor Kamensky (kamensky)" , xe-linux-external@cisco.com, Stefan Strogin , Eugene Syromiatnikov Message-ID: <20181016150934.AaoBQW8koBM1dGdkaA1cQRcY3zrI9oMYnPmKfPsng7k@z> Oleg Nesterov writes: > On 10/15, Enke Chen wrote: >> >> > I don't understand why we need valid_predump_signal() at all. >> >> Most of the signals have well-defined semantics, and would not be appropriate >> for this purpose. > > you are going to change the rules anyway. I will just add that CLD_XXX is only valid with SIGCHLD as they are signal specific si_codes. In conjunction with another signal like SIGUSR it will have another meaning. I would really appreciate it if new code does not further complicate siginfo_layout. >> That is why it is limited to only SIGCHLD, SIGUSR1, SIGUSR2. > > Which do not queue. So the parent won't get the 2nd signal if 2 children > crash at the same time. We do best effort queueing but we don't guarantee anything. So yes this makes signals a very louzy interface for sending this kind of information. >> >> if (sig_kernel_coredump(signr)) { >> >> + /* >> >> + * Notify the parent prior to the coredump if the >> >> + * parent is interested in such a notificaiton. >> >> + */ >> >> + int p_sig = current->real_parent->predump_signal; >> >> + >> >> + if (valid_predump_signal(p_sig)) { >> >> + read_lock(&tasklist_lock); >> >> + do_notify_parent_predump(current); >> >> + read_unlock(&tasklist_lock); >> >> + cond_resched(); >> > >> > perhaps this should be called by do_coredump() after coredump_wait() kills >> > all the sub-threads? >> >> proc_coredump_connector(current) is located here, they should stay together. > > Why? > > Once again, other threads are still alive. So if the parent restarts the service > after it recieves -predump_signal, the new process can "race" with the old thread. Yes. It isn't until do_coredump calls coredump_wait that all of the threads are killed. Eric