RE: [PATCH v5] net/iavf: data paths support no-polling mode

["Zhang, Qi Z" <qi.z.zhang@intel.com>]

> -----Original Message-----
> From: Mingjin Ye <mingjinx.ye@intel.com>
> Sent: Thursday, October 19, 2023 5:04 PM
> To: dev@dpdk.org
> Cc: Yang, Qiming <qiming.yang@intel.com>; Zhou, YidingX
> <yidingx.zhou@intel.com>; Ye, MingjinX <mingjinx.ye@intel.com>; Su, Simei
> <simei.su@intel.com>; Wu, Wenjun1 <wenjun1.wu@intel.com>; Zhang,
> Yuying <yuying.zhang@intel.com>; Xing, Beilei <beilei.xing@intel.com>; Wu,
> Jingjing <jingjing.wu@intel.com>
> Subject: [PATCH v5] net/iavf: data paths support no-polling mode
> 
> In a scenario involving a hot firmware upgrade, the network device on the
> host side need to be reset, potentially causing the hardware queues to
> become unreachable. In a VM, continuing to run VF PMD Rx/Tx during this
> process can lead to application crash.
> 
> The solution is to implement a 'no-polling' Rx and Tx wrapper. This wrapper
> will check the link status and return immediately if the link is down. This is
> especially important because the link down events will continue to be sent
> from the PF to the VF during firmware hot upgrades, and the event will
> always occur before the RESET IMPENDING event.
> 
> The no-polling rx/tx mechanism will only be active when the devarg "no-poll-
> on-link-down" is enabled. This devarg is typically recommended for use in
> this specific hot upgrade scenario.
> Ideally, "no-poll-on-link-down" should be used in conjunction with the
> devarg "auto-reset" to provide a seamless and user-friendly experience
> within the VM.
> 
> Signed-off-by: Mingjin Ye <mingjinx.ye@intel.com>

Acked-by: Qi Zhang <qi.z.zhang@intel.com>

Applied to dpdk-next-net-intel after minor refine on commit log and description in ice.rst.

Thanks
Qi