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;
Date:   Fri, 17 Jun 2022 16:41:05 +0200
From:   "Jason A. Donenfeld" <Jason@zx2c4.com>
To:     Petr Mladek <pmladek@suse.com>
Cc:     John Ogness <john.ogness@linutronix.de>,
        Marco Elver <elver@google.com>, linux-kernel@vger.kernel.org
Subject: Re: 5.19 printk breaks message ordering
Message-ID: <YqySgdH3sGj7o9O8@zx2c4.com>
References: <YqyANveL50uxupfQ@zx2c4.com>
 <YqyN20jpRw1SaaTw@alley>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <YqyN20jpRw1SaaTw@alley>
Precedence: bulk

On Fri, Jun 17, 2022 at 04:21:15PM +0200, Petr Mladek wrote:
> In fact, in your case you might get even better synchronization
> if you do it the other way and write userspace messages into
> the kernel log via /dev/kmsg:
> 
>      echo "Hello world" > /dev/kmsg

The idea is to interleave stderr/stdout properly, as CI has done for
ages, which threaded printk breaks, and which my patch (now posted)
fixes behind a default-off option. You can't dup2 /dev/kmsg to 1 and 2,
as the write() semantics are different, with respect to buffering and
with respect to new lines. The interfaces aren't comparable, so this
isn't quite a solution. (Besides, not breaking things as they currently
exist has a certain benefit.)

Jason