On Fri, Jul 05, 2019 at 01:13:54PM +0200, Jesper Dangaard Brouer wrote:
On Fri, 5 Jul 2019 02:14:06 +0300
Ivan Khoronzhuk <ivan.khoronzhuk@xxxxxxxxxx> wrote:
+static int cpsw_xdp_tx_frame(struct cpsw_priv *priv, struct xdp_frame *xdpf,
+ struct page *page)
+{
+ struct cpsw_common *cpsw = priv->cpsw;
+ struct cpsw_meta_xdp *xmeta;
+ struct cpdma_chan *txch;
+ dma_addr_t dma;
+ int ret, port;
+
+ xmeta = (void *)xdpf + CPSW_XMETA_OFFSET;
+ xmeta->ndev = priv->ndev;
+ xmeta->ch = 0;
+ txch = cpsw->txv[0].ch;
+
+ port = priv->emac_port + cpsw->data.dual_emac;
+ if (page) {
+ dma = page_pool_get_dma_addr(page);
+ dma += xdpf->data - (void *)xdpf;
This code is only okay because this only happens for XDP_TX, where you
know this head-room calculation will be true. The "correct"
calculation of the head-room would be:
dma += xdpf->headroom + sizeof(struct xdp_frame);
The reason behind not using xdpf pointer itself as "data_hard_start",
is to allow struct xdp_frame to be located in another memory area.
My assumption was based on:
struct xdp_frame *convert_to_xdp_frame(struct xdp_buff *xdp)
{
...
xdp_frame = xdp->data_hard_start;
...
xdp_frame->headroom = headroom - sizeof(*xdp_frame);
...
}
But agree, it doesn't contradict the reason in question.
So, better use proposed variant. Will check and do this in v8 a little later:
dma += xdpf->headroom + sizeof(struct xdp_frame);
This will be useful for e.g. AF_XDP transmit, or other zero-copy
transmit to go through ndo_xdp_xmit() (as we don't want userspace to
be-able to e.g. "race" change xdpf->len during transmit/DMA-completion).
+ ret = cpdma_chan_submit_mapped(txch, cpsw_xdpf_to_handle(xdpf),
+ dma, xdpf->len, port);
+ } else {
+ if (sizeof(*xmeta) > xdpf->headroom) {
+ xdp_return_frame_rx_napi(xdpf);
+ return -EINVAL;
+ }
+
+ ret = cpdma_chan_submit(txch, cpsw_xdpf_to_handle(xdpf),
+ xdpf->data, xdpf->len, port);
+ }
--
Best regards,
Jesper Dangaard Brouer
MSc.CS, Principal Kernel Engineer at Red Hat
LinkedIn: http://www.linkedin.com/in/brouer
--
Regards,
Ivan Khoronzhuk
