Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S934770AbeAKPix (ORCPT + 1 other); Thu, 11 Jan 2018 10:38:53 -0500 Received: from merlin.infradead.org ([205.233.59.134]:39764 "EHLO merlin.infradead.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932421AbeAKPiu (ORCPT ); Thu, 11 Jan 2018 10:38:50 -0500 Date: Thu, 11 Jan 2018 16:38:45 +0100 From: Peter Zijlstra To: Vince Weaver Cc: Ingo Molnar , linux-kernel@vger.kernel.org, Ingo Molnar , Arnaldo Carvalho de Melo , Thomas Gleixner Subject: Re: perf: perf_fuzzer quickly locks up on 4.15-rc7 Message-ID: <20180111153845.GD6176@hirez.programming.kicks-ass.net> References: <20180108173005.lkglqrixb2ota6g2@gmail.com> <20180109102507.GG6176@hirez.programming.kicks-ass.net> <20180109132602.GA2369@hirez.programming.kicks-ass.net> <20180111091353.GW6176@hirez.programming.kicks-ass.net> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.9.2 (2017-12-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Return-Path: On Thu, Jan 11, 2018 at 10:26:14AM -0500, Vince Weaver wrote: > On Thu, 11 Jan 2018, Peter Zijlstra wrote: > > > I'm seeing things like: > > > > Cannot open /sys/kernel/tracing/kprobe_events > > > > this is likely caused by me not having anything mounted there. Rostedt > > provided the magic incantation to make that work, I'll go try now. > > The perf_fuzzer krpobe code is possibly really out of date, I haven't > really kept up with recent developments and I don't usually fuzz as root. I was thinking that running as root should be worse than user.. I can switch to fuzzing as user.. who knows. > > At least I seem to have cured all lockdep splats.. of course, the moment > > I hit send on this someone bad is bound to happen. > > Should I be trying any of those patches? Can't hurt, but I doubt it'll help sone, nothing there is PTI specific. The force_early_printk stuff helps with printk deadlocks though, hit a bunch of them earlier today when I build a kernel without them in. > Currently I can still lock my machines pretty quickly. I have been doing > some more tests. Machines don't crash if paranoid is "2" but will if > paranoid is "1". OK, I'm going to try fuzzing as a user with paranoid=0, and if that doesn't help, I'm going to switch to linus' tree with my patches on.