Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20221214233106.69b2c01b@gandalf.local.home> <Y5trUep9IvCv1Uwy@google.com>
 <20221215141146.6ceb7cf2@gandalf.local.home>
In-Reply-To: <20221215141146.6ceb7cf2@gandalf.local.home>
From:   Joel Fernandes <joel@joelfernandes.org>
Date:   Thu, 15 Dec 2022 14:52:48 -0500
Message-ID: <CAEXW_YQLtK=4LMJ+LHPVWU0wbV-027HJoCEKTjZvBZ6krrn6vw@mail.gmail.com>
Subject: Re: [RFC][PATCH] ACPI: tracing: Have ACPI debug go to tracing ring buffer
To:     Steven Rostedt <rostedt@goodmis.org>
Cc:     LKML <linux-kernel@vger.kernel.org>,
        Linux Trace Kernel <linux-trace-kernel@vger.kernel.org>,
        linux-acpi@vger.kernel.org,
        "Rafael J. Wysocki" <rafael@kernel.org>,
        Len Brown <lenb@kernel.org>,
        Masami Hiramatsu <mhiramat@kernel.org>,
        Brian Norris <briannorris@chromium.org>,
        Ross Zwisler <zwisler@google.com>,
        Ching-lin Yu <chinglinyu@google.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Thu, Dec 15, 2022 at 2:11 PM Steven Rostedt <rostedt@goodmis.org> wrote:
>
> On Thu, 15 Dec 2022 18:45:37 +0000
> Joel Fernandes <joel@joelfernandes.org> wrote:
>
> > Wouldn't it be better to also check trace_acpi_print_enabled() here in the
> > else if() condition, along with IS_ENABLED()? That way if the CONFIG is
> > enabled but the tracepoint is not enabled, at least the messages will go to
> > dmesg instead of skipped.
>
> I really don't want that. This was purposely done to be mutually exclusive.
> The reason I added this in the first place, is because too much enabled
> will render the system useless if printk() is used.
>
> After boot up, if I had enabled all debug events and then I were to disable
> the acpi tracepoint, it will likely render the system useless again if it
> were to switch over to printk.

Ok, sure. I see where you were going. So you want no debugging
messages at all if the trace event is disabled. That's fine with me. I
would also add a note about the need to enable the specific trace
event, in the Kconfig message and/or the Documentation. Otherwise, you
might get someone say, "hey I enabled the CONFIG option but I see
nothing in the trace buffer".

Another approach could be to always enable the trace event by default,
if the CONFIG is turned on. Or do a printk() telling the user about
the event to enable, so they know why their trace buffer is empty.

Up to you and the ACPI maintainers. ;-)

thanks,

 - Joel