Brad Bishop | 1a4b7ee | 2018-12-16 17:11:34 -0800 | [diff] [blame^] | 1 | From 9e67b4c94b94493123d38379bd9b3eceae23a6f1 Mon Sep 17 00:00:00 2001 |
| 2 | From: Michael Jeanson <mjeanson@efficios.com> |
| 3 | Date: Fri, 7 Sep 2018 12:21:12 -0400 |
| 4 | Subject: [PATCH] Fix: net: expose sk wmem in sock_exceed_buf_limit tracepoint |
| 5 | (4.19) |
| 6 | |
| 7 | See upstream commit: |
| 8 | |
| 9 | commit d6f19938eb031ee2158272757db33258153ae59c |
| 10 | Author: Yafang Shao <laoar.shao@gmail.com> |
| 11 | Date: Sun Jul 1 23:31:30 2018 +0800 |
| 12 | |
| 13 | net: expose sk wmem in sock_exceed_buf_limit tracepoint |
| 14 | |
| 15 | Currently trace_sock_exceed_buf_limit() only show rmem info, |
| 16 | but wmem limit may also be hit. |
| 17 | So expose wmem info in this tracepoint as well. |
| 18 | |
| 19 | Regarding memcg, I think it is better to introduce a new tracepoint(if |
| 20 | that is needed), i.e. trace_memcg_limit_hit other than show memcg info in |
| 21 | trace_sock_exceed_buf_limit. |
| 22 | |
| 23 | Signed-off-by: Michael Jeanson <mjeanson@efficios.com> |
| 24 | Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com> |
| 25 | Upstream-Status: Backport |
| 26 | Signed-off-by: He Zhe <zhe.he@windriver.com> |
| 27 | --- |
| 28 | instrumentation/events/lttng-module/sock.h | 23 ++++++++++++++++++++++- |
| 29 | 1 file changed, 22 insertions(+), 1 deletion(-) |
| 30 | |
| 31 | diff --git a/instrumentation/events/lttng-module/sock.h b/instrumentation/events/lttng-module/sock.h |
| 32 | index 5cd02ca..cd0c92b 100644 |
| 33 | --- a/instrumentation/events/lttng-module/sock.h |
| 34 | +++ b/instrumentation/events/lttng-module/sock.h |
| 35 | @@ -21,7 +21,28 @@ LTTNG_TRACEPOINT_EVENT(sock_rcvqueue_full, |
| 36 | ) |
| 37 | ) |
| 38 | |
| 39 | -#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,15,0)) |
| 40 | +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,19,0)) |
| 41 | + |
| 42 | +LTTNG_TRACEPOINT_EVENT(sock_exceed_buf_limit, |
| 43 | + |
| 44 | + TP_PROTO(struct sock *sk, struct proto *prot, long allocated, int kind), |
| 45 | + |
| 46 | + TP_ARGS(sk, prot, allocated, kind), |
| 47 | + |
| 48 | + TP_FIELDS( |
| 49 | + ctf_string(name, prot->name) |
| 50 | + ctf_array(long, sysctl_mem, prot->sysctl_mem, 3) |
| 51 | + ctf_integer(long, allocated, allocated) |
| 52 | + ctf_integer(int, sysctl_rmem, sk_get_rmem0(sk, prot)) |
| 53 | + ctf_integer(int, rmem_alloc, atomic_read(&sk->sk_rmem_alloc)) |
| 54 | + ctf_integer(int, sysctl_wmem, sk_get_wmem0(sk, prot)) |
| 55 | + ctf_integer(int, wmem_alloc, refcount_read(&sk->sk_wmem_alloc)) |
| 56 | + ctf_integer(int, wmem_queued, sk->sk_wmem_queued) |
| 57 | + ctf_integer(int, kind, kind) |
| 58 | + ) |
| 59 | +) |
| 60 | + |
| 61 | +#elif (LINUX_VERSION_CODE >= KERNEL_VERSION(4,15,0)) |
| 62 | |
| 63 | LTTNG_TRACEPOINT_EVENT(sock_exceed_buf_limit, |
| 64 | |
| 65 | -- |
| 66 | 2.7.4 |
| 67 | |