Re: [PATCH] printk: Add caller information to printk() output.

From: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>
To: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Cc: Sergey Senozhatsky <sergey.senozhatsky.work@gmail.com>,
	Petr Mladek <pmladek@suse.com>,
	Dmitry Vyukov <dvyukov@google.com>,
	Sergey Senozhatsky <sergey.senozhatsky@gmail.com>,
	Steven Rostedt <rostedt@goodmis.org>,
	Linus Torvalds <torvalds@linux-foundation.org>,
	Andrew Morton <akpm@linux-foundation.org>,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH] printk: Add caller information to printk() output.
Date: Tue, 4 Dec 2018 19:38:36 +0900	[thread overview]
Message-ID: <20181204103836.GB530@jagdpanzerIV> (raw)
In-Reply-To: <be668360-279b-5c24-4fc1-10ef050bd87f@i-love.sakura.ne.jp>

On (12/04/18 19:16), Tetsuo Handa wrote:
> On 2018/12/04 11:02, Sergey Senozhatsky wrote:
> > On (12/02/18 20:23), Tetsuo Handa wrote:
> >>
> >> Some examples for console output:
> >>
> >>   [    0.919699]@T1 x86: Booting SMP configuration:
> >>   [    4.152681]@T271 Fusion MPT base driver 3.04.20
> >>   [    5.070470]@C0 random: fast init done
> >>   [    6.587900]@C3 random: crng init done
> > 
> > This is hard to read. Let's have a fixed width space for from_id.
> 
> Console might be only 80 columns. When we have to check from screen capture of
> crash, triggering needless newlines due to occupying columns more than needed
> will not be nice. Also, syzbot has file size limit for saving console output.
> Wasting with needless spaces increases possibility of loosing past lines.
> Given that said, do you still want to waste precious columns with spaces?

Well, if the feature will be useful for non-syzbot cases as well, as
you mention below, then I'd say let's make the output human readable.
To some extent, at least.

I think we consume much more bytes on pr_fmt prefixes; and not all of them
are very useful all the time.

[..]
[    1.223872] nouveau 0000:01:00.0: DRM: DCB outp 00: 01000f02 00020030
[    1.224090] nouveau 0000:01:00.0: DRM: DCB outp 01: 02011f62 00020010
[..]
[    2.676030] snd_hda_codec_realtek hdaudioC0D0: autoconfig for ALC662 rev3: line_outs=1 (0x14/0x0/0x0/0x0/0x0) type:line
[    2.676643] snd_hda_codec_realtek hdaudioC0D0:    speaker_outs=0 (0x0/0x0/0x0/0x0/0x0)
[    2.677258] snd_hda_codec_realtek hdaudioC0D0:    hp_outs=1 (0x1b/0x0/0x0/0x0/0x0)
[    2.677879] snd_hda_codec_realtek hdaudioC0D0:    mono: mono_out=0x0
[    2.678578] snd_hda_codec_realtek hdaudioC0D0:    inputs:
[    2.679150] snd_hda_codec_realtek hdaudioC0D0:      Front Mic=0x19
[..]
[  430.768498] rndis_host 1-1:1.0 enp0s20f0u1: unregister 'rndis_host' usb-0000:00:14.0-1, RNDIS device

So, to me, PRINTK_FROM probably does not look like a really bad
serial console abuser. How many bytes do we need for PRINTK_FROM?
7-8?

> CONFIG_DEBUG_AID_FOR_SYZBOT is currently linux-next only option.
> This feature will be useful for non-syzbot environments.

	-ss