Path: csiph.com!goblin2!goblin.stu.neva.ru!news.unit0.net!news.panservice.it!bofh.it!news.nic.it!robomod
From: Andy Lutomirski <luto@kernel.org>
Newsgroups: linux.kernel
Subject: Re: perf: race with automatic rdpmc() disabling
Date: Mon, 13 Mar 2017 22:10:03 +0100
Message-ID: <tkFo7-6nZ-29@gated-at.bofh.it>
References: <tkyFZ-1ik-29@gated-at.bofh.it> <tkBku-3fs-17@gated-at.bofh.it> <tkBu9-3kO-5@gated-at.bofh.it>
X-Original-To: Peter Zijlstra <peterz@infradead.org>
X-Gm-Message-State: AMke39mVcitfoZvfMonwlax3tBOwAsw+Z1Gt+EByA1FLIvpCA3zSz/05+oQu+xCaX9n1uMZsytyyKB7/DvfVo2Mo
X-Received: by 10.31.108.89 with SMTP id h86mr15617825vkc.27.1489439148142; Mon, 13 Mar 2017 14:05:48 -0700 (PDT)
MIME-Version: 1.0
X-Gmail-Original-Message-ID: <CALCETrVRzPwxnHmron_J+6nkBy24cm_G8kX=cxa5+qYoeQLcsw@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8
Sender: robomod@news.nic.it
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org
Approved: robomod@news.nic.it
Lines: 27
Organization: linux.* mail to news gateway
X-Original-Cc: Andy Lutomirski <luto@kernel.org>, Vince Weaver <vincent.weaver@maine.edu>, "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>, Ingo Molnar <mingo@redhat.com>, Arnaldo Carvalho de Melo <acme@kernel.org>
X-Original-Date: Mon, 13 Mar 2017 14:05:27 -0700
X-Original-Message-ID: <CALCETrVRzPwxnHmron_J+6nkBy24cm_G8kX=cxa5+qYoeQLcsw@mail.gmail.com>
X-Original-References: <alpine.DEB.2.20.1703130945120.12942@macbook-air> <CALCETrWoB9UNThOdbMjquA0uBhU74DqF2RB9ERX1Xi_ujyOwEA@mail.gmail.com> <20170313165507.GJ3312@twins.programming.kicks-ass.net>
X-Original-Sender: linux-kernel-owner@vger.kernel.org
Xref: csiph.com linux.kernel:1599823

On Mon, Mar 13, 2017 at 9:55 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> On Mon, Mar 13, 2017 at 09:44:02AM -0700, Andy Lutomirski wrote:
>> static void x86_pmu_event_mapped(struct perf_event *event)
>> {
>>     if (!(event->hw.flags & PERF_X86_EVENT_RDPMC_ALLOWED))
>>         return;
>>
>>     if (atomic_inc_return(&current->mm->context.perf_rdpmc_allowed) == 1)
>>
>> <-- thread 1 stalls here
>>
>>         on_each_cpu_mask(mm_cpumask(current->mm), refresh_pce, NULL, 1);
>> }
>>
>> Suppose you start with perf_rdpmc_allowed == 0.  Thread 1 runs
>> x86_pmu_event_mapped and gets preempted (or just runs slowly) where I
>> marked.  Then thread 2 runs the whole function, does *not* update CR4,
>> returns to userspace, and GPFs.
>>
>> The big hammer solution is to stick a per-mm mutex around it.  Let me
>> ponder whether a smaller hammer is available.
>
> Reminds me a bit of what we ended up with in kernel/jump_label.c:static_key_slow_inc().
>
>

One thing I don't get: isn't mmap_sem held for write the whole time?