* [xdp-hints] [PATCH net-next v5 0/3] XDP Rx HWTS metadata for stmmac driver
@ 2023-04-14 5:26 Song Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff Song Yoong Siang
` (2 more replies)
0 siblings, 3 replies; 7+ messages in thread
From: Song Yoong Siang @ 2023-04-14 5:26 UTC (permalink / raw)
To: Giuseppe Cavallaro, Alexandre Torgue, Jose Abreu,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Maxime Coquelin, Alexei Starovoitov, Daniel Borkmann,
Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev,
Alexander Duyck, Jesper Dangaard Brouer, Ong Boon Leong,
Jacob Keller
Cc: netdev, linux-stm32, linux-arm-kernel, linux-kernel, bpf,
xdp-hints, Song Yoong Siang
Implemented XDP receive hardware timestamp metadata for stmmac driver.
This patchset is tested with tools/testing/selftests/bpf/xdp_hw_metadata.
Below are the test steps and results.
Command on DUT:
sudo ./xdp_hw_metadata <interface name>
Command on Link Partner:
echo -n xdp | nc -u -q1 <destination IPv4 addr> 9091
echo -n skb | nc -u -q1 <destination IPv4 addr> 9092
Result for port 9091:
0x55fdb5f006d0: rx_desc[3]->addr=1000000003bd000 addr=3bd100 comp_addr=3bd000
rx_timestamp: 1677762474360150047
rx_hash: 0
0x55fdb5f006d0: complete idx=515 addr=3bd000
Result for port 9092:
found skb hwtstamp = 1677762476.320146161
changelog:
v4 -> v5: remove zeroing operation on ctx variable
v3 -> v4: directly retrieve Rx HWTS in stmmac_xdp_rx_timestamp(), instead
of reuse stmmac_get_rx_hwtstamp()
v2 -> v3: To reduce packet processing cost, get the Rx HWTS only when
xmo_rx_timestamp() is called
v1 -> v2: Add static to stmmac_xdp_metadata_ops declaration
---
Song Yoong Siang (3):
net: stmmac: introduce wrapper for struct xdp_buff
net: stmmac: add Rx HWTS metadata to XDP receive pkt
net: stmmac: add Rx HWTS metadata to XDP ZC receive pkt
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 7 ++
.../net/ethernet/stmicro/stmmac/stmmac_main.c | 80 ++++++++++++++++---
2 files changed, 77 insertions(+), 10 deletions(-)
--
2.34.1
^ permalink raw reply [flat|nested] 7+ messages in thread
* [xdp-hints] [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff
2023-04-14 5:26 [xdp-hints] [PATCH net-next v5 0/3] XDP Rx HWTS metadata for stmmac driver Song Yoong Siang
@ 2023-04-14 5:26 ` Song Yoong Siang
2023-04-14 16:04 ` [xdp-hints] " Jesper Dangaard Brouer
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt Song Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 3/3] net: stmmac: add Rx HWTS metadata to XDP ZC " Song Yoong Siang
2 siblings, 1 reply; 7+ messages in thread
From: Song Yoong Siang @ 2023-04-14 5:26 UTC (permalink / raw)
To: Giuseppe Cavallaro, Alexandre Torgue, Jose Abreu,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Maxime Coquelin, Alexei Starovoitov, Daniel Borkmann,
Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev,
Alexander Duyck, Jesper Dangaard Brouer, Ong Boon Leong,
Jacob Keller
Cc: netdev, linux-stm32, linux-arm-kernel, linux-kernel, bpf,
xdp-hints, Song Yoong Siang
Introduce struct stmmac_xdp_buff as a preparation to support XDP Rx
metadata via kfuncs.
Signed-off-by: Song Yoong Siang <yoong.siang.song@intel.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
---
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 4 ++++
.../net/ethernet/stmicro/stmmac/stmmac_main.c | 18 +++++++++---------
2 files changed, 13 insertions(+), 9 deletions(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index 3d15e1e92e18..ac8ccf851708 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -92,6 +92,10 @@ struct stmmac_rx_buffer {
dma_addr_t sec_addr;
};
+struct stmmac_xdp_buff {
+ struct xdp_buff xdp;
+};
+
struct stmmac_rx_queue {
u32 rx_count_frames;
u32 queue_index;
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index d7fcab057032..10b9f8912bb2 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -5190,7 +5190,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
enum dma_data_direction dma_dir;
unsigned int desc_size;
struct sk_buff *skb = NULL;
- struct xdp_buff xdp;
+ struct stmmac_xdp_buff ctx;
int xdp_status = 0;
int buf_sz;
@@ -5311,17 +5311,17 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
dma_sync_single_for_cpu(priv->device, buf->addr,
buf1_len, dma_dir);
- xdp_init_buff(&xdp, buf_sz, &rx_q->xdp_rxq);
- xdp_prepare_buff(&xdp, page_address(buf->page),
+ xdp_init_buff(&ctx.xdp, buf_sz, &rx_q->xdp_rxq);
+ xdp_prepare_buff(&ctx.xdp, page_address(buf->page),
buf->page_offset, buf1_len, false);
- pre_len = xdp.data_end - xdp.data_hard_start -
+ pre_len = ctx.xdp.data_end - ctx.xdp.data_hard_start -
buf->page_offset;
- skb = stmmac_xdp_run_prog(priv, &xdp);
+ skb = stmmac_xdp_run_prog(priv, &ctx.xdp);
/* Due xdp_adjust_tail: DMA sync for_device
* cover max len CPU touch
*/
- sync_len = xdp.data_end - xdp.data_hard_start -
+ sync_len = ctx.xdp.data_end - ctx.xdp.data_hard_start -
buf->page_offset;
sync_len = max(sync_len, pre_len);
@@ -5331,7 +5331,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
if (xdp_res & STMMAC_XDP_CONSUMED) {
page_pool_put_page(rx_q->page_pool,
- virt_to_head_page(xdp.data),
+ virt_to_head_page(ctx.xdp.data),
sync_len, true);
buf->page = NULL;
priv->dev->stats.rx_dropped++;
@@ -5359,7 +5359,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
if (!skb) {
/* XDP program may expand or reduce tail */
- buf1_len = xdp.data_end - xdp.data;
+ buf1_len = ctx.xdp.data_end - ctx.xdp.data;
skb = napi_alloc_skb(&ch->rx_napi, buf1_len);
if (!skb) {
@@ -5369,7 +5369,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
}
/* XDP program may adjust header */
- skb_copy_to_linear_data(skb, xdp.data, buf1_len);
+ skb_copy_to_linear_data(skb, ctx.xdp.data, buf1_len);
skb_put(skb, buf1_len);
/* Data payload copied into SKB, page ready for recycle */
--
2.34.1
^ permalink raw reply [flat|nested] 7+ messages in thread
* [xdp-hints] [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt
2023-04-14 5:26 [xdp-hints] [PATCH net-next v5 0/3] XDP Rx HWTS metadata for stmmac driver Song Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff Song Yoong Siang
@ 2023-04-14 5:26 ` Song Yoong Siang
2023-04-14 16:38 ` [xdp-hints] " Jesper Dangaard Brouer
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 3/3] net: stmmac: add Rx HWTS metadata to XDP ZC " Song Yoong Siang
2 siblings, 1 reply; 7+ messages in thread
From: Song Yoong Siang @ 2023-04-14 5:26 UTC (permalink / raw)
To: Giuseppe Cavallaro, Alexandre Torgue, Jose Abreu,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Maxime Coquelin, Alexei Starovoitov, Daniel Borkmann,
Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev,
Alexander Duyck, Jesper Dangaard Brouer, Ong Boon Leong,
Jacob Keller
Cc: netdev, linux-stm32, linux-arm-kernel, linux-kernel, bpf,
xdp-hints, Song Yoong Siang
Add receive hardware timestamp metadata support via kfunc to XDP receive
packets.
Suggested-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Song Yoong Siang <yoong.siang.song@intel.com>
Acked-by: Stanislav Fomichev <sdf@google.com>
---
drivers/net/ethernet/stmicro/stmmac/stmmac.h | 3 ++
.../net/ethernet/stmicro/stmmac/stmmac_main.c | 40 ++++++++++++++++++-
2 files changed, 42 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
index ac8ccf851708..826ac0ec88c6 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
@@ -94,6 +94,9 @@ struct stmmac_rx_buffer {
struct stmmac_xdp_buff {
struct xdp_buff xdp;
+ struct stmmac_priv *priv;
+ struct dma_desc *p;
+ struct dma_desc *np;
};
struct stmmac_rx_queue {
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 10b9f8912bb2..74f78e5537a3 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -5313,10 +5313,15 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
xdp_init_buff(&ctx.xdp, buf_sz, &rx_q->xdp_rxq);
xdp_prepare_buff(&ctx.xdp, page_address(buf->page),
- buf->page_offset, buf1_len, false);
+ buf->page_offset, buf1_len, true);
pre_len = ctx.xdp.data_end - ctx.xdp.data_hard_start -
buf->page_offset;
+
+ ctx.priv = priv;
+ ctx.p = p;
+ ctx.np = np;
+
skb = stmmac_xdp_run_prog(priv, &ctx.xdp);
/* Due xdp_adjust_tail: DMA sync for_device
* cover max len CPU touch
@@ -7060,6 +7065,37 @@ void stmmac_fpe_handshake(struct stmmac_priv *priv, bool enable)
}
}
+static int stmmac_xdp_rx_timestamp(const struct xdp_md *_ctx, u64 *timestamp)
+{
+ const struct stmmac_xdp_buff *ctx = (void *)_ctx;
+ struct stmmac_priv *priv = ctx->priv;
+ struct dma_desc *desc = ctx->p;
+ struct dma_desc *np = ctx->np;
+ struct dma_desc *p = ctx->p;
+ u64 ns = 0;
+
+ if (!priv->hwts_rx_en)
+ return -ENODATA;
+
+ /* For GMAC4, the valid timestamp is from CTX next desc. */
+ if (priv->plat->has_gmac4 || priv->plat->has_xgmac)
+ desc = np;
+
+ /* Check if timestamp is available */
+ if (stmmac_get_rx_timestamp_status(priv, p, np, priv->adv_ts)) {
+ stmmac_get_timestamp(priv, desc, priv->adv_ts, &ns);
+ ns -= priv->plat->cdc_error_adj;
+ *timestamp = ns_to_ktime(ns);
+ return 0;
+ }
+
+ return -ENODATA;
+}
+
+static const struct xdp_metadata_ops stmmac_xdp_metadata_ops = {
+ .xmo_rx_timestamp = stmmac_xdp_rx_timestamp,
+};
+
/**
* stmmac_dvr_probe
* @device: device pointer
@@ -7167,6 +7203,8 @@ int stmmac_dvr_probe(struct device *device,
ndev->netdev_ops = &stmmac_netdev_ops;
+ ndev->xdp_metadata_ops = &stmmac_xdp_metadata_ops;
+
ndev->hw_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
NETIF_F_RXCSUM;
ndev->xdp_features = NETDEV_XDP_ACT_BASIC | NETDEV_XDP_ACT_REDIRECT |
--
2.34.1
^ permalink raw reply [flat|nested] 7+ messages in thread
* [xdp-hints] [PATCH net-next v5 3/3] net: stmmac: add Rx HWTS metadata to XDP ZC receive pkt
2023-04-14 5:26 [xdp-hints] [PATCH net-next v5 0/3] XDP Rx HWTS metadata for stmmac driver Song Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff Song Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt Song Yoong Siang
@ 2023-04-14 5:26 ` Song Yoong Siang
2 siblings, 0 replies; 7+ messages in thread
From: Song Yoong Siang @ 2023-04-14 5:26 UTC (permalink / raw)
To: Giuseppe Cavallaro, Alexandre Torgue, Jose Abreu,
David S . Miller, Eric Dumazet, Jakub Kicinski, Paolo Abeni,
Maxime Coquelin, Alexei Starovoitov, Daniel Borkmann,
Jesper Dangaard Brouer, John Fastabend, Stanislav Fomichev,
Alexander Duyck, Jesper Dangaard Brouer, Ong Boon Leong,
Jacob Keller
Cc: netdev, linux-stm32, linux-arm-kernel, linux-kernel, bpf,
xdp-hints, Song Yoong Siang
Add receive hardware timestamp metadata support via kfunc to XDP Zero Copy
receive packets.
Signed-off-by: Song Yoong Siang <yoong.siang.song@intel.com>
---
.../net/ethernet/stmicro/stmmac/stmmac_main.c | 22 +++++++++++++++++++
1 file changed, 22 insertions(+)
diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
index 74f78e5537a3..f3b8eae0846e 100644
--- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
+++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
@@ -1614,6 +1614,12 @@ static int stmmac_alloc_rx_buffers_zc(struct stmmac_priv *priv,
struct stmmac_rx_queue *rx_q = &dma_conf->rx_queue[queue];
int i;
+ /* struct stmmac_xdp_buff is using cb field (maximum size of 24 bytes)
+ * in struct xdp_buff_xsk to stash driver specific information. Thus,
+ * use this macro to make sure no size violations.
+ */
+ XSK_CHECK_PRIV_TYPE(struct stmmac_xdp_buff);
+
for (i = 0; i < dma_conf->dma_rx_size; i++) {
struct stmmac_rx_buffer *buf;
dma_addr_t dma_addr;
@@ -4998,6 +5004,16 @@ static bool stmmac_rx_refill_zc(struct stmmac_priv *priv, u32 queue, u32 budget)
return ret;
}
+static struct stmmac_xdp_buff *xsk_buff_to_stmmac_ctx(struct xdp_buff *xdp)
+{
+ /* In XDP zero copy data path, xdp field in struct xdp_buff_xsk is used
+ * to represent incoming packet, whereas cb field in the same structure
+ * is used to store driver specific info. Thus, struct stmmac_xdp_buff
+ * is laid on top of xdp and cb fields of struct xdp_buff_xsk.
+ */
+ return (struct stmmac_xdp_buff *)xdp;
+}
+
static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
{
struct stmmac_rx_queue *rx_q = &priv->dma_conf.rx_queue[queue];
@@ -5027,6 +5043,7 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
}
while (count < limit) {
struct stmmac_rx_buffer *buf;
+ struct stmmac_xdp_buff *ctx;
unsigned int buf1_len = 0;
struct dma_desc *np, *p;
int entry;
@@ -5112,6 +5129,11 @@ static int stmmac_rx_zc(struct stmmac_priv *priv, int limit, u32 queue)
goto read_again;
}
+ ctx = xsk_buff_to_stmmac_ctx(buf->xdp);
+ ctx->priv = priv;
+ ctx->p = p;
+ ctx->np = np;
+
/* XDP ZC Frame only support primary buffers for now */
buf1_len = stmmac_rx_buf1_len(priv, p, status, len);
len += buf1_len;
--
2.34.1
^ permalink raw reply [flat|nested] 7+ messages in thread
* [xdp-hints] Re: [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff Song Yoong Siang
@ 2023-04-14 16:04 ` Jesper Dangaard Brouer
0 siblings, 0 replies; 7+ messages in thread
From: Jesper Dangaard Brouer @ 2023-04-14 16:04 UTC (permalink / raw)
To: Song Yoong Siang, Giuseppe Cavallaro, Alexandre Torgue,
Jose Abreu, David S . Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Maxime Coquelin, Alexei Starovoitov,
Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
Stanislav Fomichev, Alexander Duyck, Ong Boon Leong,
Jacob Keller
Cc: brouer, netdev, linux-stm32, linux-arm-kernel, linux-kernel, bpf,
xdp-hints
On 14/04/2023 07.26, Song Yoong Siang wrote:
> Introduce struct stmmac_xdp_buff as a preparation to support XDP Rx
> metadata via kfuncs.
>
> Signed-off-by: Song Yoong Siang<yoong.siang.song@intel.com>
> Reviewed-by: Jacob Keller<jacob.e.keller@intel.com>
> ---
> drivers/net/ethernet/stmicro/stmmac/stmmac.h | 4 ++++
> .../net/ethernet/stmicro/stmmac/stmmac_main.c | 18 +++++++++---------
> 2 files changed, 13 insertions(+), 9 deletions(-)
Acked-by: Jesper Dangaard Brouer <brouer@redhat.com>
^ permalink raw reply [flat|nested] 7+ messages in thread
* [xdp-hints] Re: [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt Song Yoong Siang
@ 2023-04-14 16:38 ` Jesper Dangaard Brouer
2023-04-15 5:16 ` Song, Yoong Siang
0 siblings, 1 reply; 7+ messages in thread
From: Jesper Dangaard Brouer @ 2023-04-14 16:38 UTC (permalink / raw)
To: Song Yoong Siang, Giuseppe Cavallaro, Alexandre Torgue,
Jose Abreu, David S . Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Maxime Coquelin, Alexei Starovoitov,
Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
Stanislav Fomichev, Alexander Duyck, Ong Boon Leong,
Jacob Keller
Cc: brouer, netdev, linux-stm32, linux-arm-kernel, linux-kernel, bpf,
xdp-hints
On 14/04/2023 07.26, Song Yoong Siang wrote:
> Add receive hardware timestamp metadata support via kfunc to XDP receive
> packets.
>
> Suggested-by: Stanislav Fomichev <sdf@google.com>
> Signed-off-by: Song Yoong Siang <yoong.siang.song@intel.com>
> Acked-by: Stanislav Fomichev <sdf@google.com>
> ---
> drivers/net/ethernet/stmicro/stmmac/stmmac.h | 3 ++
> .../net/ethernet/stmicro/stmmac/stmmac_main.c | 40 ++++++++++++++++++-
> 2 files changed, 42 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
> index ac8ccf851708..826ac0ec88c6 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
> @@ -94,6 +94,9 @@ struct stmmac_rx_buffer {
>
> struct stmmac_xdp_buff {
> struct xdp_buff xdp;
> + struct stmmac_priv *priv;
> + struct dma_desc *p;
> + struct dma_desc *np;
Hmm, I don't like the naming of the descriptors as "p" and "np".
If you insist on this naming, at least we need comments describing that
this is.
Does "desc" and "ndesc" make sense? (where "n" means "next")
> };
>
> struct stmmac_rx_queue {
> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> index 10b9f8912bb2..74f78e5537a3 100644
> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> @@ -5313,10 +5313,15 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue)
>
> xdp_init_buff(&ctx.xdp, buf_sz, &rx_q->xdp_rxq);
> xdp_prepare_buff(&ctx.xdp, page_address(buf->page),
> - buf->page_offset, buf1_len, false);
> + buf->page_offset, buf1_len, true);
>
> pre_len = ctx.xdp.data_end - ctx.xdp.data_hard_start -
> buf->page_offset;
> +
> + ctx.priv = priv;
> + ctx.p = p;
> + ctx.np = np;
> +
> skb = stmmac_xdp_run_prog(priv, &ctx.xdp);
> /* Due xdp_adjust_tail: DMA sync for_device
> * cover max len CPU touch
> @@ -7060,6 +7065,37 @@ void stmmac_fpe_handshake(struct stmmac_priv *priv, bool enable)
> }
> }
>
> +static int stmmac_xdp_rx_timestamp(const struct xdp_md *_ctx, u64 *timestamp)
> +{
> + const struct stmmac_xdp_buff *ctx = (void *)_ctx;
> + struct stmmac_priv *priv = ctx->priv;
> + struct dma_desc *desc = ctx->p;
> + struct dma_desc *np = ctx->np;
> + struct dma_desc *p = ctx->p;
> + u64 ns = 0;
> +
> + if (!priv->hwts_rx_en)
> + return -ENODATA;
> +
> + /* For GMAC4, the valid timestamp is from CTX next desc. */
> + if (priv->plat->has_gmac4 || priv->plat->has_xgmac)
> + desc = np;
> +
> + /* Check if timestamp is available */
> + if (stmmac_get_rx_timestamp_status(priv, p, np, priv->adv_ts)) {
> + stmmac_get_timestamp(priv, desc, priv->adv_ts, &ns);
> + ns -= priv->plat->cdc_error_adj;
> + *timestamp = ns_to_ktime(ns);
> + return 0;
> + }
> +
> + return -ENODATA;
> +}
> +
> +static const struct xdp_metadata_ops stmmac_xdp_metadata_ops = {
> + .xmo_rx_timestamp = stmmac_xdp_rx_timestamp,
> +};
> +
> /**
> * stmmac_dvr_probe
> * @device: device pointer
> @@ -7167,6 +7203,8 @@ int stmmac_dvr_probe(struct device *device,
>
> ndev->netdev_ops = &stmmac_netdev_ops;
>
> + ndev->xdp_metadata_ops = &stmmac_xdp_metadata_ops;
> +
> ndev->hw_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
> NETIF_F_RXCSUM;
> ndev->xdp_features = NETDEV_XDP_ACT_BASIC | NETDEV_XDP_ACT_REDIRECT |
^ permalink raw reply [flat|nested] 7+ messages in thread
* [xdp-hints] Re: [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt
2023-04-14 16:38 ` [xdp-hints] " Jesper Dangaard Brouer
@ 2023-04-15 5:16 ` Song, Yoong Siang
0 siblings, 0 replies; 7+ messages in thread
From: Song, Yoong Siang @ 2023-04-15 5:16 UTC (permalink / raw)
To: Jesper Dangaard Brouer, Giuseppe Cavallaro, Alexandre Torgue,
Jose Abreu, David S . Miller, Eric Dumazet, Jakub Kicinski,
Paolo Abeni, Maxime Coquelin, Alexei Starovoitov,
Daniel Borkmann, Jesper Dangaard Brouer, John Fastabend,
Stanislav Fomichev, Alexander Duyck, Ong, Boon Leong, Keller,
Jacob E
Cc: Brouer, Jesper, netdev, linux-stm32, linux-arm-kernel,
linux-kernel, bpf, xdp-hints
On Saturday, April 15, 2023 12:39 AM, Jesper Dangaard Brouer <jbrouer@redhat.com> wrote:
>On 14/04/2023 07.26, Song Yoong Siang wrote:
>> Add receive hardware timestamp metadata support via kfunc to XDP
>> receive packets.
>>
>> Suggested-by: Stanislav Fomichev <sdf@google.com>
>> Signed-off-by: Song Yoong Siang <yoong.siang.song@intel.com>
>> Acked-by: Stanislav Fomichev <sdf@google.com>
>> ---
>> drivers/net/ethernet/stmicro/stmmac/stmmac.h | 3 ++
>> .../net/ethernet/stmicro/stmmac/stmmac_main.c | 40 ++++++++++++++++++-
>> 2 files changed, 42 insertions(+), 1 deletion(-)
>>
>> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
>> b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
>> index ac8ccf851708..826ac0ec88c6 100644
>> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h
>> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h
>> @@ -94,6 +94,9 @@ struct stmmac_rx_buffer {
>>
>> struct stmmac_xdp_buff {
>> struct xdp_buff xdp;
>> + struct stmmac_priv *priv;
>> + struct dma_desc *p;
>> + struct dma_desc *np;
>
>Hmm, I don't like the naming of the descriptors as "p" and "np".
>If you insist on this naming, at least we need comments describing that this is.
>
>Does "desc" and "ndesc" make sense? (where "n" means "next")
>
Yup, make sense to have desc and ndesc. I will update the naming in V6.
Thanks & Regards
Siang
>> };
>>
>> struct stmmac_rx_queue {
>> diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
>> b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
>> index 10b9f8912bb2..74f78e5537a3 100644
>> --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
>> +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
>> @@ -5313,10 +5313,15 @@ static int stmmac_rx(struct stmmac_priv *priv,
>> int limit, u32 queue)
>>
>> xdp_init_buff(&ctx.xdp, buf_sz, &rx_q->xdp_rxq);
>> xdp_prepare_buff(&ctx.xdp, page_address(buf->page),
>> - buf->page_offset, buf1_len, false);
>> + buf->page_offset, buf1_len, true);
>>
>> pre_len = ctx.xdp.data_end - ctx.xdp.data_hard_start -
>> buf->page_offset;
>> +
>> + ctx.priv = priv;
>> + ctx.p = p;
>> + ctx.np = np;
>> +
>> skb = stmmac_xdp_run_prog(priv, &ctx.xdp);
>> /* Due xdp_adjust_tail: DMA sync for_device
>> * cover max len CPU touch
>> @@ -7060,6 +7065,37 @@ void stmmac_fpe_handshake(struct stmmac_priv *priv, bool enable)
>> }
>> }
>>
>> +static int stmmac_xdp_rx_timestamp(const struct xdp_md *_ctx, u64
>> +*timestamp) {
>> + const struct stmmac_xdp_buff *ctx = (void *)_ctx;
>> + struct stmmac_priv *priv = ctx->priv;
>> + struct dma_desc *desc = ctx->p;
>> + struct dma_desc *np = ctx->np;
>> + struct dma_desc *p = ctx->p;
>> + u64 ns = 0;
>> +
>> + if (!priv->hwts_rx_en)
>> + return -ENODATA;
>> +
>> + /* For GMAC4, the valid timestamp is from CTX next desc. */
>> + if (priv->plat->has_gmac4 || priv->plat->has_xgmac)
>> + desc = np;
>> +
>> + /* Check if timestamp is available */
>> + if (stmmac_get_rx_timestamp_status(priv, p, np, priv->adv_ts)) {
>> + stmmac_get_timestamp(priv, desc, priv->adv_ts, &ns);
>> + ns -= priv->plat->cdc_error_adj;
>> + *timestamp = ns_to_ktime(ns);
>> + return 0;
>> + }
>> +
>> + return -ENODATA;
>> +}
>> +
>> +static const struct xdp_metadata_ops stmmac_xdp_metadata_ops = {
>> + .xmo_rx_timestamp = stmmac_xdp_rx_timestamp,
>> +};
>> +
>> /**
>> * stmmac_dvr_probe
>> * @device: device pointer
>> @@ -7167,6 +7203,8 @@ int stmmac_dvr_probe(struct device *device,
>>
>> ndev->netdev_ops = &stmmac_netdev_ops;
>>
>> + ndev->xdp_metadata_ops = &stmmac_xdp_metadata_ops;
>> +
>> ndev->hw_features = NETIF_F_SG | NETIF_F_IP_CSUM | NETIF_F_IPV6_CSUM |
>> NETIF_F_RXCSUM;
>> ndev->xdp_features = NETDEV_XDP_ACT_BASIC | NETDEV_XDP_ACT_REDIRECT |
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2023-04-15 5:16 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-04-14 5:26 [xdp-hints] [PATCH net-next v5 0/3] XDP Rx HWTS metadata for stmmac driver Song Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 1/3] net: stmmac: introduce wrapper for struct xdp_buff Song Yoong Siang
2023-04-14 16:04 ` [xdp-hints] " Jesper Dangaard Brouer
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 2/3] net: stmmac: add Rx HWTS metadata to XDP receive pkt Song Yoong Siang
2023-04-14 16:38 ` [xdp-hints] " Jesper Dangaard Brouer
2023-04-15 5:16 ` Song, Yoong Siang
2023-04-14 5:26 ` [xdp-hints] [PATCH net-next v5 3/3] net: stmmac: add Rx HWTS metadata to XDP ZC " Song Yoong Siang
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox