linux-kernel - Re: [PATCH] panic: fix incomplete panic log in panic()

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-Id: <20121011141845.1ed347df.akpm@linux-foundation.org>
Date:	Thu, 11 Oct 2012 14:18:45 -0700
From:	Andrew Morton <akpm@...ux-foundation.org>
To:	Qing Zhu <qzhu@...vell.com>
Cc:	mingo@...nel.org, ben@...adent.org.uk, markivx@...eaurora.org,
	ak@...ux.intel.com, linux-kernel@...r.kernel.org,
	cxie4@...vell.com, binw@...vell.com, wwang27@...vell.com,
	njun@...vell.com
Subject: Re: [PATCH] panic: fix incomplete panic log in panic()

On Thu, 11 Oct 2012 16:03:07 +0800
Qing Zhu <qzhu@...vell.com> wrote:

> Panic log should be printed on the console, but if someone lock the
> console when panic, console won't print out panic log.
> 
> The incomplete panic log issue will happen in below scenarios:
> 1. One task call console_lock(), then panic happend before it call
> console_unlock(). No panic log can be printed.
> 2. Cpu 0 call panic()->Cpu 1 call console_lock()->Cpu 0 call
> smp_send_stop()
> Cpu 1 will be stopped and can't unlock console,only top part of panic log
> will be printed.
> 
> So unlock console anyway in panic() to keep panic log printed.
> 
> Signed-off-by: Qing Zhu <qzhu@...vell.com>
> ---
>  kernel/panic.c |    8 ++++++++
>  1 files changed, 8 insertions(+), 0 deletions(-)
> 
> diff --git a/kernel/panic.c b/kernel/panic.c
> index e1b2822..3924d25 100644
> --- a/kernel/panic.c
> +++ b/kernel/panic.c
> @@ -23,6 +23,7 @@
>  #include <linux/init.h>
>  #include <linux/nmi.h>
>  #include <linux/dmi.h>
> +#include <linux/console.h>
>  
>  #define PANIC_TIMER_STEP 100
>  #define PANIC_BLINK_SPD 18
> @@ -127,6 +128,13 @@ void panic(const char *fmt, ...)
>  
>  	atomic_notifier_call_chain(&panic_notifier_list, 0, buf);
>  
> +	/*
> +	 * Unlock the console anyway here, in case it's occupied by another
> +	 * one which has no chance to unlock the console thus prevents the
> +	 * panic log prints on the console.
> +	 */
> +	console_unlock();
> +
>  	bust_spinlocks(0);
>  
>  	if (!panic_blink)

hm.  console_unlock() does a large amount of work, and it seems risky
to do all of that when the system is in a bad-and-getting-worse state.

Is there some more modest thing we could do here, for example,

	if (console_locked) {
		up(&console_sem);
		console_locked = 0;
	}

or something along those lines?

Also, perhaps this operation should be moved into bust_spinlocks(). 
What would have happened if your code had triggered an oops, rather
than called panic()?


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/