[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20200619010144.GH40119@google.com>
Date: Thu, 18 Jun 2020 21:01:44 -0400
From: Joel Fernandes <joel@...lfernandes.org>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: "Paul E. McKenney" <paulmck@...nel.org>,
linux-kernel@...r.kernel.org, Davidlohr Bueso <dave@...olabs.net>,
Ingo Molnar <mingo@...hat.com>,
Josh Triplett <josh@...htriplett.org>,
Lai Jiangshan <jiangshanlai@...il.com>,
Marco Elver <elver@...gle.com>,
Mathieu Desnoyers <mathieu.desnoyers@...icios.com>,
rcu@...r.kernel.org, "Uladzislau Rezki (Sony)" <urezki@...il.com>
Subject: Re: [PATCH 3/7] rcu/trace: Add name of the source for gp_seq
On Thu, Jun 18, 2020 at 08:01:38PM -0400, Steven Rostedt wrote:
> On Thu, 18 Jun 2020 15:19:01 -0700
> "Paul E. McKenney" <paulmck@...nel.org> wrote:
>
> > For future reference, the TPS() around strings is not optional. Without
> > it, trace messages from crash dumps are garbled, if I remember correctly.
>
> When you pass in a string like this, only the pointer to the string is
> saved in the ring buffer. User space tools have no idea what those
> pointers are. The TPS() around strings maps those pointers to the
> string and shows them in the /sys/kernel/tracing/printk_formats file,
> such that perf and trace-cmd know how to make sense of those strings.
Makes sense. Quite a valuable feature!
thanks,
- Joel
Powered by blists - more mailing lists