Date: Tue, 11 Oct 2016 23:05:56 -0300
From: Mauro Carvalho Chehab <mchehab@s-opensource.com>
To: Joe Perches <joe@perches.com>
Cc: LKML <linux-kernel@vger.kernel.org>,
        Mauro Carvalho Chehab <mchehab@kernel.org>,
        "Theodore Ts'o" <tytso@mit.edu>
Subject: Re: fyi: Recent commits require printk("fmt", ...) ->
 printk(PR_CONT/pr_cont( conversions
Message-ID: <20161011230556.6afe2764@vento.lan>
In-Reply-To: <1476235782.2116.5.camel@perches.com>
References: <1476235782.2116.5.camel@perches.com>
Organization: Samsung
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8BIT
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 2973
Lines: 79

Em Tue, 11 Oct 2016 18:29:42 -0700
Joe Perches <joe@perches.com> escreveu:

> Linus merged this recently (unfortunately without cc'ing LKML)
> 
> ----------------------------------------------------------------------------
> commit 563873318d328d9bbab4b00dfd835ac7c7e28697
> Merge: 24532f768121 bfd8d3f23b51
> Author: Linus Torvalds <torvalds@linux-foundation.org>
> Date:   Mon Oct 10 09:29:50 2016 -0700
> 
> Merge branch 'printk-cleanups'
> 
> Merge my system logging cleanups, triggered by the broken '\n' patches.
> 
> The line continuation handling has been broken basically forever, and
> the code to handle the system log records was both confusing and
> dubious.  And it would do entirely the wrong thing unless you always had
> a terminating newline, partly because it couldn't actually see whether a
> message was marked KERN_CONT or not (but partly because the LOG_CONT
> handling in the recording code was rather confusing too).
> 
> 
> This re-introduces a real semantically meaningful KERN_CONT, and fixes
> the few places I noticed where it was missing.  There are probably more
> missing cases, since KERN_CONT hasn't actually had any semantic meaning
> for at least four years (other than the checkpatch meaning of "no log
> level necessary, this is a continuation line").
> ----------------------------------------------------------------------------
> 
> These changes are not bad as they allow a message to remove the
> now unnecessary terminating newline or not have the newline
> added in the first place.  But there is also a modified behavior
> to existing logging message code.
> 
> Given the new behavior of bare printk that continues a line now
> requires a PR_CONT.  There are several places that will need update
> to have the same logging output as before this series.
> 
> For instance:
> 
> fs/ext4/super.c
> mm/page_alloc.c
> 
> And many files in these directories:
> 
> drivers/media/
> drivers/gpu/
> drivers/block/
> 
> and other directories as well.'
> 
> Many drivers for ancient and out of production hardware, and the old, crufty,
> and unmaintained drivers also have this issue, but likely no one will be
> impacted by this change to those driver's logging code.

Thanks for warning! Yeah, there are lots of media drivers written before
we start using KERN_CONT for continuation on media. On a rough estimation,
I'd say we have about 100 drivers with such stuff:

$ git grep -E '\bprintk' drivers/media/|grep -v KERN_|grep -v '\\n'|cut -d: -f1|sort|uniq|wc -l
97

But the above grep doesn't catch things like:

		printk(DRIVER_NAME ": Device must be connected to a high-speed"
		       " USB 2.0 port.\n");

(with should actually be calling pr_info instead of printk)

Yet, there are a lot of drivers to check.

Thankfully, most (if not all)  of KERNEL_CONT cases are used for
debugging purposes. So, normal users shouldn't notice it.

I'll seek for some time to review it, at least on the most used drivers.

Thanks,
Mauro