Re: [patch] x86: test case for the RODATA config option

[Arjan van de Ven <arjan@infradead.org>]

On 23 Jan 2008 10:09:58 +0100
Andi Kleen <andi@firstfloor.org> wrote:

> Arjan van de Ven <arjan@infradead.org> writes:
> 
> > This patch adds a test module for the DEBUG_RODATA config
> > option to make sure change_page_attr() did indeed make
> > "const" data read only.
> 
> The only that this does that is not done pretty much equivalently
> by pageattr-test.c (it just checks G instead of W) is testing the
> kernel text mapping on 64bit. On 32bit it is fully redundant I
> believe.

What it does is check if the rodata marking was succesful. That's more than just checking 
change_page_attr() itself, and basically a reasonably self self-check of mark_rodata_const().


> Testing the kernel mapping might be a good idea, but I would
> suggest adding it to pageattr-test.c

where we move the actual test function... I don't really care. Once pageattr-test.c
is merged I have absolutely no problem moving it there if it's the right place...
(one can argue that it should be in or near the mark_rodata_const() function instead
since it tests that one).
Just like the NX test I wrote yesterday (and will post later today) probably wants to live
somewhere close to the NX code.
 
> Yes I'm sure these basic facts will not stop it from being 
> applied anyways; just wanted to point it out for the benefit
> of the list readers.

Please spare me the bitter, undeserved sarcasm; I don't remember doing anything
to you to deserve that.
-- 
If you want to reach me at my work email, use arjan@linux.intel.com
For development, discussion and tips for power savings, 
visit http://www.lesswatts.org