Hi Stephen,
I appreciate your careful review and detailed suggestions.
Thank you.
------------------------------------------------------------------
发件人：Stephen Hemminger <[email protected]>
发送时间：2025年11月10日(周一) 04:16
收件人：Dimon<[email protected]>
抄 送：dev<[email protected]>; Thomas Monjalon<[email protected]>;
Alvin<[email protected]>; Leon<[email protected]>;
Sam<[email protected]>
主 题：Re: [PATCH v1 4/4] net/nbl: update documentation and maintainers
On Thu, 6 Nov 2025 23:34:59 -0800
Dimon Zhao <[email protected]> wrote:
> diff --git a/doc/guides/nics/nbl.rst b/doc/guides/nics/nbl.rst
> index f41da20eea..9657ee2c87 100644
> --- a/doc/guides/nics/nbl.rst
> +++ b/doc/guides/nics/nbl.rst
> @@ -70,6 +70,13 @@ Anolis community:
> Alternatively, you can contact us to obtain the `nbl_core` code and
> installation package.
>
>
> +Coexistence Between DPDK And Kernel Driver
> +-------------
> +When the device is bind to the nbl_core driver, and IOMMU in passthrough
> mode,
> +it is necessary to force IOVA to PA via EAL --iova-mode=pa option,
> +DPDK works normally only under this configuration.
> +
This paragraph is has awkward wording because of verb tense.
Also in RST files the header needs to have as many dash characters as the
header above
and use highlighting with `` for options.
Better as:
When this device is bound to the nbl_core driver and IOMMU is in passthrough
mode,
it is necessary to force I/O virtual address (IOVA) to be mapped to physical
address (PA)
with the EAL command line option ``--iova-mode=pa``.
Probably should also check for this in the driver, and warn user if not set.
