>
>On 08.12.2017 18:44, Bodireddy, Bhanuprakash wrote:
>>>
>>> On 08.12.2017 16:45, Stokes, Ian wrote:
> All instances of struct dp_netdev_pmd_thread are allocated by
> xzalloc and therefore doesn't guarantee memory allocation aligned
> on CACHE_LINE_SIZE boundary. Due to this any paddi
On 08.12.2017 18:44, Bodireddy, Bhanuprakash wrote:
>>
>> On 08.12.2017 16:45, Stokes, Ian wrote:
All instances of struct dp_netdev_pmd_thread are allocated by xzalloc
and therefore doesn't guarantee memory allocation aligned on
CACHE_LINE_SIZE boundary. Due to this any padding done
>
>On 08.12.2017 16:45, Stokes, Ian wrote:
>>> All instances of struct dp_netdev_pmd_thread are allocated by xzalloc
>>> and therefore doesn't guarantee memory allocation aligned on
>>> CACHE_LINE_SIZE boundary. Due to this any padding done inside the
>>> structure with this assumption might create
On 08.12.2017 16:45, Stokes, Ian wrote:
>> All instances of struct dp_netdev_pmd_thread are allocated by xzalloc and
>> therefore doesn't guarantee memory allocation aligned on CACHE_LINE_SIZE
>> boundary. Due to this any padding done inside the structure with this
>> assumption might create holes.
> All instances of struct dp_netdev_pmd_thread are allocated by xzalloc and
> therefore doesn't guarantee memory allocation aligned on CACHE_LINE_SIZE
> boundary. Due to this any padding done inside the structure with this
> assumption might create holes.
>
> This commit replaces xzalloc, free wit
All instances of struct dp_netdev_pmd_thread are allocated by xzalloc
and therefore doesn't guarantee memory allocation aligned on
CACHE_LINE_SIZE boundary. Due to this any padding done inside
the structure with this assumption might create holes.
This commit replaces xzalloc, free with xzalloc_ca