[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Date: Tue, 7 Aug 2018 18:17:42 -0700
From: Joel Fernandes <joelaf@...gle.com>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: Joel Fernandes <joel@...lfernandes.org>,
LKML <linux-kernel@...r.kernel.org>,
"Cc: Android Kernel" <kernel-team@...roid.com>,
Boqun Feng <boqun.feng@...il.com>,
Byungchul Park <byungchul.park@....com>,
Ingo Molnar <mingo@...hat.com>,
Masami Hiramatsu <mhiramat@...nel.org>,
Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
Namhyung Kim <namhyung@...nel.org>,
Paul McKenney <paulmck@...ux.vnet.ibm.com>,
Peter Zijlstra <peterz@...radead.org>,
Thomas Glexiner <tglx@...utronix.de>,
Tom Zanussi <tom.zanussi@...ux.intel.com>
Subject: Re: [PATCH v12 3/3] tracing: Centralize preemptirq tracepoints and
unify their usage
On Tue, Aug 7, 2018 at 5:48 PM, Steven Rostedt <rostedt@...dmis.org> wrote:
> On Tue, 07 Aug 2018 19:54:13 -0400
> Joel Fernandes <joelaf@...gle.com> wrote:
>
>
>> >OK, I hit this bug, but it's not because of the partial revert. This
>> >bug seems it needs to be another partial revert. I like you movement of
>> >the code, but I'm starting to doubt that we can use a trace event as a
>> >hook for critical areas yet. Well, not until we can use srcu in NMI.
>> >
>>
>> I sent a patch to use srcu for all tracepoints including nmi. That
>> patch also removes this warning and fixes the one other issue masami
>> reported (hot plug causing a warning).
>
> Is it one I can take, or does Paul have it? I'll need it to continue
> with this as is, because I can't pass my tests without triggering that
> warning (and that fails the test).
I think you could take it if you're Ok with it, its purely in the
tracing code and I tested it yesterday morning. Its attached to this
email. I tested that it fixes the mmio trace (hotplug related) issue..
>>
>> If Paul and Mathieu can confirm SRCU works on offline CPUs that would
>> be great.
>
> Yes please.
>
BTW I found this in Paul's docs RCU Requirements docs, "SRCU is insensitive
to whether or not a CPU is online" so I think it will work.
The paragraph was..
Also unlike other RCU flavors, SRCU's callbacks-wait function
srcu_barrier() may be invoked from CPU-hotplug notifiers,
though this is not necessarily a good idea.
The reason that this is possible is that SRCU is insensitive
to whether or not a CPU is online, which means that srcu_barrier()
need not exclude CPU-hotplug operations.
>>
>> It's just this one warning or anything else that makes you feel
>> tracepoints for critical paths isn't feasible?
>
> Currently I'm down to this, but I can't get past my first test in my
> ktest suite, because it fails here.
Ok hopefully this will get you past that. Sorry for the frustration.
View attachment "0001-tracepoint-Run-tracepoints-even-after-CPU-is-offline.patch" of type "text/x-patch" (6127 bytes)
Powered by blists - more mailing lists