[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <2a6f414b-a6cf-dd33-fb70-9dd36e93841e@davidwei.uk>
Date: Sat, 26 Aug 2023 14:37:18 -0700
From: David Wei <dw@...idwei.uk>
To: David Ahern <dsahern@...nel.org>, Jens Axboe <axboe@...nel.dk>,
Pavel Begunkov <asml.silence@...il.com>
Cc: io-uring@...r.kernel.org, netdev@...r.kernel.org,
Mina Almasry <almasrymina@...gle.com>, Jakub Kicinski <kuba@...nel.org>
Subject: Re: [PATCH 03/11] netdev: add XDP_SETUP_ZC_RX command
On 25/08/2023 19:21, David Ahern wrote:
> On 8/25/23 6:19 PM, David Wei wrote:
>> diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
>> index 08fbd4622ccf..a20a5c847916 100644
>> --- a/include/linux/netdevice.h
>> +++ b/include/linux/netdevice.h
>> @@ -1000,6 +1000,7 @@ enum bpf_netdev_command {
>> BPF_OFFLOAD_MAP_ALLOC,
>> BPF_OFFLOAD_MAP_FREE,
>> XDP_SETUP_XSK_POOL,
>> + XDP_SETUP_ZC_RX,
>
> Why XDP in the name? Packets go from nic to driver to stack to io_uring,
> no? That is not XDP.
This new bpf_netdev_command XDP_SETUP_ZC_RX mirrors XDP_SETUP_XSK_POOL above,
both in terms of naming and purpose (the arg structs are almost identical).
However, I dislike this way of toggling ZC RX support anyway and am happy to
change it to a better method e.g. properly supports namespaces.
>
>
>> };
>>
>> struct bpf_prog_offload_ops;
>> @@ -1038,6 +1039,11 @@ struct netdev_bpf {
>> struct xsk_buff_pool *pool;
>> u16 queue_id;
>> } xsk;
>> + /* XDP_SETUP_ZC_RX */
>> + struct {
>> + struct io_zc_rx_ifq *ifq;
>> + u16 queue_id;
>> + } zc_rx;
>> };
>> };
>>
>
Powered by blists - more mailing lists