linux-kernel - Re: [PATCH 1/3] printk: Fix kdb_trap

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

Message-ID: <20171009150556.GA15129@pathway.suse.cz>
Date:   Mon, 9 Oct 2017 17:05:56 +0200
From:   Petr Mladek <pmladek@...e.com>
To:     Peter Zijlstra <peterz@...radead.org>
Cc:     sergey.senozhatsky@...il.com, linux-kernel@...r.kernel.org,
        rostedt@...dmis.org, mingo@...nel.org, tglx@...utronix.de,
        Jason Wessel <jason.wessel@...driver.com>
Subject: Re: [PATCH 1/3] printk: Fix kdb_trap_printk placement

On Thu 2017-10-05 15:42:27, Peter Zijlstra wrote:
> On Thu, Oct 05, 2017 at 03:38:44PM +0200, Petr Mladek wrote:
> > Or can the kdb console commands be called in NMI context?
> 
> IIRC most of KDB runs from NMI context.

To be honest, I am not familiar with kdb. I tried the following
from Documentation/dev-tools/kgdb.rst:

$> echo ttyS0 > /sys/module/kgdboc/parameters/kgdboc
$> echo g >/proc/sysrq-trigger

Result: I was able to do kdb commands on the serial console.
Note: It seems that this stuff was _not_ running in NMI.

Then I tried to set breakpoint for a function that is
called in NMI context:

$kdb> bt show_regs

Where show_regs() is called from nmi_cpu_backtrace(). I unblocked
the system and triggered ''l'' sysrq to show stacks from all CPUs:

$kdb> go
$> echo l >/proc/sysrq-trigger

Result: The system frozen and I had to reboot using a power switch.

I wonder if the break point in NMI is supposed to work and if the
kdb commands are handled in NMI context on the serial console then.

By other words, do you need this patch for a particular use-case?
Or did you added this patch just to fix a theoretical problem?

Best Regards,
Petr