linux-kernel - Re: [lkp-robot] [include/linux/string.h] 6974f0c455: kernel_BUG_at

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]

Date:   Thu, 20 Jul 2017 22:48:01 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Ye Xiaolong <xiaolong.ye@...el.com>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Ananth N Mavinakayanahalli <ananth@...ux.vnet.ibm.com>,
        Anil S Keshavamurthy <anil.s.keshavamurthy@...el.com>,
        Masami Hiramatsu <mhiramat@...nel.org>,
        Daniel Micay <danielmicay@...il.com>,
        Arnd Bergmann <arnd@...db.de>,
        Mark Rutland <mark.rutland@....com>,
        Daniel Axtens <dja@...ens.net>,
        Rasmus Villemoes <linux@...musvillemoes.dk>,
        Andy Shevchenko <andriy.shevchenko@...ux.intel.com>,
        Chris Metcalf <cmetcalf@...hip.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        "H. Peter Anvin" <hpa@...or.com>, Ingo Molnar <mingo@...e.hu>,
        Andrew Morton <akpm@...ux-foundation.org>,
        LKML <linux-kernel@...r.kernel.org>, LKP <lkp@...org>
Subject: Re: [lkp-robot] [include/linux/string.h] 6974f0c455: kernel_BUG_at_lib/string.c

On Thu, Jul 20, 2017 at 6:59 PM, Ye Xiaolong <xiaolong.ye@...el.com> wrote:
> On 07/19, Linus Torvalds wrote:
>>Hmm. I wonder why the kernel test robot ends up having that annoying
>>line doubling for the dmesg.
>>
>
> Hmm, this line doubling issue should be caused by we set both
> 'earlyprintk=ttyS0,115200' and 'console=ttyS0,115200' in cmdline, after I
> remove any of it, this issue is gone, is it an inappropriate setting?

This is a weird behavior. Normally the boot console (in this case
"earlyser0") should be disabled when the regular console starts. In
your dmesg I see:

[    0.000000] bootconsole [earlyser0] enabled
...
[    0.003333] console [ttyS0] enabled
[    0.003333] console [ttyS0] enabled

This is where repeating starts, and I would have expected to see:

[    0.000000] bootconsole [earlyser0] enabled
...
[    0.000000] console [ttyS0] enabled
[    0.000000] console [ttyS0] enabled
[    0.000000] bootconsole [earlyser0] disabled
[    0.000000] bootconsole [earlyser0] disabled

And that would be the end of doubling.

Actually, with your command-line ("earlyprintk=ttyS0,115200
console=ttyS0,115200 console=tty0") I would expect:

[    0.000000] console [tty0] enabled
[    0.000000] bootconsole [earlyser0] disabled
[    0.000000] console [ttyS0] enabled

I don't see any mention of tty0, though. I wonder if that is confusing
the disable code. The only way I would normally expect the boot
console not to get unregistered is if ",keep" was appended to it...
The code is in kernel/printk/printk.c, FWIW.

Also, it looks like your kernel command line is cut off:

... earlyprintk=ttyS0,115200 console=ttyS0,115200 console=tty0 vga=

With your script showing "... vga=normal ..."

That's only 968 characters, well shy of x86's 2048 limit. Is this a
line length issue with qemu output, or does /proc/cmdline report the
same truncation?

-Kees

-- 
Kees Cook
Pixel Security