From:   Chuck Lever <chuck.lever@oracle.com>
To:     linux-nfs@vger.kernel.org
Cc:     linux-trace-devel@vger.kernel.org
Subject: [PATCH v2 0/2] Fix sockaddr handling in NFSD trace points
Date:   Mon, 10 Jan 2022 10:56:49 -0500
Message-Id:  <164182978641.8391.8277203495236105391.stgit@bazille.1015granger.net>
User-Agent: StGit/1.4
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 8bit
Precedence: bulk

The patches in this series address a simple buffer over-read bug in
the Linux NFS server.

However I was thinking it would be nice to have trace helpers to
deal safely with generic socket addresses. I'd like to be able to
treat them the same way we currently treat strings. So for example:


#define field_sockaddr(field, len)  __dynamic_array(u8, field, len)
#define assign_sockaddr(dest, src, len)  memcpy(__get_dynamic_array(dest), src, len)
#define __get_sockaddr(field)  ((struct sockaddr *)__get_dynamic_array(field))

TRACE_EVENT(sockaddr_example,
        TP_PROTO(
                const struct sockaddr *sap,
                size_t salen
        ),  
        TP_ARGS(sap, salen),
        TP_STRUCT__entry(
                __field_sockaddr(addr, salen)
        ),  
        TP_fast_assign(
                __assign_sockaddr(addr, sap, salen);
        ),  
        TP_printk("addr=%pIS", __get_sockaddr(addr))
);


should be able to store any address family in a dynamically-sized
array field (addr).

I haven't quite been able to figure out how to handle the 
TP_printk() part of this equation. `trace-cmd report` displays
something like "addr=ARG TYPE NOT FIELD BUT 7". 

Thoughts or advice appreciated.

---

Chuck Lever (2):
      SUNRPC: Fix sockaddr handling in the svc_xprt_create_error trace point
      SUNRPC: Fix sockaddr handling in svcsock_accept_class trace points


 include/trace/events/sunrpc.h | 13 +++++++------
 net/sunrpc/svc_xprt.c         |  2 +-
 2 files changed, 8 insertions(+), 7 deletions(-)

--
Chuck Lever