[dpdk-dev] [PATCH v5 2/2] net/mlx5: fix instruction hotspot on replenishing Rx buffer

[Yongseok Koh]

Oops, I missed this email somehow,
probably out of mind during my long vacation. :-)

I've also encountered this issue with 17.11.6.
http://git.dpdk.org/dpdk-stable/commit/?h=17.11&id=63f06f3fccc87c55adb33248e5c68a0175d213f1

I'll send a backport to you.


Sorry for late reply.
Yongseok

> On Feb 21, 2019, at 11:10 AM, Kevin Traynor <ktraynor at redhat.com> wrote:
> 
> Hi Yongseok,
> 
> Can you let me know how you want to proceed with the below. I think we
> could just drop as it's a performance optimization, or maybe you have a
> different idea.
> 
> thanks,
> Kevin.
> 
> On 06/02/2019 15:54, Kevin Traynor wrote:
>> On 01/14/2019 09:16 PM, Yongseok Koh wrote:
>>> On replenishing Rx buffers for vectorized Rx, mbuf->buf_addr isn't needed
>>> to be accessed as it is static and easily calculated from the mbuf address.
>>> Accessing the mbuf content causes unnecessary load stall and it is worsened
>>> on ARM.
>>> 
>>> Fixes: 545b884b1da3 ("net/mlx5: fix buffer address posting in SSE Rx")
>>> Cc: stable at dpdk.org
>>> 
>> 
>> This is using the API introduced in 1/2, so it's not really suitable for
>> backport. Maybe you want to send an alternative for stable?
>> 
>>> Signed-off-by: Yongseok Koh <yskoh at mellanox.com>
>>> Acked-by: Shahaf Shuler <shahafs at mellanox.com>
>>> ---
>