[dpdk-dev] [PATCH v3 11/68] eal: enable reserving physically contiguous memzones
Burakov, Anatoly
anatoly.burakov at intel.com
Wed Apr 4 10:01:22 CEST 2018
On 04-Apr-18 12:41 AM, Stephen Hemminger wrote:
> On Wed, 4 Apr 2018 00:21:23 +0100
> Anatoly Burakov <anatoly.burakov at intel.com> wrote:
>
>> This adds a new set of _contig API's to rte_memzone. For now,
>> hugepage memory is always contiguous, but we need to prepare the
>> drivers for the switch.
>>
>> Signed-off-by: Anatoly Burakov <anatoly.burakov at intel.com>
>
> Why not make fragmentation an optional flag to the memzone_reserved
> rather than a new API. That way less drivers need to be changed.
>
> #define RTE_MEMZONE_SPARSE 0x00100000 /* Allow zone to be non-contiguous */
>
That is a good idea, however, i would perhaps do it the other way
around. Most of the time we are reserving memzones, we do not need those
to be IOVA-contiguous (i.e. creating rings, hashtables etc.). So it
would rather be a RTE_MEMZONE_CONTIG (or some such) flag.
--
Thanks,
Anatoly
More information about the dev
mailing list