Re: Fedora Core 4 Test Update: kernel-2.6.14-1.1633_FC4

[Dave Jones <davej redhat com>]

On Thu, Nov 03, 2005 at 05:31:53PM -0600, Marc Schwartz (via MN) wrote:

 > >From kernel-2.6.12-1.1456_FC4:
 > 
 > $ cat /proc/i8k
 > 1.0 A38 ? 54 -22 1 -22 79260 -1 2
 > 
 > 
 > >From kernel-2.6.14-1.1633_FC4:
 > 
 > $ cat /proc/i8k
 > 1.0 A38  52 -22 1 -22 77340 -1 2

Ok, that looks like unintentional breakage. I've brought this
up upstream. Will try and get any resulting fix backported to FC4
when that appears.

 > What is different is the field that should show the service tag ID for
 > the laptop. This is the '?' in the first instance (3rd field) and
 > appears to be missing in the second.
 > 
 > This would perhaps explain the SEGFAULT, since in reviewing the code in
 > i8krellm.c for the __i8k_get_proc_info() function, there are a series of
 > strcpy's, parsing the fields from the result of getting the /proc/i8k
 > info.

That code is really quite icky, and screams "scanf" to me.
Whilst that kernel interface *shouldnt* change, things like this
do happen, and it could have taken defensive action instead of
being as fragile as it is.

		Dave