Free list initialization

Free list initialization

[Anand Eswaran]

Hi all:

  Could anybody point me out to the part of the mm code where the  zone
free-lists are initialized to the remaining system memory  just
subsequent to setting up of the zone structures . ( so that  say when
the very first time _alloc_pages executes, the system can use (
__alloc_pages ()  ->   rmqueue()  free-list to allocate the required
memory block.

  I dont seem to be able to find any such code in free_area_init_core().

  Im using a 2.4.18 kernel.

Thanks in advance,
-----
Anand
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: Free list initialization

[Martin J. Bligh]

>   Could anybody point me out to the part of the mm code where the  zone
> free-lists are initialized to the remaining system memory  just
> subsequent to setting up of the zone structures . ( so that  say when
> the very first time _alloc_pages executes, the system can use (
> __alloc_pages ()  ->   rmqueue()  free-list to allocate the required
> memory block.
> 
>   I dont seem to be able to find any such code in free_area_init_core().
> 
>   Im using a 2.4.18 kernel.

Suggest you start at free_all_bootmem. IIRC, basically we just call a 
free on every page we have, and the normal buddy free routines populate
the lists. Not very efficient, but who cares? ... it's boottime! ;-)

M.

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: Free list initialization

[William Lee Irwin III]

On Wed, Aug 06, 2003 at 04:12:46PM -0700, Martin J. Bligh wrote:
> Suggest you start at free_all_bootmem. IIRC, basically we just call a 
> free on every page we have, and the normal buddy free routines populate
> the lists. Not very efficient, but who cares? ... it's boottime! ;-)

If it ever turns out that someone does care, I've gone through and done
useful things like freeing higher-order pages at a time for both
bootmem.c and highmem (as isolated patches) I could resurrect if the
issue ever arises in the field.

The whole affair is still asymptotically O(pages) since the coremap
initialization still touches every struct page, so there aren't large
amounts of improvement that can be made without dynamically allocating
coremap elements (which raises more issues than it resolves, though
there are valid reasons to want to do it beyond this).


-- wli
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>

Re: Free list initialization

[Mel Gorman]

On Wed, 6 Aug 2003, Anand Eswaran wrote:

>   Could anybody point me out to the part of the mm code where the  zone
> free-lists are initialized to the remaining system memory  just
> subsequent to setting up of the zone structures .

Read:

http://www.csn.ul.ie/~mel/projects/vm/guide/html/understand/node42.html

and then

http://www.csn.ul.ie/~mel/projects/vm/guide/html/code/node9.html#SECTION00450300000000000000

-- 
Mel Gorman
http://www.csn.ul.ie/~mel
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"aart@kvack.org"> aart@kvack.org </a>