Re: [PATCH net-next v4 0/8] bpf: cpumap: enable GRO for XDP_PASS frames

[Alexander Lobakin <aleksander.lobakin@intel.com>]

From: Eric Dumazet <edumazet@google.com>
Date: Mon, 10 Feb 2025 16:31:22 +0100

> On Mon, Feb 10, 2025 at 3:10 PM Alexander Lobakin
> <aleksander.lobakin@intel.com> wrote:
>>
>> From: Alexander Lobakin <aleksander.lobakin@intel.com>
>> Date: Wed, 5 Feb 2025 17:36:01 +0100
>>
>>> Several months ago, I had been looking through my old XDP hints tree[0]
>>> to check whether some patches not directly related to hints can be sent
>>> standalone. Roughly at the same time, Daniel appeared and asked[1] about
>>> GRO for cpumap from that tree.
>>
>> I see "Changes requested" on Patchwork. Which ones?
>>
>> 1/8 regarding gro_node? Nobody proposed a solution which would be as
>> efficient, but avoid using struct_group(), I don't see such as well.
>> I explain in the commitmsgs and cover letter everything. Jakub gave me
>> Acked-by on struct_group() in the v3 thread.
> 
> One of the points of your nice series is to dissociate GRO from NAPI,
> so defining gro_node inside napi_struct is not appealing.
> 
> I suggested not putting napi_id in the new structure.
> 
> If you need to cache a copy in it for "performance/whatever reason",
> you can cache napi_id, because napi->napi_id is only set once
> in __napi_hash_add_with_id()
> 
> gro->napi_id_cache = napi->napi_id;

This was rejected by Kuba in v2.
He didn't like to have napi_id two times within napi_struct (one inside
gro_node, one outside).

Thanks,
Olek