Re: [RFC] per thread page reservation patch

[Nikita Danilov <nikita@clusterfs.com>]

Marcelo Tosatti <marcelo.tosatti@cyclades.com> writes:

[...]

>
> - all instances of an allocator from the current thread will eat from the perthread
>   reserves, you probably want only a few special allocations to eat from the reserves?
>   Thing is its not really a reservation intended for emergency situations,
>   rather a "generic per-thread pool" the way things are now.

Per-thread reservations are only useful when they are transparent. If
users have to call special function, or to pass special flag to
__alloc_pages() they might just use mempool as well. Idea is to reserve
some number of pages before starting complex operation so that generic
function (like find_get_page()) called as part of this operation are
guaranteed to succeed.

>
> - its a real fast path, we're adding quite some instructions there which are only
>   used by reiserfs now.

Yes, this worries me too. Possibly we should move this check below in
__alloc_pages(), so that per-thread reservation is only checked if
fast-path allocation failed.

Nikita.
--
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>