[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20181130094523.1bc91052@gandalf.local.home>
Date: Fri, 30 Nov 2018 09:45:23 -0500
From: Steven Rostedt <rostedt@...dmis.org>
To: Jiri Olsa <jolsa@...hat.com>
Cc: LKML <linux-kernel@...r.kernel.org>,
Thomas Gleixner <tglx@...utronix.de>,
Ingo Molnar <mingo@...nel.org>,
Peter Zijlstra <peterz@...radead.org>,
"Luis Claudio R. Goncalves" <lclaudio@...g.org>,
Arnaldo Carvalho de Melo <acme@...nel.org>,
ldv@...linux.org, esyr@...hat.com,
Frederic Weisbecker <fweisbec@...il.com>,
Masami Hiramatsu <mhiramat@...nel.org>,
Namhyung Kim <namhyung@...nel.org>
Subject: Re: Rough idea of implementing blocking perf calls for system call
tracepoints
On Fri, 30 Nov 2018 15:01:58 +0100
Jiri Olsa <jolsa@...hat.com> wrote:
> > or perhaps just tracepoints.. does not seem to make much
> > sense for he events
>
> actualy the syscalls seems to be enough for now ;-) I tried
> something and ended up with hack below
Actually, we *only* want to do syscalls. We really don't want to add
blocking to trace events. That would be a nightmare for deadlocks and
such.
I was thinking about adding a PTRACE feature that makes this part of
PTRACE. That is, we start this via PTRACE and let PTRACE signal the
tracer process when the tracee process blocks, and it can then wake up
the tracee process.
A tracer must have PTRACE permissions anyway. The goal of this is to
speed up syscall tracing.
-- Steve
Powered by blists - more mailing lists