[PATCH 1/2] mempool: fix comments for mempool create functions

[PATCH 1/2] mempool: fix comments for mempool create functions

[Ferruh Yigit]

Fixes: 85226f9c526b ("mempool: introduce a function to create an empty pool")
Fixes: d1d914ebbc25 ("mempool: allocate in several memory chunks by default")

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 lib/librte_mempool/rte_mempool.c | 4 +---
 lib/librte_mempool/rte_mempool.h | 4 +---
 2 files changed, 2 insertions(+), 6 deletions(-)

diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
index df8b68c..e96d14f 100644
--- a/lib/librte_mempool/rte_mempool.c
+++ b/lib/librte_mempool/rte_mempool.c
@@ -908,9 +908,7 @@ rte_mempool_create(const char *name, unsigned n, unsigned elt_size,
 /*
  * Create the mempool over already allocated chunk of memory.
  * That external memory buffer can consists of physically disjoint pages.
- * Setting vaddr to NULL, makes mempool to fallback to original behaviour
- * and allocate space for mempool and it's elements as one big chunk of
- * physically continuos memory.
+ * Setting vaddr to NULL, makes mempool to fallback to original behaviour.
  */
 struct rte_mempool *
 rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
index 0243f9e..6fc331a 100644
--- a/lib/librte_mempool/rte_mempool.h
+++ b/lib/librte_mempool/rte_mempool.h
@@ -591,8 +591,6 @@ typedef void (rte_mempool_ctor_t)(struct rte_mempool *, void *);
  *
  * This function uses ``memzone_reserve()`` to allocate memory. The
  * pool contains n elements of elt_size. Its size is set to n.
- * All elements of the mempool are allocated together with the mempool header,
- * in one physically continuous chunk of memory.
  *
  * @param name
  *   The name of the mempool.
@@ -748,7 +746,7 @@ rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
  *
  * The mempool is allocated and initialized, but it is not populated: no
  * memory is allocated for the mempool elements. The user has to call
- * rte_mempool_populate_*() or to add memory chunks to the pool. Once
+ * rte_mempool_populate_*() to add memory chunks to the pool. Once
  * populated, the user may also want to initialize each object with
  * rte_mempool_obj_iter().
  *
-- 
2.7.4

[PATCH 2/2] mempool: fix comments for no contiguous flag

[Ferruh Yigit]

Fixes: ce94a51ff05c ("mempool: add flag for removing phys contiguous constraint")

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 lib/librte_mempool/rte_mempool.c | 4 +++-
 lib/librte_mempool/rte_mempool.h | 3 +++
 2 files changed, 6 insertions(+), 1 deletion(-)

diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
index e96d14f..d767f39 100644
--- a/lib/librte_mempool/rte_mempool.c
+++ b/lib/librte_mempool/rte_mempool.c
@@ -340,7 +340,9 @@ rte_mempool_free_memchunks(struct rte_mempool *mp)
 }
 
 /* Add objects in the pool, using a physically contiguous memory
- * zone. Return the number of objects added, or a negative value
+ * zone if MEMPOOL_F_NO_PHYS_CONTIG flag is not set.
+ *
+ * Return the number of objects added, or a negative value
  * on error.
  */
 int
diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
index 6fc331a..291c168 100644
--- a/lib/librte_mempool/rte_mempool.h
+++ b/lib/librte_mempool/rte_mempool.h
@@ -798,6 +798,9 @@ rte_mempool_free(struct rte_mempool *mp);
  * Add a virtually and physically contiguous memory chunk in the pool
  * where objects can be instanciated.
  *
+ * mempool flag MEMPOOL_F_NO_PHYS_CONTIG changes behavior of the function
+ * and removes physical contiguous constraint.
+ *
  * @param mp
  *   A pointer to the mempool structure.
  * @param vaddr
-- 
2.7.4

Re: [PATCH 2/2] mempool: fix comments for no contiguous flag

[Olivier Matz]

Hi Ferruh,

On 09/20/2016 06:17 PM, Ferruh Yigit wrote:
> Fixes: ce94a51ff05c ("mempool: add flag for removing phys contiguous constraint")
> 
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---
>  lib/librte_mempool/rte_mempool.c | 4 +++-
>  lib/librte_mempool/rte_mempool.h | 3 +++
>  2 files changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
> index e96d14f..d767f39 100644
> --- a/lib/librte_mempool/rte_mempool.c
> +++ b/lib/librte_mempool/rte_mempool.c
> @@ -340,7 +340,9 @@ rte_mempool_free_memchunks(struct rte_mempool *mp)
>  }
>  
>  /* Add objects in the pool, using a physically contiguous memory
> - * zone. Return the number of objects added, or a negative value
> + * zone if MEMPOOL_F_NO_PHYS_CONTIG flag is not set.
> + *
> + * Return the number of objects added, or a negative value
>   * on error.
>   */
>  int
> diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
> index 6fc331a..291c168 100644
> --- a/lib/librte_mempool/rte_mempool.h
> +++ b/lib/librte_mempool/rte_mempool.h
> @@ -798,6 +798,9 @@ rte_mempool_free(struct rte_mempool *mp);
>   * Add a virtually and physically contiguous memory chunk in the pool
>   * where objects can be instanciated.
>   *
> + * mempool flag MEMPOOL_F_NO_PHYS_CONTIG changes behavior of the function
> + * and removes physical contiguous constraint.
> + *
>   * @param mp
>   *   A pointer to the mempool structure.
>   * @param vaddr
> 

Not sure I'm getting why you add these comments, as I don't see any
usage of MEMPOOL_F_NO_PHYS_CONTIG in rte_mempool_populate_phys().

Could you describe what makes you think the API comments are wrong here?

Thanks,
Olivier

Re: [PATCH 2/2] mempool: fix comments for no contiguous flag

[Ferruh Yigit]

Hi Olivier,

On 9/21/2016 4:12 PM, Olivier Matz wrote:
> Hi Ferruh,
> 
> On 09/20/2016 06:17 PM, Ferruh Yigit wrote:
>> Fixes: ce94a51ff05c ("mempool: add flag for removing phys contiguous constraint")
>>
>> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
>> ---
>>  lib/librte_mempool/rte_mempool.c | 4 +++-
>>  lib/librte_mempool/rte_mempool.h | 3 +++
>>  2 files changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
>> index e96d14f..d767f39 100644
>> --- a/lib/librte_mempool/rte_mempool.c
>> +++ b/lib/librte_mempool/rte_mempool.c
>> @@ -340,7 +340,9 @@ rte_mempool_free_memchunks(struct rte_mempool *mp)
>>  }
>>  
>>  /* Add objects in the pool, using a physically contiguous memory
>> - * zone. Return the number of objects added, or a negative value
>> + * zone if MEMPOOL_F_NO_PHYS_CONTIG flag is not set.
>> + *
>> + * Return the number of objects added, or a negative value
>>   * on error.
>>   */
>>  int
>> diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
>> index 6fc331a..291c168 100644
>> --- a/lib/librte_mempool/rte_mempool.h
>> +++ b/lib/librte_mempool/rte_mempool.h
>> @@ -798,6 +798,9 @@ rte_mempool_free(struct rte_mempool *mp);
>>   * Add a virtually and physically contiguous memory chunk in the pool
>>   * where objects can be instanciated.
>>   *
>> + * mempool flag MEMPOOL_F_NO_PHYS_CONTIG changes behavior of the function
>> + * and removes physical contiguous constraint.
>> + *
>>   * @param mp
>>   *   A pointer to the mempool structure.
>>   * @param vaddr
>>
> 
> Not sure I'm getting why you add these comments, as I don't see any
> usage of MEMPOOL_F_NO_PHYS_CONTIG in rte_mempool_populate_phys().
> 
> Could you describe what makes you think the API comments are wrong here?

rte_mempool_populate_phys() now can get RTE_BAD_PHYS_ADDR as "paddr"
variable, and when paddr == RTE_BAD_PHYS_ADDR, function no more
quarantines that populated items are physically continuous.

Like how this is used in rte_mempool_populate_phys_tab(), if
MEMPOOL_F_NO_PHYS_CONTIG is set rte_mempool_populate_phys() called for
whole virtually continuous memory (len: pg_num * pg_size), so resulting
items may not be in physically continuous addresses, so function comment
becomes wrong.

Although MEMPOOL_F_NO_PHYS_CONTIG is not used within the
rte_mempool_populate_phys(), as far as I can see flag has a side effect
on the functionality.

And please help on wording if it is not accurate.


> 
> Thanks,
> Olivier
> 

Re: [PATCH 2/2] mempool: fix comments for no contiguous flag

[Olivier Matz]

Hi Ferruh,

On 09/21/2016 06:15 PM, Ferruh Yigit wrote:
> Hi Olivier,
> 
> On 9/21/2016 4:12 PM, Olivier Matz wrote:
>> Hi Ferruh,
>>
>> On 09/20/2016 06:17 PM, Ferruh Yigit wrote:
>>> Fixes: ce94a51ff05c ("mempool: add flag for removing phys contiguous constraint")
>>>
>>> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
>>> ---
>>>  lib/librte_mempool/rte_mempool.c | 4 +++-
>>>  lib/librte_mempool/rte_mempool.h | 3 +++
>>>  2 files changed, 6 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
>>> index e96d14f..d767f39 100644
>>> --- a/lib/librte_mempool/rte_mempool.c
>>> +++ b/lib/librte_mempool/rte_mempool.c
>>> @@ -340,7 +340,9 @@ rte_mempool_free_memchunks(struct rte_mempool *mp)
>>>  }
>>>  
>>>  /* Add objects in the pool, using a physically contiguous memory
>>> - * zone. Return the number of objects added, or a negative value
>>> + * zone if MEMPOOL_F_NO_PHYS_CONTIG flag is not set.
>>> + *
>>> + * Return the number of objects added, or a negative value
>>>   * on error.
>>>   */
>>>  int
>>> diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
>>> index 6fc331a..291c168 100644
>>> --- a/lib/librte_mempool/rte_mempool.h
>>> +++ b/lib/librte_mempool/rte_mempool.h
>>> @@ -798,6 +798,9 @@ rte_mempool_free(struct rte_mempool *mp);
>>>   * Add a virtually and physically contiguous memory chunk in the pool
>>>   * where objects can be instanciated.
>>>   *
>>> + * mempool flag MEMPOOL_F_NO_PHYS_CONTIG changes behavior of the function
>>> + * and removes physical contiguous constraint.
>>> + *
>>>   * @param mp
>>>   *   A pointer to the mempool structure.
>>>   * @param vaddr
>>>
>>
>> Not sure I'm getting why you add these comments, as I don't see any
>> usage of MEMPOOL_F_NO_PHYS_CONTIG in rte_mempool_populate_phys().
>>
>> Could you describe what makes you think the API comments are wrong here?
> 
> rte_mempool_populate_phys() now can get RTE_BAD_PHYS_ADDR as "paddr"
> variable, and when paddr == RTE_BAD_PHYS_ADDR, function no more
> quarantines that populated items are physically continuous.
> 
> Like how this is used in rte_mempool_populate_phys_tab(), if
> MEMPOOL_F_NO_PHYS_CONTIG is set rte_mempool_populate_phys() called for
> whole virtually continuous memory (len: pg_num * pg_size), so resulting
> items may not be in physically continuous addresses, so function comment
> becomes wrong.
> 
> Although MEMPOOL_F_NO_PHYS_CONTIG is not used within the
> rte_mempool_populate_phys(), as far as I can see flag has a side effect
> on the functionality.
> 
> And please help on wording if it is not accurate.

OK, so I think we should just talk about paddr in the API comment of
rte_mempool_populate_phys(), not about the flag which is already
documented in mempool_create*().


I suggest the following:

--- a/lib/librte_mempool/rte_mempool.h
+++ b/lib/librte_mempool/rte_mempool.h
@@ -800,6 +800,10 @@ rte_mempool_free(struct rte_mempool *mp);
  * Add a virtually and physically contiguous memory chunk in the pool
  * where objects can be instanciated.
  *
+ * If the given physical address is unknown (paddr = RTE_BAD_PHYS_ADDR),
+ * the chunk doesn't need to be physically contiguous (only virtually),
+ * and allocated objects may span two pages.
+ *
  * @param mp
  *   A pointer to the mempool structure.
  * @param vaddr



What do you think?

Regards
Olivier

Re: [PATCH 1/2] mempool: fix comments for mempool create functions

[Olivier Matz]

Hi Ferruh,

Thanks for checking the API documentation.
Please see few comments below.

On 09/20/2016 06:17 PM, Ferruh Yigit wrote:
> Fixes: 85226f9c526b ("mempool: introduce a function to create an empty pool")
> Fixes: d1d914ebbc25 ("mempool: allocate in several memory chunks by default")
> 
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---
>  lib/librte_mempool/rte_mempool.c | 4 +---
>  lib/librte_mempool/rte_mempool.h | 4 +---
>  2 files changed, 2 insertions(+), 6 deletions(-)
> 
> diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
> index df8b68c..e96d14f 100644
> --- a/lib/librte_mempool/rte_mempool.c
> +++ b/lib/librte_mempool/rte_mempool.c
> @@ -908,9 +908,7 @@ rte_mempool_create(const char *name, unsigned n, unsigned elt_size,
>  /*
>   * Create the mempool over already allocated chunk of memory.
>   * That external memory buffer can consists of physically disjoint pages.
> - * Setting vaddr to NULL, makes mempool to fallback to original behaviour
> - * and allocate space for mempool and it's elements as one big chunk of
> - * physically continuos memory.
> + * Setting vaddr to NULL, makes mempool to fallback to original behaviour.

I would replace "original behaviour" by "rte_mempool_create() behaviour".

>   */
>  struct rte_mempool *
>  rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
> diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
> index 0243f9e..6fc331a 100644
> --- a/lib/librte_mempool/rte_mempool.h
> +++ b/lib/librte_mempool/rte_mempool.h
> @@ -591,8 +591,6 @@ typedef void (rte_mempool_ctor_t)(struct rte_mempool *, void *);
>   *
>   * This function uses ``memzone_reserve()`` to allocate memory. The
>   * pool contains n elements of elt_size. Its size is set to n.
> - * All elements of the mempool are allocated together with the mempool header,
> - * in one physically continuous chunk of memory.

While there, maybe we could replace ``memzone_reserve()`` by
``rte_memzone_reserve()``.

>   *
>   * @param name
>   *   The name of the mempool.
> @@ -748,7 +746,7 @@ rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
>   *
>   * The mempool is allocated and initialized, but it is not populated: no
>   * memory is allocated for the mempool elements. The user has to call
> - * rte_mempool_populate_*() or to add memory chunks to the pool. Once
> + * rte_mempool_populate_*() to add memory chunks to the pool. Once
>   * populated, the user may also want to initialize each object with
>   * rte_mempool_obj_iter().
>   *
> 

Re: [PATCH 1/2] mempool: fix comments for mempool create functions

[Ferruh Yigit]

On 9/21/2016 4:04 PM, Olivier Matz wrote:
> Hi Ferruh,
> 
> Thanks for checking the API documentation.
> Please see few comments below.
> 
> On 09/20/2016 06:17 PM, Ferruh Yigit wrote:
>> Fixes: 85226f9c526b ("mempool: introduce a function to create an empty pool")
>> Fixes: d1d914ebbc25 ("mempool: allocate in several memory chunks by default")
>>
>> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
>> ---
>>  lib/librte_mempool/rte_mempool.c | 4 +---
>>  lib/librte_mempool/rte_mempool.h | 4 +---
>>  2 files changed, 2 insertions(+), 6 deletions(-)
>>
>> diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
>> index df8b68c..e96d14f 100644
>> --- a/lib/librte_mempool/rte_mempool.c
>> +++ b/lib/librte_mempool/rte_mempool.c
>> @@ -908,9 +908,7 @@ rte_mempool_create(const char *name, unsigned n, unsigned elt_size,
>>  /*
>>   * Create the mempool over already allocated chunk of memory.
>>   * That external memory buffer can consists of physically disjoint pages.
>> - * Setting vaddr to NULL, makes mempool to fallback to original behaviour
>> - * and allocate space for mempool and it's elements as one big chunk of
>> - * physically continuos memory.
>> + * Setting vaddr to NULL, makes mempool to fallback to original behaviour.
> 
> I would replace "original behaviour" by "rte_mempool_create() behaviour".
> 
>>   */
>>  struct rte_mempool *
>>  rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
>> diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
>> index 0243f9e..6fc331a 100644
>> --- a/lib/librte_mempool/rte_mempool.h
>> +++ b/lib/librte_mempool/rte_mempool.h
>> @@ -591,8 +591,6 @@ typedef void (rte_mempool_ctor_t)(struct rte_mempool *, void *);
>>   *
>>   * This function uses ``memzone_reserve()`` to allocate memory. The
>>   * pool contains n elements of elt_size. Its size is set to n.
>> - * All elements of the mempool are allocated together with the mempool header,
>> - * in one physically continuous chunk of memory.
> 
> While there, maybe we could replace ``memzone_reserve()`` by
> ``rte_memzone_reserve()``.
> 
>>   *
>>   * @param name
>>   *   The name of the mempool.
>> @@ -748,7 +746,7 @@ rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
>>   *
>>   * The mempool is allocated and initialized, but it is not populated: no
>>   * memory is allocated for the mempool elements. The user has to call
>> - * rte_mempool_populate_*() or to add memory chunks to the pool. Once
>> + * rte_mempool_populate_*() to add memory chunks to the pool. Once
>>   * populated, the user may also want to initialize each object with
>>   * rte_mempool_obj_iter().
>>   *
>>


Thanks for the comments, I will send a new version of the patch with
suggested fixes.

Thanks,
ferruh

[PATCH v2 1/2] mempool: fix comments for mempool create functions

[Ferruh Yigit]

Fixes: 85226f9c526b ("mempool: introduce a function to create an empty pool")
Fixes: d1d914ebbc25 ("mempool: allocate in several memory chunks by default")

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 lib/librte_mempool/rte_mempool.c | 5 ++---
 lib/librte_mempool/rte_mempool.h | 6 ++----
 2 files changed, 4 insertions(+), 7 deletions(-)

diff --git a/lib/librte_mempool/rte_mempool.c b/lib/librte_mempool/rte_mempool.c
index df8b68c..71017e1 100644
--- a/lib/librte_mempool/rte_mempool.c
+++ b/lib/librte_mempool/rte_mempool.c
@@ -908,9 +908,8 @@ rte_mempool_create(const char *name, unsigned n, unsigned elt_size,
 /*
  * Create the mempool over already allocated chunk of memory.
  * That external memory buffer can consists of physically disjoint pages.
- * Setting vaddr to NULL, makes mempool to fallback to original behaviour
- * and allocate space for mempool and it's elements as one big chunk of
- * physically continuos memory.
+ * Setting vaddr to NULL, makes mempool to fallback to rte_mempool_create()
+ * behavior.
  */
 struct rte_mempool *
 rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
index 0243f9e..7a9cc3a 100644
--- a/lib/librte_mempool/rte_mempool.h
+++ b/lib/librte_mempool/rte_mempool.h
@@ -589,10 +589,8 @@ typedef void (rte_mempool_ctor_t)(struct rte_mempool *, void *);
 /**
  * Create a new mempool named *name* in memory.
  *
- * This function uses ``memzone_reserve()`` to allocate memory. The
+ * This function uses ``rte_memzone_reserve()`` to allocate memory. The
  * pool contains n elements of elt_size. Its size is set to n.
- * All elements of the mempool are allocated together with the mempool header,
- * in one physically continuous chunk of memory.
  *
  * @param name
  *   The name of the mempool.
@@ -748,7 +746,7 @@ rte_mempool_xmem_create(const char *name, unsigned n, unsigned elt_size,
  *
  * The mempool is allocated and initialized, but it is not populated: no
  * memory is allocated for the mempool elements. The user has to call
- * rte_mempool_populate_*() or to add memory chunks to the pool. Once
+ * rte_mempool_populate_*() to add memory chunks to the pool. Once
  * populated, the user may also want to initialize each object with
  * rte_mempool_obj_iter().
  *
-- 
2.7.4

[PATCH v2 2/2] mempool: fix comments for no contiguous flag

[Ferruh Yigit]

Fixes: ce94a51ff05c ("mempool: add flag for removing phys contiguous constraint")

Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
---
 lib/librte_mempool/rte_mempool.h | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h
index 7a9cc3a..e7a10e0 100644
--- a/lib/librte_mempool/rte_mempool.h
+++ b/lib/librte_mempool/rte_mempool.h
@@ -798,6 +798,10 @@ rte_mempool_free(struct rte_mempool *mp);
  * Add a virtually and physically contiguous memory chunk in the pool
  * where objects can be instanciated.
  *
+ * If the given physical address is unknown (paddr = RTE_BAD_PHYS_ADDR),
+ * the chunk doesn't need to be physically contiguous (only virtually),
+ * and allocated objects may span two pages.
+ *
  * @param mp
  *   A pointer to the mempool structure.
  * @param vaddr
-- 
2.7.4

Re: [PATCH v2 1/2] mempool: fix comments for mempool create functions

[Olivier Matz]

On 09/28/2016 03:59 PM, Ferruh Yigit wrote:
> Fixes: 85226f9c526b ("mempool: introduce a function to create an empty pool")
> Fixes: d1d914ebbc25 ("mempool: allocate in several memory chunks by default")
> 
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> ---

Series:
Acked-by: Olivier Matz <olivier.matz@6wind.com>

Thanks

Re: [PATCH v2 1/2] mempool: fix comments for mempool create functions

[Thomas Monjalon]

2016-10-03 17:06, Olivier Matz:
> On 09/28/2016 03:59 PM, Ferruh Yigit wrote:
> > Fixes: 85226f9c526b ("mempool: introduce a function to create an empty pool")
> > Fixes: d1d914ebbc25 ("mempool: allocate in several memory chunks by default")
> > 
> > Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> 
> Series:
> Acked-by: Olivier Matz <olivier.matz@6wind.com>

Applied, thanks