@@ -95,6 +95,7 @@ static const struct ena_stats ena_stats_rx_strings[] = {
ENA_STAT_RX_ENTRY(xdp_pass),
ENA_STAT_RX_ENTRY(xdp_tx),
ENA_STAT_RX_ENTRY(xdp_invalid),
+ ENA_STAT_RX_ENTRY(xdp_redirect),
};
static const struct ena_stats ena_stats_ena_com_strings[] = {
@@ -289,21 +289,17 @@ static int ena_xdp_xmit_frame(struct net_device *dev,
struct ena_com_tx_ctx ena_tx_ctx = {};
struct ena_tx_buffer *tx_info;
struct ena_ring *xdp_ring;
- struct page *rx_buff_page;
u16 next_to_use, req_id;
int rc;
void *push_hdr;
u32 push_len;
- rx_buff_page = virt_to_page(xdpf->data);
-
xdp_ring = &adapter->tx_ring[qid];
next_to_use = xdp_ring->next_to_use;
req_id = xdp_ring->free_ids[next_to_use];
tx_info = &xdp_ring->tx_buffer_info[req_id];
tx_info->num_of_bufs = 0;
- page_ref_inc(rx_buff_page);
- tx_info->xdp_rx_page = rx_buff_page;
+ tx_info->xdp_rx_page = virt_to_page(xdpf->data);
rc = ena_xdp_tx_map_frame(xdp_ring, tx_info, xdpf, &push_hdr, &push_len);
if (unlikely(rc))
@@ -335,7 +331,7 @@ static int ena_xdp_xmit_frame(struct net_device *dev,
ena_unmap_tx_buff(xdp_ring, tx_info);
tx_info->xdpf = NULL;
error_drop_packet:
- __free_page(tx_info->xdp_rx_page);
+ xdp_return_frame(xdpf);
return NETDEV_TX_OK;
}
@@ -354,25 +350,36 @@ static int ena_xdp_execute(struct ena_ring *rx_ring, struct xdp_buff *xdp)
verdict = bpf_prog_run_xdp(xdp_prog, xdp);
- if (verdict == XDP_TX) {
+ switch (verdict) {
+ case XDP_TX:
xdpf = xdp_convert_buff_to_frame(xdp);
if (unlikely(!xdpf)) {
trace_xdp_exception(rx_ring->netdev, xdp_prog, verdict);
xdp_stat = &rx_ring->rx_stats.xdp_aborted;
- } else {
- ena_xdp_xmit_frame(rx_ring->netdev, xdpf,
- rx_ring->qid + rx_ring->adapter->num_io_queues);
+ break;
+ }
- xdp_stat = &rx_ring->rx_stats.xdp_tx;
+ ena_xdp_xmit_frame(rx_ring->netdev, xdpf,
+ rx_ring->qid + rx_ring->adapter->num_io_queues);
+ xdp_stat = &rx_ring->rx_stats.xdp_tx;
+ break;
+ case XDP_REDIRECT:
+ if (likely(!xdp_do_redirect(rx_ring->netdev, xdp, xdp_prog))) {
+ xdp_stat = &rx_ring->rx_stats.xdp_redirect;
+ break;
}
- } else if (unlikely(verdict == XDP_ABORTED)) {
+ fallthrough;
+ case XDP_ABORTED:
trace_xdp_exception(rx_ring->netdev, xdp_prog, verdict);
xdp_stat = &rx_ring->rx_stats.xdp_aborted;
- } else if (unlikely(verdict == XDP_DROP)) {
+ break;
+ case XDP_DROP:
xdp_stat = &rx_ring->rx_stats.xdp_drop;
- } else if (unlikely(verdict == XDP_PASS)) {
+ break;
+ case XDP_PASS:
xdp_stat = &rx_ring->rx_stats.xdp_pass;
- } else {
+ break;
+ default:
bpf_warn_invalid_xdp_action(verdict);
xdp_stat = &rx_ring->rx_stats.xdp_invalid;
}
@@ -958,11 +965,20 @@ static int ena_alloc_rx_page(struct ena_ring *rx_ring,
return 0;
}
+static void ena_unmap_rx_buff(struct ena_ring *rx_ring,
+ struct ena_rx_buffer *rx_info)
+{
+ struct ena_com_buf *ena_buf = &rx_info->ena_buf;
+
+ dma_unmap_page(rx_ring->dev, ena_buf->paddr - rx_ring->rx_headroom,
+ ENA_PAGE_SIZE,
+ DMA_BIDIRECTIONAL);
+}
+
static void ena_free_rx_page(struct ena_ring *rx_ring,
struct ena_rx_buffer *rx_info)
{
struct page *page = rx_info->page;
- struct ena_com_buf *ena_buf = &rx_info->ena_buf;
if (unlikely(!page)) {
netif_warn(rx_ring->adapter, rx_err, rx_ring->netdev,
@@ -970,9 +986,7 @@ static void ena_free_rx_page(struct ena_ring *rx_ring,
return;
}
- dma_unmap_page(rx_ring->dev, ena_buf->paddr - rx_ring->rx_headroom,
- ENA_PAGE_SIZE,
- DMA_BIDIRECTIONAL);
+ ena_unmap_rx_buff(rx_ring, rx_info);
__free_page(page);
rx_info->page = NULL;
@@ -1396,9 +1410,7 @@ static struct sk_buff *ena_rx_skb(struct ena_ring *rx_ring,
return NULL;
do {
- dma_unmap_page(rx_ring->dev,
- dma_unmap_addr(&rx_info->ena_buf, paddr),
- ENA_PAGE_SIZE, DMA_BIDIRECTIONAL);
+ ena_unmap_rx_buff(rx_ring, rx_info);
skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, rx_info->page,
rx_info->page_offset, len, ENA_PAGE_SIZE);
@@ -1556,6 +1568,7 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi,
struct sk_buff *skb;
int refill_required;
struct xdp_buff xdp;
+ int xdp_flags = 0;
int total_len = 0;
int xdp_verdict;
int rc = 0;
@@ -1603,22 +1616,25 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi,
&next_to_clean);
if (unlikely(!skb)) {
- /* The page might not actually be freed here since the
- * page reference count is incremented in
- * ena_xdp_xmit_frame(), and it will be decreased only
- * when send completion was received from the device
- */
- if (xdp_verdict == XDP_TX)
- ena_free_rx_page(rx_ring,
- &rx_ring->rx_buffer_info[rx_ring->ena_bufs[0].req_id]);
for (i = 0; i < ena_rx_ctx.descs; i++) {
- rx_ring->free_ids[next_to_clean] =
- rx_ring->ena_bufs[i].req_id;
+ int req_id = rx_ring->ena_bufs[i].req_id;
+
+ rx_ring->free_ids[next_to_clean] = req_id;
next_to_clean =
ENA_RX_RING_IDX_NEXT(next_to_clean,
rx_ring->ring_size);
+
+ /* Packets was passed for transmission, unmap it
+ * from RX side.
+ */
+ if (xdp_verdict == XDP_TX || xdp_verdict == XDP_REDIRECT) {
+ ena_unmap_rx_buff(rx_ring,
+ &rx_ring->rx_buffer_info[req_id]);
+ rx_ring->rx_buffer_info[req_id].page = NULL;
+ }
}
if (xdp_verdict != XDP_PASS) {
+ xdp_flags |= xdp_verdict;
res_budget--;
continue;
}
@@ -1664,6 +1680,9 @@ static int ena_clean_rx_irq(struct ena_ring *rx_ring, struct napi_struct *napi,
ena_refill_rx_bufs(rx_ring, refill_required);
}
+ if (xdp_flags & XDP_REDIRECT)
+ xdp_do_flush_map();
+
return work_done;
error:
@@ -239,6 +239,7 @@ struct ena_stats_rx {
u64 xdp_pass;
u64 xdp_tx;
u64 xdp_invalid;
+ u64 xdp_redirect;
};
struct ena_ring {
This patch adds a partial support for the XDP_REDIRECT directive which instructs the driver to pass the packet to an interface specified by the program. The directive is passed to the driver by calling bpf_redirect() or bpf_redirect_map() functions from the eBPF program. To lay the ground for integration with the existing XDP TX implementation the patch removes the redundant page ref count increase in ena_xdp_xmit_frame() and then decrease in ena_clean_rx_irq(). Instead it only DMA unmaps descriptors for which XDP TX or REDIRECT directive was received. The XDP Redirect support is still missing .ndo_xdp_xmit function implementation, which allows to redirect packet to an ENA interface, which would be added in a later patch. Signed-off-by: Shay Agroskin <shayagr@amazon.com> --- drivers/net/ethernet/amazon/ena/ena_ethtool.c | 1 + drivers/net/ethernet/amazon/ena/ena_netdev.c | 83 ++++++++++++------- drivers/net/ethernet/amazon/ena/ena_netdev.h | 1 + 3 files changed, 53 insertions(+), 32 deletions(-)