[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20221205121119.537fa922bbd8147b9df5ae6c@kernel.org>
Date: Mon, 5 Dec 2022 12:11:19 +0900
From: Masami Hiramatsu (Google) <mhiramat@...nel.org>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: kernel test robot <lkp@...el.com>,
David Howells <dhowells@...hat.com>,
oe-kbuild-all@...ts.linux.dev, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] trace: Fix some checker warnings
On Sun, 4 Dec 2022 21:39:29 -0500
Steven Rostedt <rostedt@...dmis.org> wrote:
> On Mon, 5 Dec 2022 11:22:36 +0900
> Masami Hiramatsu (Google) <mhiramat@...nel.org> wrote:
>
> > diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c
> > index 5cfc95a52bc3..14f18edfe5bc 100644
> > --- a/kernel/trace/trace.c
> > +++ b/kernel/trace/trace.c
> > @@ -6572,7 +6572,8 @@ tracing_thresh_write(struct file *filp, const char __user *ubuf,
> > return ret;
> > }
> >
> > -#if defined(CONFIG_TRACER_MAX_TRACE) || defined(CONFIG_HWLAT_TRACER)
> > +#if defined(CONFIG_TRACER_MAX_TRACE) || defined(CONFIG_HWLAT_TRACER) \
> > + || defined(CONFIG_OSNOISE_TRACER)
> >
>
> When it gets this much, we need to put it into the trace.h header and
> define it. Actually, we have something that handles this too.
>
> #if (defined(CONFIG_TRACER_MAX_TRACE) || defined(CONFIG_HWLAT_TRACER) \
> || defined(CONFIG_OSNOISE_TRACER)) && defined(CONFIG_FSNOTIFY)
> #define LATENCY_FS_NOTIFY
> #endif
>
>
> We can add:
>
> #if (defined(CONFIG_TRACER_MAX_TRACE) || defined(CONFIG_HWLAT_TRACER) \
> || defined(CONFIG_OSNOISE_TRACER)) && defined(CONFIG_FSNOTIFY)
> #define LATENCY_FS_NOTIFY
> +#define USES_MAX_TRACE
> #endif
>
> And use that instead.
BTW, why can't HWLAT_TRACER and OSNOISE_TRACER depend on TRACER_MAX_TRACE?
I think it is better to reduce combinations of those, especially partially
enabling a feature seems a bit dangerous.
Thank you,
--
Masami Hiramatsu (Google) <mhiramat@...nel.org>
Powered by blists - more mailing lists