Re: [stable] Linux 2.6.25.10 (resume)

[pageexec@freemail.hu]

On 23 Jul 2008 at 11:31, Henrique de Moraes Holschuh wrote:

> On Wed, 23 Jul 2008, pageexec@freemail.hu wrote:
> > it's apparently not true when foo = "kernel's security model", hence the
> > suggested change to reflect reality.
> 
> I heavily suggest using something else than "disclose".  
> 
> For the security community, "disclose" doesn't mean you have the source code
> for the buggy code and the source code for the fix.  It means you have the
> information that it is a "foo = kernel's security model" bug, and a
> description of the consequences of the bug for foo (the security model).
> 
> This is NOT what "disclose" means for the Linux kernel, right now.  Here,
> "disclose" means "you know there is a bug, you have the code, you have the
> bug fix".  But you don't know that "foo = kernel's security bug", or the
> consequences of the bug for the security model.

i think you misunderstood the whole thread here ;). we were explicitly
talking about bugs where the kernel devs *knew* they were fixing one
with an impact on security yet they chose not to say so.

determining whether a bug is a security one is a whole different kettle
of fish, that was not the topic here at all.

IOW, Documentation/SecurityBugs talks about bugs where the security impact
is known, not about bugs in general where such determination has yet to be
done.

> So just use another word, or properly qualify WHAT is going to be disclosed,
> (and in this case, WHAT is not going to be *usually* disclosed).