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: <CAMdYzYpF4FNTBPZsEFeWRuEwSies36QM_As8osPWZSr2q-viEA@mail.gmail.com>
 <87y1y48spg.fsf@jogness.linutronix.de> <CAMdYzYr-Wo713Y4qjboTpoK6GcrYfKCfRJAEizwXw6-=dymVzg@mail.gmail.com>
 <87zgihlbst.fsf@jogness.linutronix.de>
In-Reply-To: <87zgihlbst.fsf@jogness.linutronix.de>
From:   Peter Geis <pgwipeout@gmail.com>
Date:   Sun, 12 Jun 2022 19:30:27 -0400
Message-ID: <CAMdYzYosXiQc9=t7daPaWWS=rnTVT6nnZvVBXDycBQvfR-1FAA@mail.gmail.com>
Subject: Re: [BUG] Threaded printk breaks early debugging
To:     John Ogness <john.ogness@linutronix.de>
Cc:     Petr Mladek <pmladek@suse.com>,
        Sergey Senozhatsky <senozhatsky@chromium.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        "open list:ARM/Rockchip SoC..." <linux-rockchip@lists.infradead.org>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Sun, Jun 12, 2022 at 7:08 PM John Ogness <john.ogness@linutronix.de> wrote:
>
> On 2022-06-10, Peter Geis <pgwipeout@gmail.com> wrote:
> > This might be a side effect of the fact that this is on a low powered
> > arm64 board. I noticed with threading enabled during large bursts the
> > console drops an excessive amount of messages. It's especially
> > apparent during the handover from earlycon to the normal console.
>
> I guess you have a very small kernel buffer and are generating unusually
> high amounts of messages? Is there a reason you cannot use a larger
> buffer?

The buffer isn't an issue here, everything is available in dmesg when
userspace becomes available. Instead some messages bound for the
serial console are never output.

>
> Or maybe you are generating a constant amount of messages that a serial
> port could never keep up with unless it was interfering with the system
> in such a way as to slow everything down? This is exactly what printk
> should _not_ be used for.

We run a serial console at 1.5m baud which is significantly higher
than most SoCs which default to 115200. I have noticed some timing
differences since the introduction of the threaded console. A
significant amount of information is dumped very early in the boot
process (between 0 and 4 seconds into boot), as most drivers are
probing during this time. It also happens to be when the earlycon
console hands over to the normal console. There is no abnormal
debugging enabled, the output is a standard (non-quiet) boot log. The
question is why is direct mode not triggering during a panic?

>
> John Ogness