[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20230421074959.t2ttsy6qpfjgngcr@soft-dev3-1>
Date: Fri, 21 Apr 2023 09:49:59 +0200
From: Horatiu Vultur <horatiu.vultur@...rochip.com>
To: Alexander Lobakin <aleksander.lobakin@...el.com>
CC: <netdev@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
<bpf@...r.kernel.org>, <davem@...emloft.net>,
<edumazet@...gle.com>, <kuba@...nel.org>, <pabeni@...hat.com>,
<daniel@...earbox.net>, <hawk@...nel.org>,
<john.fastabend@...il.com>, <richardcochran@...il.com>,
<UNGLinuxDriver@...rochip.com>, <maciej.fijalkowski@...el.com>,
<alexandr.lobakin@...el.com>
Subject: Re: [PATCH net-next] net: lan966x: Don't use xdp_frame when action
is XDP_TX
The 04/20/2023 16:49, Alexander Lobakin wrote:
>
> From: Horatiu Vultur <horatiu.vultur@...rochip.com>
> Date: Thu, 20 Apr 2023 14:11:52 +0200
Hi Olek,
>
> > When the action of an xdp program was XDP_TX, lan966x was creating
> > a xdp_frame and use this one to send the frame back. But it is also
> > possible to send back the frame without needing a xdp_frame, because
> > it possible to send it back using the page.
> > And then once the frame is transmitted is possible to use directly
> > page_pool_recycle_direct as lan966x is using page pools.
> > This would save some CPU usage on this path.
> >
> > Signed-off-by: Horatiu Vultur <horatiu.vultur@...rochip.com>
>
> [...]
>
> > @@ -702,6 +704,7 @@ static void lan966x_fdma_tx_start(struct lan966x_tx *tx, int next_to_use)
> > int lan966x_fdma_xmit_xdpf(struct lan966x_port *port,
> > struct xdp_frame *xdpf,
> > struct page *page,
> > + u32 len,
> > bool dma_map)
>
> I think you can cut the number of arguments by almost a half:
>
> int lan966x_fdma_xmit_xdpf(struct lan966x_port *port,
> void *ptr, u32 len)
> {
> if (len) {
> /* XDP_TX, ptr is page */
> page = ptr;
>
> dma_sync_here(page, len);
> } else {
> /* XDP_REDIR, ptr is xdp_frame */
> xdpf = ptr;
>
> dma_map_here(xdpf->data, xdpf->len);
> }
>
> @page and @xdpf are mutually exclusive. When @xdpf is non-null, @len is
> excessive (xdpf->len is here), so you can use @len as logical
> `len * !dma_map`, i.e. zero for REDIR and the actual frame length for TX.
Thanks for the review. You are right, I can reduce number of arguments,
the reason why I have done it like this, I thought it is a little bit more
clear this way. But I will update as you propose in the next version
>
> I generally enjoy seeing how you constantly improve stuff in your driver :)
>
> > {
> > struct lan966x *lan966x = port->lan966x;
> > @@ -722,6 +725,15 @@ int lan966x_fdma_xmit_xdpf(struct lan966x_port *port,
> > goto out;
> > }
> [...]
>
> Thanks,
> Olek
--
/Horatiu
Powered by blists - more mailing lists