From: Zhu Yanjun <yanjun.zhu@linux.dev>
To: zhenwei pi <zhenwei.pi@linux.dev>,
linux-kernel@vger.kernel.org, linux-rdma@vger.kernel.org,
"yanjun.zhu@linux.dev" <yanjun.zhu@linux.dev>
Cc: zyjzyj2000@gmail.com, jgg@ziepe.ca, leon@kernel.org
Subject: Re: [PATCH v4 3/4] RDMA/rxe: add SENT/RCVD bytes
Date: Mon, 6 Apr 2026 07:55:41 -0700 [thread overview]
Message-ID: <92943fb0-b0ec-443e-a04f-204faba0c9cf@linux.dev> (raw)
In-Reply-To: <20260406132830.435381-4-zhenwei.pi@linux.dev>
在 2026/4/6 6:28, zhenwei pi 写道:
> There is a lack of sent/received counter in bytes.
>
> Signed-off-by: zhenwei pi <zhenwei.pi@linux.dev>
> ---
> drivers/infiniband/sw/rxe/rxe_hw_counters.c | 2 ++
> drivers/infiniband/sw/rxe/rxe_hw_counters.h | 2 ++
> drivers/infiniband/sw/rxe/rxe_net.c | 2 ++
> drivers/infiniband/sw/rxe/rxe_recv.c | 6 ++++++
> drivers/infiniband/sw/rxe/rxe_verbs.h | 6 ++++++
> 5 files changed, 18 insertions(+)
>
> diff --git a/drivers/infiniband/sw/rxe/rxe_hw_counters.c b/drivers/infiniband/sw/rxe/rxe_hw_counters.c
> index 437917a7d8f2..17edaa9a9b9b 100644
> --- a/drivers/infiniband/sw/rxe/rxe_hw_counters.c
> +++ b/drivers/infiniband/sw/rxe/rxe_hw_counters.c
> @@ -22,6 +22,8 @@ static const struct rdma_stat_desc rxe_counter_descs[] = {
> [RXE_CNT_LINK_DOWNED].name = "link_downed",
> [RXE_CNT_RDMA_SEND].name = "rdma_sends",
> [RXE_CNT_RDMA_RECV].name = "rdma_recvs",
> + [RXE_CNT_SENT_BYTES].name = "sent_bytes",
> + [RXE_CNT_RCVD_BYTES].name = "rcvd_bytes",
> };
>
> int rxe_ib_get_hw_stats(struct ib_device *ibdev,
> diff --git a/drivers/infiniband/sw/rxe/rxe_hw_counters.h b/drivers/infiniband/sw/rxe/rxe_hw_counters.h
> index 051f9e1c3852..01b355103cbc 100644
> --- a/drivers/infiniband/sw/rxe/rxe_hw_counters.h
> +++ b/drivers/infiniband/sw/rxe/rxe_hw_counters.h
> @@ -26,6 +26,8 @@ enum rxe_counters {
> RXE_CNT_LINK_DOWNED,
> RXE_CNT_RDMA_SEND,
> RXE_CNT_RDMA_RECV,
> + RXE_CNT_SENT_BYTES,
> + RXE_CNT_RCVD_BYTES,
> RXE_NUM_OF_COUNTERS
> };
>
> diff --git a/drivers/infiniband/sw/rxe/rxe_net.c b/drivers/infiniband/sw/rxe/rxe_net.c
> index 6621d01ac32d..86660031ffa2 100644
> --- a/drivers/infiniband/sw/rxe/rxe_net.c
> +++ b/drivers/infiniband/sw/rxe/rxe_net.c
> @@ -503,6 +503,7 @@ int rxe_xmit_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt,
> int err;
> int is_request = pkt->mask & RXE_REQ_MASK;
> struct rxe_dev *rxe = to_rdev(qp->ibqp.device);
> + unsigned int skblen = skb->len;
> unsigned long flags;
>
> spin_lock_irqsave(&qp->state_lock, flags);
> @@ -526,6 +527,7 @@ int rxe_xmit_packet(struct rxe_qp *qp, struct rxe_pkt_info *pkt,
> }
>
> rxe_counter_inc(rxe, RXE_CNT_SENT_PKTS);
> + rxe_counter_add(rxe, RXE_CNT_SENT_BYTES, skblen);
> goto done;
>
> drop:
> diff --git a/drivers/infiniband/sw/rxe/rxe_recv.c b/drivers/infiniband/sw/rxe/rxe_recv.c
> index 5861e4244049..0d9112e95eae 100644
> --- a/drivers/infiniband/sw/rxe/rxe_recv.c
> +++ b/drivers/infiniband/sw/rxe/rxe_recv.c
> @@ -318,6 +318,7 @@ void rxe_rcv(struct sk_buff *skb)
> int err;
> struct rxe_pkt_info *pkt = SKB_TO_PKT(skb);
> struct rxe_dev *rxe = pkt->rxe;
> + unsigned int skblen = skb->len + sizeof(struct udphdr);
>
> if (unlikely(skb->len < RXE_BTH_BYTES))
> goto drop;
> @@ -341,6 +342,11 @@ void rxe_rcv(struct sk_buff *skb)
> if (unlikely(err))
> goto drop;
>
> + if (skb->protocol == htons(ETH_P_IP))
> + skblen += sizeof(struct iphdr);
> + else if (skb->protocol == htons(ETH_P_IPV6))
> + skblen += sizeof(struct ipv6hdr);
> + rxe_counter_add(rxe, RXE_CNT_RCVD_BYTES, skblen);
From the above source code, I think that you want to calculate total
length starting from the Network Layer (IP Header).
Maybe the following is compact.
"
unsigned int skblen = skb->len - skb_network_offset(skb);
rxe_counter_add(rxe, RXE_CNT_RCVD_BYTES, skblen);
"
Zhu Yanjun
> rxe_counter_inc(rxe, RXE_CNT_RCVD_PKTS);
>
> if (unlikely(bth_qpn(pkt) == IB_MULTICAST_QPN))
> diff --git a/drivers/infiniband/sw/rxe/rxe_verbs.h b/drivers/infiniband/sw/rxe/rxe_verbs.h
> index e800545d1046..0f5ffd94643f 100644
> --- a/drivers/infiniband/sw/rxe/rxe_verbs.h
> +++ b/drivers/infiniband/sw/rxe/rxe_verbs.h
> @@ -455,6 +455,12 @@ static inline void rxe_counter_inc(struct rxe_dev *rxe, enum rxe_counters index)
> atomic64_inc(&rxe->stats_counters[index]);
> }
>
> +static inline void rxe_counter_add(struct rxe_dev *rxe, enum rxe_counters index,
> + s64 val)
> +{
> + atomic64_add(val, &rxe->stats_counters[index]);
> +}
> +
> static inline struct rxe_dev *to_rdev(struct ib_device *dev)
> {
> return dev ? container_of(dev, struct rxe_dev, ib_dev) : NULL;
next prev parent reply other threads:[~2026-04-06 14:55 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-06 13:28 [PATCH v4 0/4] Support PERF MGMT for RXE zhenwei pi
2026-04-06 13:28 ` [PATCH v4 1/4] RDMA/core: Fix memory free for GID table zhenwei pi
2026-04-06 13:28 ` [PATCH v4 2/4] RDMA/rxe: remove rxe_ib_device_get_netdev() and RXE_PORT zhenwei pi
2026-04-06 13:28 ` [PATCH v4 3/4] RDMA/rxe: add SENT/RCVD bytes zhenwei pi
2026-04-06 14:55 ` Zhu Yanjun [this message]
2026-04-07 0:58 ` zhenwei pi
2026-04-06 13:28 ` [PATCH v4 4/4] RDMA/rxe: support perf mgmt GET method zhenwei pi
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=92943fb0-b0ec-443e-a04f-204faba0c9cf@linux.dev \
--to=yanjun.zhu@linux.dev \
--cc=jgg@ziepe.ca \
--cc=leon@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-rdma@vger.kernel.org \
--cc=zhenwei.pi@linux.dev \
--cc=zyjzyj2000@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox