| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-ID: <Y0ZXiDiqxYb7yYmS@unreal>
Date: Wed, 12 Oct 2022 08:58:32 +0300
From: Leon Romanovsky <leon@...nel.org>
To: Peilin Ye <yepeilin.cs@...il.com>
Cc: Eric Dumazet <edumazet@...gle.com>,
"David S. Miller" <davem@...emloft.net>,
Hideaki YOSHIFUJI <yoshfuji@...ux-ipv6.org>,
David Ahern <dsahern@...nel.org>,
Jakub Kicinski <kuba@...nel.org>,
Paolo Abeni <pabeni@...hat.com>,
Peilin Ye <peilin.ye@...edance.com>,
Cong Wang <cong.wang@...edance.com>, netdev@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH net-next v3] net/sock: Introduce trace_sk_data_ready()
On Tue, Oct 11, 2022 at 12:58:56PM -0700, Peilin Ye wrote:
> From: Peilin Ye <peilin.ye@...edance.com>
>
> As suggested by Cong, introduce a tracepoint for all ->sk_data_ready()
> callback implementations. For example:
>
> <...>
> dpkg-deb-7752 [000] ..... 145.660735: sk_data_ready: family=16 protocol=16 func=sock_def_readable
> dpkg-deb-7757 [000] ..... 145.759168: sk_data_ready: family=16 protocol=16 func=sock_def_readable
> dpkg-deb-7758 [000] ..... 145.763956: sk_data_ready: family=16 protocol=16 func=sock_def_readable
> <...>
>
> Suggested-by: Cong Wang <cong.wang@...edance.com>
> Signed-off-by: Peilin Ye <peilin.ye@...edance.com>
> ---
> change since v2:
> - Fix modpost error for modules (kernel test robot)
>
> changes since v1:
> - Move tracepoint into ->sk_data_ready() callback implementations
> (Eric Dumazet)
> - Fix W=1 warning (Jakub Kicinski)
>
> drivers/infiniband/hw/erdma/erdma_cm.c | 3 +++
> drivers/infiniband/sw/siw/siw_cm.c | 5 +++++
> drivers/infiniband/sw/siw/siw_qp.c | 3 +++
> drivers/nvme/host/tcp.c | 3 +++
> drivers/nvme/target/tcp.c | 5 +++++
> drivers/scsi/iscsi_tcp.c | 3 +++
> drivers/soc/qcom/qmi_interface.c | 3 +++
> drivers/target/iscsi/iscsi_target_nego.c | 2 ++
> drivers/xen/pvcalls-back.c | 5 +++++
> fs/dlm/lowcomms.c | 5 +++++
> fs/ocfs2/cluster/tcp.c | 5 +++++
> include/trace/events/sock.h | 24 ++++++++++++++++++++++++
> net/ceph/messenger.c | 4 ++++
> net/core/net-traces.c | 2 ++
> net/core/skmsg.c | 3 +++
> net/core/sock.c | 2 ++
> net/kcm/kcmsock.c | 3 +++
> net/mptcp/subflow.c | 3 +++
> net/qrtr/ns.c | 3 +++
> net/rds/tcp_listen.c | 2 ++
> net/rds/tcp_recv.c | 2 ++
> net/sctp/socket.c | 3 +++
> net/smc/smc_rx.c | 3 +++
> net/sunrpc/svcsock.c | 5 +++++
> net/sunrpc/xprtsock.c | 3 +++
> net/tipc/socket.c | 3 +++
> net/tipc/topsrv.c | 5 +++++
> net/tls/tls_sw.c | 3 +++
> net/xfrm/espintcp.c | 3 +++
> 29 files changed, 118 insertions(+)
>
> diff --git a/drivers/infiniband/hw/erdma/erdma_cm.c b/drivers/infiniband/hw/erdma/erdma_cm.c
> index f13f16479eca..63f314222813 100644
> --- a/drivers/infiniband/hw/erdma/erdma_cm.c
> +++ b/drivers/infiniband/hw/erdma/erdma_cm.c
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
> pr_debug("Entering iscsi_target_sk_data_ready: conn: %p\n", conn);
This can go.
<...>
> + trace_sk_data_ready(sock, __func__);
<...>
> + trace_sk_data_ready(sock, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
> + trace_sk_data_ready(sk, __func__);
<...>
__func__ repetitive pattern hints that it is not best API interface.
> +TRACE_EVENT(sk_data_ready,
> +
> + TP_PROTO(const struct sock *sk, const char *func),
> +
> + TP_ARGS(sk, func),
> +
> + TP_STRUCT__entry(
> + __field(const void *, skaddr)
> + __field(__u16, family)
> + __field(__u16, protocol)
> + __string(func, func)
TRACE_EVENT() is macro defined in .h file, you can safely put __func__
instead.
> + ),
> +
> + TP_fast_assign(
> + __entry->skaddr = sk;
> + __entry->family = sk->sk_family;
> + __entry->protocol = sk->sk_protocol;
> + __assign_str(func, func)
> + ),
> +
> + TP_printk("family=%u protocol=%u func=%s",
> + __entry->family, __entry->protocol, __get_str(func))
> +);
> +
> #endif /* _TRACE_SOCK_H */
Powered by blists - more mailing lists