Re: [PATCH 1 of 13] Add apply_to_page_range() which applies a function to a pte range

["Antonio Vargas" <windenntw@gmail.com>]

On 8/1/06, Eric W. Biederman <ebiederm@xmission.com> wrote:
> Jeremy Fitzhardinge <jeremy@xensource.com> writes:
>
> > 2 files changed, 99 insertions(+)
> > include/linux/mm.h |    5 ++
> > mm/memory.c        |   94 ++++++++++++++++++++++++++++++++++++++++++++++++++++
> >
> >
> > Add a new mm function apply_to_page_range() which applies a given
> > function to every pte in a given virtual address range in a given mm
> > structure. This is a generic alternative to cut-and-pasting the Linux
> > idiomatic pagetable walking code in every place that a sequence of
> > PTEs must be accessed.
> >
> > Although this interface is intended to be useful in a wide range of
> > situations, it is currently used specifically by several Xen
> > subsystems, for example: to ensure that pagetables have been allocated
> > for a virtual address range, and to construct batched special
> > pagetable update requests to map I/O memory (in ioremap()).
>
> - You don't handle huge pages.  For a generic function
>   that sounds like a problem.
> - I believe there is a reason the kernel doesn't already have
>   a function like this.  I seem to recall there being efficiency
>   and fast path arguments.

The proper trick for this is:

1. place you "for each page" code in a #define like so:

#define FOR_EACH_PAGE_INNER do{ ... code ... }while(0);

2. create your function in a separate .h file without the double-include guard

3. inside this code, exchange the indirect function call with your define name:

(*fn)(args); --> FOR_EACH_PAGE_INNER

4. document how the macro will receive certain variables from it's
outer scope, and should leave the "function result" in another one.

this in effect creates a different copy of the page walker for each
function, and inlines your code in it.. just like it would do with a
C++ template.

A place where you can see this technique working is the software
triangle filler from MESA.

The doubt is... is this acceptable regarding linux-kernel coding-style?


> - Placing this code in mm/memory.c without a common consumer is
>   pure kernel bloat for everyone who doesn't use this function,
>   which is just about everyone.
>


-- 
Greetz, Antonio Vargas aka winden of network

http://network.amigascne.org/
windNOenSPAMntw@gmail.com
thesameasabove@amigascne.org

Every day, every year
you have to work
you have to study
you have to scene.