[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20190326112958.2668a4a1@gandalf.local.home>
Date: Tue, 26 Mar 2019 11:29:58 -0400
From: Steven Rostedt <rostedt@...dmis.org>
To: "Paul E. McKenney" <paulmck@...ux.ibm.com>
Cc: Yafang Shao <laoar.shao@...il.com>, mingo@...hat.com,
peterz@...radead.org, josh@...htriplett.org,
mathieu.desnoyers@...icios.com, jiangshanlai@...il.com,
joel@...lfernandes.org, linux-kernel@...r.kernel.org,
shaoyafang@...iglobal.com
Subject: Re: [PATCH v2 3/3] rcu: validate arguments for rcu tracepoints
On Tue, 26 Mar 2019 08:18:15 -0700
"Paul E. McKenney" <paulmck@...ux.ibm.com> wrote:
> On Tue, Mar 26, 2019 at 08:13:11PM +0800, Yafang Shao wrote:
> > When CONFIG_RCU_TRACE is not set, all these tracepoints are defined as
> > do-nothing macro.
> > We'd better make those inline functions that take proper arguments.
> >
> > As RCU_TRACE() is defined as do-nothing marco as well when
> > CONFIG_RCU_TRACE is not set, so we can clean it up.
>
> How about this for the commit log?
>
> Unless the CONFIG_RCU_TRACE kconfig option is set, almost all
> of RCU's tracepoints are defined as empty macros. It would
> be better if these tracepoints could instead be empty inline
> functions with proper arguments and type checking. It would
> also be good to get rid of the RCU_TRACE() macro, which
> compiles its argument in CONFIG_RCU_TRACE=y kernels and
> omits them otherwise.
>
> This commit therefore creates a TRACE_EVENT_RCU macro that
> is defined as TRACE_EVENT in CONFIG_RCU_TRACE=y kernels and
> as the new TRACE_EVENT_NOP otherwise, which allows the
> empty macros and the RCU_TRACE() macro to be eliminated.
>
> With that:
>
> Reviewed-by: Paul E. McKenney <paulmck@...ux.ibm.com>
Yafang,
If you are OK with the above changes, I'll take this patch with the
updated change log.
-- Steve
Powered by blists - more mailing lists