[PATCH] mm: mempool: Factor out mempool_refill()

[PATCH] mm: mempool: Factor out mempool_refill()

[Zhi Wang]

This patch factors out mempool_refill() from mempool_resize(). It's reasonable
that the mempool user wants to refill the pool immdiately when it has chance
e.g. inside a sleepible context, so that next time in the IRQ context the pool
would have much more available elements to allocate.

After the refactor, mempool_refill() can also executes with mempool_resize()
/mempool_alloc/mempool_free() or another mempool_refill().

Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>
---
 include/linux/mempool.h |  1 +
 mm/mempool.c            | 61 ++++++++++++++++++++++++++++++++++++-------------
 2 files changed, 46 insertions(+), 16 deletions(-)

diff --git a/include/linux/mempool.h b/include/linux/mempool.h
index 69b6951..71f7460 100644
--- a/include/linux/mempool.h
+++ b/include/linux/mempool.h
@@ -30,6 +30,7 @@ extern mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
 			gfp_t gfp_mask, int nid);
 
 extern int mempool_resize(mempool_t *pool, int new_min_nr);
+extern void mempool_refill(mempool_t *pool);
 extern void mempool_destroy(mempool_t *pool);
 extern void * mempool_alloc(mempool_t *pool, gfp_t gfp_mask);
 extern void mempool_free(void *element, mempool_t *pool);
diff --git a/mm/mempool.c b/mm/mempool.c
index 004d42b..139c477 100644
--- a/mm/mempool.c
+++ b/mm/mempool.c
@@ -223,6 +223,47 @@ mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
 EXPORT_SYMBOL(mempool_create_node);
 
 /**
+ * mempool_refill - refill an existing memory pool immediately
+ * @pool:       pointer to the memory pool which was allocated via
+ *              mempool_create().
+ *
+ * This function tries to refill the pool with new elements
+ * immediately. Similar with mempool_resize(), it cannot be
+ * guaranteed that the pool will be fully filled immediately.
+ *
+ * Note, the caller must guarantee that no mempool_destroy is called
+ * while this function is running. mempool_alloc() & mempool_free()
+ * might be called (eg. from IRQ contexts) while this function executes.
+ */
+void mempool_refill(mempool_t *pool)
+{
+	void *element;
+	unsigned long flags;
+
+	spin_lock_irqsave(&pool->lock, flags);
+	if (pool->curr_nr >= pool->min_nr) {
+		spin_unlock_irqrestore(&pool->lock, flags);
+		return;
+	}
+
+	while (pool->curr_nr < pool->min_nr) {
+		spin_unlock_irqrestore(&pool->lock, flags);
+		element = pool->alloc(GFP_KERNEL, pool->pool_data);
+		if (!element)
+			return;
+		spin_lock_irqsave(&pool->lock, flags);
+		if (pool->curr_nr < pool->min_nr) {
+			add_element(pool, element);
+		} else {
+			spin_unlock_irqrestore(&pool->lock, flags);
+			pool->free(element, pool->pool_data);	/* Raced */
+			return;
+		}
+	}
+}
+EXPORT_SYMBOL(mempool_refill);
+
+/**
  * mempool_resize - resize an existing memory pool
  * @pool:       pointer to the memory pool which was allocated via
  *              mempool_create().
@@ -256,7 +297,8 @@ int mempool_resize(mempool_t *pool, int new_min_nr)
 			spin_lock_irqsave(&pool->lock, flags);
 		}
 		pool->min_nr = new_min_nr;
-		goto out_unlock;
+		spin_unlock_irqrestore(&pool->lock, flags);
+		goto out;
 	}
 	spin_unlock_irqrestore(&pool->lock, flags);
 
@@ -279,22 +321,9 @@ int mempool_resize(mempool_t *pool, int new_min_nr)
 	pool->elements = new_elements;
 	pool->min_nr = new_min_nr;
 
-	while (pool->curr_nr < pool->min_nr) {
-		spin_unlock_irqrestore(&pool->lock, flags);
-		element = pool->alloc(GFP_KERNEL, pool->pool_data);
-		if (!element)
-			goto out;
-		spin_lock_irqsave(&pool->lock, flags);
-		if (pool->curr_nr < pool->min_nr) {
-			add_element(pool, element);
-		} else {
-			spin_unlock_irqrestore(&pool->lock, flags);
-			pool->free(element, pool->pool_data);	/* Raced */
-			goto out;
-		}
-	}
-out_unlock:
 	spin_unlock_irqrestore(&pool->lock, flags);
+
+	mempool_refill(pool);
 out:
 	return 0;
 }
-- 
1.9.1

--
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:"dont@kvack.org"> email@kvack.org </a>

[PATCH] mm: mempool: Factor out mempool_refill()

[Wang, Zhi A]

This patch factors out mempool_refill() from mempool_resize(). It's reasonable
that the mempool user wants to refill the pool immdiately when it has chance
e.g. inside a sleepible context, so that next time in the IRQ context the pool
would have much more available elements to allocate.

After the refactor, mempool_refill() can also executes with mempool_resize()
/mempool_alloc/mempool_free() or another mempool_refill().

Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>
---
 include/linux/mempool.h |  1 +
 mm/mempool.c            | 61 ++++++++++++++++++++++++++++++++++++-------------
 2 files changed, 46 insertions(+), 16 deletions(-)

diff --git a/include/linux/mempool.h b/include/linux/mempool.h
index 69b6951..71f7460 100644
--- a/include/linux/mempool.h
+++ b/include/linux/mempool.h
@@ -30,6 +30,7 @@ extern mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
 			gfp_t gfp_mask, int nid);
 
 extern int mempool_resize(mempool_t *pool, int new_min_nr);
+extern void mempool_refill(mempool_t *pool);
 extern void mempool_destroy(mempool_t *pool);
 extern void * mempool_alloc(mempool_t *pool, gfp_t gfp_mask);
 extern void mempool_free(void *element, mempool_t *pool);
diff --git a/mm/mempool.c b/mm/mempool.c
index 004d42b..139c477 100644
--- a/mm/mempool.c
+++ b/mm/mempool.c
@@ -223,6 +223,47 @@ mempool_t *mempool_create_node(int min_nr, mempool_alloc_t *alloc_fn,
 EXPORT_SYMBOL(mempool_create_node);
 
 /**
+ * mempool_refill - refill an existing memory pool immediately
+ * @pool:       pointer to the memory pool which was allocated via
+ *              mempool_create().
+ *
+ * This function tries to refill the pool with new elements
+ * immediately. Similar with mempool_resize(), it cannot be
+ * guaranteed that the pool will be fully filled immediately.
+ *
+ * Note, the caller must guarantee that no mempool_destroy is called
+ * while this function is running. mempool_alloc() & mempool_free()
+ * might be called (eg. from IRQ contexts) while this function executes.
+ */
+void mempool_refill(mempool_t *pool)
+{
+	void *element;
+	unsigned long flags;
+
+	spin_lock_irqsave(&pool->lock, flags);
+	if (pool->curr_nr >= pool->min_nr) {
+		spin_unlock_irqrestore(&pool->lock, flags);
+		return;
+	}
+
+	while (pool->curr_nr < pool->min_nr) {
+		spin_unlock_irqrestore(&pool->lock, flags);
+		element = pool->alloc(GFP_KERNEL, pool->pool_data);
+		if (!element)
+			return;
+		spin_lock_irqsave(&pool->lock, flags);
+		if (pool->curr_nr < pool->min_nr) {
+			add_element(pool, element);
+		} else {
+			spin_unlock_irqrestore(&pool->lock, flags);
+			pool->free(element, pool->pool_data);	/* Raced */
+			return;
+		}
+	}
+}
+EXPORT_SYMBOL(mempool_refill);
+
+/**
  * mempool_resize - resize an existing memory pool
  * @pool:       pointer to the memory pool which was allocated via
  *              mempool_create().
@@ -256,7 +297,8 @@ int mempool_resize(mempool_t *pool, int new_min_nr)
 			spin_lock_irqsave(&pool->lock, flags);
 		}
 		pool->min_nr = new_min_nr;
-		goto out_unlock;
+		spin_unlock_irqrestore(&pool->lock, flags);
+		goto out;
 	}
 	spin_unlock_irqrestore(&pool->lock, flags);
 
@@ -279,22 +321,9 @@ int mempool_resize(mempool_t *pool, int new_min_nr)
 	pool->elements = new_elements;
 	pool->min_nr = new_min_nr;
 
-	while (pool->curr_nr < pool->min_nr) {
-		spin_unlock_irqrestore(&pool->lock, flags);
-		element = pool->alloc(GFP_KERNEL, pool->pool_data);
-		if (!element)
-			goto out;
-		spin_lock_irqsave(&pool->lock, flags);
-		if (pool->curr_nr < pool->min_nr) {
-			add_element(pool, element);
-		} else {
-			spin_unlock_irqrestore(&pool->lock, flags);
-			pool->free(element, pool->pool_data);	/* Raced */
-			goto out;
-		}
-	}
-out_unlock:
 	spin_unlock_irqrestore(&pool->lock, flags);
+
+	mempool_refill(pool);
 out:
 	return 0;
 }
-- 
1.9.1

--
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:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] mm: mempool: Factor out mempool_refill()

[Johannes Weiner]

On Mon, Dec 14, 2015 at 11:09:43AM +0000, Wang, Zhi A wrote:
> This patch factors out mempool_refill() from mempool_resize(). It's reasonable
> that the mempool user wants to refill the pool immdiately when it has chance
> e.g. inside a sleepible context, so that next time in the IRQ context the pool
> would have much more available elements to allocate.
> 
> After the refactor, mempool_refill() can also executes with mempool_resize()
> /mempool_alloc/mempool_free() or another mempool_refill().
> 
> Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>

Who is going to call that function? Adding a new interace usually
comes with a user, or as part of a series that adds users.

--
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:"dont@kvack.org"> email@kvack.org </a>

Re: [PATCH] mm: mempool: Factor out mempool_refill()

[Zhi Wang]

Hi Johannes:
     Thanks for the reply. In the end of the mempool_resize(), it will 
call the mempool_refill() to do the rest of the work. So this is not one 
of the "no-caller" case. If you insist this is a "no-caller" case, 
perhaps I should change it to a "static" function without exposing a new 
interface?

Personally I think mempool_refill() should be one of the typical 
interfaces in an implementation of a mempool. Currently the mempool will 
not grow only if pool->min_nr > new_min_nr.

So when user wants to refill the mempool immediately, not resize a 
mempool, in the current implementation, it has to do 2x 
mempool_resize(). First one is mempool_resize(pool->min_nr - 1), second 
one is mempool_resize(new_min_nr). So the refill action would truly 
happen. This is ugly and not convenient.

On 12/16/15 05:26, Johannes Weiner wrote:
> On Mon, Dec 14, 2015 at 11:09:43AM +0000, Wang, Zhi A wrote:
>> This patch factors out mempool_refill() from mempool_resize(). It's reasonable
>> that the mempool user wants to refill the pool immdiately when it has chance
>> e.g. inside a sleepible context, so that next time in the IRQ context the pool
>> would have much more available elements to allocate.
>>
>> After the refactor, mempool_refill() can also executes with mempool_resize()
>> /mempool_alloc/mempool_free() or another mempool_refill().
>>
>> Signed-off-by: Zhi Wang <zhi.a.wang@intel.com>
>
> Who is going to call that function? Adding a new interace usually
> comes with a user, or as part of a series that adds users.
>

--
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:"dont@kvack.org"> email@kvack.org </a>