Re: kernel: sysctl table check failed: /dev/parport/parport0/devices/ppdev0/timeslice Sysctl already exists

Previous thread: 活力健康減重!索取試用包 by kmlsgxrvlxzqpqb on Friday, May 30, 2008 - 10:54 am. (1 message)

Next thread: Bug in arch/i386/lib/delay.c file, delay_loop function by Jiri Hladky on Friday, May 30, 2008 - 11:15 am. (5 messages)
To: Linux Kernel <linux-kernel@...>
Cc: Eric W. Biederman <ebiederm@...>
Date: Friday, May 30, 2008 - 11:05 am

Well, that's umm, special.
I wonder if powering off the printer isn't causing the sysctl to be removed,
causing the next power-on to look like a new device ?
Eric, you poked around the sysctl paths in parport last year, any ideas?

Dave

--
http://www.codemonkey.org.uk

To: Dave Jones <davej@...>
Cc: Linux Kernel <linux-kernel@...>
Date: Friday, May 30, 2008 - 2:19 pm

Yes. The short answer is that this is a real bug in the parport code
that has existed forever. We just now complain.

http://bugzilla.kernel.org/show_bug.cgi?id=9749

It takes something like two accessors of the parallel port to trigger
this. Though not in a bad way.

In particular each open creates the sysctl for the duration of the
open (or something like that) and so if you open (I think it is
/dev/parport0) twice you try and create the sysctls twice.

I got as far as basic confirmation of this bug, and a rough cause of
it. Unfortunately I have not had the time to unscramble it, and neither
it seems has anyone else.

Eric
--

Previous thread: 活力健康減重!索取試用包 by kmlsgxrvlxzqpqb on Friday, May 30, 2008 - 10:54 am. (1 message)

Next thread: Bug in arch/i386/lib/delay.c file, delay_loop function by Jiri Hladky on Friday, May 30, 2008 - 11:15 am. (5 messages)