Commit 20a77776 authored by Nick Piggin's avatar Nick Piggin Committed by Linus Torvalds

[PATCH] mempool: simplify alloc

Mempool is pretty clever.  Looks too clever for its own good :) It
shouldn't really know so much about page reclaim internals.

- don't guess about what effective page reclaim might involve.

- don't randomly flush out all dirty data if some unlikely thing
  happens (alloc returns NULL). page reclaim can (sort of :P) handle
  it.

I think the main motivation is trying to avoid pool->lock at all costs.
However the first allocation is attempted with __GFP_WAIT cleared, so it
will be 'can_try_harder' if it hits the page allocator.  So if allocation
still fails, then we can probably afford to hit the pool->lock - and what's
the alternative?  Try page reclaim and hit zone->lru_lock?

A nice upshot is that we don't need to do any fancy memory barriers or do
(intentionally) racy access to pool-> fields outside the lock.
Signed-off-by: default avatarNick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: default avatarAndrew Morton <akpm@osdl.org>
Signed-off-by: default avatarLinus Torvalds <torvalds@osdl.org>
parent b84a35be
...@@ -198,36 +198,22 @@ void * mempool_alloc(mempool_t *pool, unsigned int __nocast gfp_mask) ...@@ -198,36 +198,22 @@ void * mempool_alloc(mempool_t *pool, unsigned int __nocast gfp_mask)
void *element; void *element;
unsigned long flags; unsigned long flags;
DEFINE_WAIT(wait); DEFINE_WAIT(wait);
int gfp_nowait; int gfp_temp;
might_sleep_if(gfp_mask & __GFP_WAIT);
gfp_mask |= __GFP_NOMEMALLOC; /* don't allocate emergency reserves */ gfp_mask |= __GFP_NOMEMALLOC; /* don't allocate emergency reserves */
gfp_mask |= __GFP_NORETRY; /* don't loop in __alloc_pages */ gfp_mask |= __GFP_NORETRY; /* don't loop in __alloc_pages */
gfp_mask |= __GFP_NOWARN; /* failures are OK */ gfp_mask |= __GFP_NOWARN; /* failures are OK */
gfp_nowait = gfp_mask & ~(__GFP_WAIT | __GFP_IO);
might_sleep_if(gfp_mask & __GFP_WAIT); gfp_temp = gfp_mask & ~(__GFP_WAIT|__GFP_IO);
repeat_alloc: repeat_alloc:
element = pool->alloc(gfp_nowait, pool->pool_data);
element = pool->alloc(gfp_temp, pool->pool_data);
if (likely(element != NULL)) if (likely(element != NULL))
return element; return element;
/*
* If the pool is less than 50% full and we can perform effective
* page reclaim then try harder to allocate an element.
*/
mb();
if ((gfp_mask & __GFP_FS) && (gfp_mask != gfp_nowait) &&
(pool->curr_nr <= pool->min_nr/2)) {
element = pool->alloc(gfp_mask, pool->pool_data);
if (likely(element != NULL))
return element;
}
/*
* Kick the VM at this point.
*/
wakeup_bdflush(0);
spin_lock_irqsave(&pool->lock, flags); spin_lock_irqsave(&pool->lock, flags);
if (likely(pool->curr_nr)) { if (likely(pool->curr_nr)) {
element = remove_element(pool); element = remove_element(pool);
...@@ -240,6 +226,8 @@ void * mempool_alloc(mempool_t *pool, unsigned int __nocast gfp_mask) ...@@ -240,6 +226,8 @@ void * mempool_alloc(mempool_t *pool, unsigned int __nocast gfp_mask)
if (!(gfp_mask & __GFP_WAIT)) if (!(gfp_mask & __GFP_WAIT))
return NULL; return NULL;
/* Now start performing page reclaim */
gfp_temp = gfp_mask;
prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE); prepare_to_wait(&pool->wait, &wait, TASK_UNINTERRUPTIBLE);
mb(); mb();
if (!pool->curr_nr) if (!pool->curr_nr)
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment