Re: [PREEMPT_RT][PATCH] printk: Enhance the condition check of msleep in pr_flush()

From: Joe Perches <joe@perches.com>
To: chao.qin@intel.com, linux-kernel@vger.kernel.org,
	linux-rt-users@vger.kernel.org, bigeasy@linutronix.de,
	tglx@linutronix.de, john.ogness@linutronix.de,
	rostedt@goodmis.org
Cc: mgross@linux.intel.com, paul.mei@intel.com, lili.li@intel.com
Subject: Re: [PREEMPT_RT][PATCH] printk: Enhance the condition check of msleep in pr_flush()
Date: Tue, 20 Jul 2021 04:55:29 -0700	[thread overview]
Message-ID: <ea6c6a2b5482d62038d7d0a1d46b87665051222f.camel@perches.com> (raw)
In-Reply-To: <20210719022649.3444072-1-chao.qin@intel.com>

On Mon, 2021-07-19 at 10:26 +0800, chao.qin@intel.com wrote:
> From: Chao Qin <chao.qin@intel.com>
> 
> There is msleep in pr_flush(). If call WARN() in the early boot
> stage such as in early_initcall, pr_flush() will run into msleep
> when process scheduler is not ready yet. And then the system will
> sleep forever.
> 
> Before the system_state is SYSTEM_RUNNING, make sure DO NOT sleep
> in pr_flush().

Makes sense, thanks.

> diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c
[]
> @@ -3620,7 +3620,8 @@ bool pr_flush(int timeout_ms, bool reset_on_progress)
>  	u64 diff;
>  	u64 seq;
> 
> -	may_sleep = (preemptible() && !in_softirq());
> +	may_sleep = (preemptible() && !in_softirq()
> +			&& (system_state >= SYSTEM_RUNNING));

trivial style note:

Logic continuations are typically at the end of the previous line.
And there are few too many parentheses for my taste.

Maybe exceed 80 columns in a single line

	may_sleep = preemptible() && !in_softirq() && system_state >= SYSTEM_RUNNING;

or align the continuation

	may_sleep = (preemptible() && !in_softirq() &&
		     system_state >= SYSTEM_RUNNING);

or use individual lines

	may_sleep = (preemptible() &&
		     !in_softirq() &&
		     system_state >= SYSTEM_RUNNING);