See upstream commit:
commit d6f19938eb031ee2158272757db33258153ae59c
Author: Yafang Shao <laoar.s...@gmail.com>
Date: Sun Jul 1 23:31:30 2018 +0800
net: expose sk wmem in sock_exceed_buf_limit tracepoint
Currently trace_sock_exceed_buf_limit() only show rmem info,
but wmem limit may also be hit.
So expose wmem info in this tracepoint as well.
Regarding memcg, I think it is better to introduce a new tracepoint(if
that is needed), i.e. trace_memcg_limit_hit other than show memcg info in
trace_sock_exceed_buf_limit.
Signed-off-by: Michael Jeanson <mjean...@efficios.com>
---
instrumentation/events/lttng-module/sock.h | 23 +++++++++++++++++++++-
1 file changed, 22 insertions(+), 1 deletion(-)
diff --git a/instrumentation/events/lttng-module/sock.h
b/instrumentation/events/lttng-module/sock.h
index 4bde039..a1032b3 100644
--- a/instrumentation/events/lttng-module/sock.h
+++ b/instrumentation/events/lttng-module/sock.h
@@ -22,7 +22,28 @@ LTTNG_TRACEPOINT_EVENT(sock_rcvqueue_full,
)
)
-#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,15,0))
+#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,19,0))
+
+LTTNG_TRACEPOINT_EVENT(sock_exceed_buf_limit,
+
+ TP_PROTO(struct sock *sk, struct proto *prot, long allocated, int kind),
+
+ TP_ARGS(sk, prot, allocated, kind),
+
+ TP_FIELDS(
+ ctf_string(name, prot->name)
+ ctf_array(long, sysctl_mem, prot->sysctl_mem, 3)
+ ctf_integer(long, allocated, allocated)
+ ctf_integer(int, sysctl_rmem, sk_get_rmem0(sk, prot))
+ ctf_integer(int, rmem_alloc, atomic_read(&sk->sk_rmem_alloc))
+ ctf_integer(int, sysctl_wmem, sk_get_wmem0(sk, prot))
+ ctf_integer(int, wmem_alloc, refcount_read(&sk->sk_wmem_alloc))
+ ctf_integer(int, wmem_queued, sk->sk_wmem_queued)
+ ctf_integer(int, kind, kind)
+ )
+)
+
+#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(4,15,0))
LTTNG_TRACEPOINT_EVENT(sock_exceed_buf_limit,
--
2.17.1
_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
https://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
