* [PATCH v1 1/3] svcrdma: Split the svcrdma_wc_receive() tracepoint
@ 2021-09-22 17:00 Chuck Lever
2021-09-22 17:01 ` [PATCH v1 2/3] svcrdma: Split the svcrdma_wc_send() tracepoint Chuck Lever
2021-09-22 17:01 ` [PATCH v1 3/3] svcrdma: Split svcrmda_wc_{read,write} tracepoints Chuck Lever
0 siblings, 2 replies; 3+ messages in thread
From: Chuck Lever @ 2021-09-22 17:00 UTC (permalink / raw)
To: linux-nfs, linux-rdma
There are currently three separate purposes being served by a single
tracepoint here. They need to be split up.
svcrdma_wc_recv:
- status is always zero, so there's no value in recording it.
- vendor_err is meaningless unless status is not zero, so
there's no value in recording it.
- This tracepoint is needed only when developing modifications,
so it should be left disabled most of the time.
svcrdma_wc_recv_flush:
- As above, needed only rarely, and not an error.
svcrdma_wc_recv_err:
- received is always zero, so there's no value in recording it.
- This tracepoint can be left enabled because completion
errors are run-time problems (except for FLUSHED_ERR).
- Tracepoint name now ends in _err to reflect its purpose.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
include/trace/events/rpcrdma.h | 75 +++++++++++++++++++++++++++++++
net/sunrpc/xprtrdma/svc_rdma_recvfrom.c | 9 +++-
2 files changed, 81 insertions(+), 3 deletions(-)
diff --git a/include/trace/events/rpcrdma.h b/include/trace/events/rpcrdma.h
index de4195499592..342d6d7b5cd9 100644
--- a/include/trace/events/rpcrdma.h
+++ b/include/trace/events/rpcrdma.h
@@ -145,6 +145,77 @@ DECLARE_EVENT_CLASS(rpcrdma_receive_completion_class,
), \
TP_ARGS(wc, cid))
+DECLARE_EVENT_CLASS(rpcrdma_receive_success_class,
+ TP_PROTO(
+ const struct ib_wc *wc,
+ const struct rpc_rdma_cid *cid
+ ),
+
+ TP_ARGS(wc, cid),
+
+ TP_STRUCT__entry(
+ __field(u32, cq_id)
+ __field(int, completion_id)
+ __field(u32, received)
+ ),
+
+ TP_fast_assign(
+ __entry->cq_id = cid->ci_queue_id;
+ __entry->completion_id = cid->ci_completion_id;
+ __entry->received = wc->byte_len;
+ ),
+
+ TP_printk("cq.id=%u cid=%d received=%u",
+ __entry->cq_id, __entry->completion_id,
+ __entry->received
+ )
+);
+
+#define DEFINE_RECEIVE_SUCCESS_EVENT(name) \
+ DEFINE_EVENT(rpcrdma_receive_success_class, name, \
+ TP_PROTO( \
+ const struct ib_wc *wc, \
+ const struct rpc_rdma_cid *cid \
+ ), \
+ TP_ARGS(wc, cid))
+
+DECLARE_EVENT_CLASS(rpcrdma_receive_flush_class,
+ TP_PROTO(
+ const struct ib_wc *wc,
+ const struct rpc_rdma_cid *cid
+ ),
+
+ TP_ARGS(wc, cid),
+
+ TP_STRUCT__entry(
+ __field(u32, cq_id)
+ __field(int, completion_id)
+ __field(unsigned long, status)
+ __field(unsigned int, vendor_err)
+ ),
+
+ TP_fast_assign(
+ __entry->cq_id = cid->ci_queue_id;
+ __entry->completion_id = cid->ci_completion_id;
+ __entry->status = wc->status;
+ __entry->vendor_err = wc->vendor_err;
+ ),
+
+ TP_printk("cq.id=%u cid=%d status=%s (%lu/0x%x)",
+ __entry->cq_id, __entry->completion_id,
+ rdma_show_wc_status(__entry->status),
+ __entry->status, __entry->vendor_err
+ )
+);
+
+#define DEFINE_RECEIVE_FLUSH_EVENT(name) \
+ DEFINE_EVENT(rpcrdma_receive_flush_class, name, \
+ TP_PROTO( \
+ const struct ib_wc *wc, \
+ const struct rpc_rdma_cid *cid \
+ ), \
+ TP_ARGS(wc, cid))
+
DECLARE_EVENT_CLASS(xprtrdma_reply_class,
TP_PROTO(
const struct rpcrdma_rep *rep
@@ -1892,7 +1963,9 @@ TRACE_EVENT(svcrdma_post_recv,
)
);
-DEFINE_RECEIVE_COMPLETION_EVENT(svcrdma_wc_receive);
+DEFINE_RECEIVE_SUCCESS_EVENT(svcrdma_wc_recv);
+DEFINE_RECEIVE_FLUSH_EVENT(svcrdma_wc_recv_flush);
+DEFINE_RECEIVE_FLUSH_EVENT(svcrdma_wc_recv_err);
TRACE_EVENT(svcrdma_rq_post_err,
TP_PROTO(
diff --git a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
index 6be23ce7a93d..cf76a6ad127b 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c
@@ -330,9 +330,9 @@ static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
/* WARNING: Only wc->wr_cqe and wc->status are reliable */
ctxt = container_of(cqe, struct svc_rdma_recv_ctxt, rc_cqe);
- trace_svcrdma_wc_receive(wc, &ctxt->rc_cid);
if (wc->status != IB_WC_SUCCESS)
goto flushed;
+ trace_svcrdma_wc_recv(wc, &ctxt->rc_cid);
/* If receive posting fails, the connection is about to be
* lost anyway. The server will not be able to send a reply
@@ -345,7 +345,7 @@ static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
*/
if (rdma->sc_pending_recvs < rdma->sc_max_requests)
if (!svc_rdma_refresh_recvs(rdma, rdma->sc_recv_batch, false))
- goto flushed;
+ goto dropped;
/* All wc fields are now known to be valid */
ctxt->rc_byte_len = wc->byte_len;
@@ -360,6 +360,11 @@ static void svc_rdma_wc_receive(struct ib_cq *cq, struct ib_wc *wc)
return;
flushed:
+ if (wc->status == IB_WC_WR_FLUSH_ERR)
+ trace_svcrdma_wc_recv_flush(wc, &ctxt->rc_cid);
+ else
+ trace_svcrdma_wc_recv_err(wc, &ctxt->rc_cid);
+dropped:
svc_rdma_recv_ctxt_put(rdma, ctxt);
svc_xprt_deferred_close(&rdma->sc_xprt);
}
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [PATCH v1 2/3] svcrdma: Split the svcrdma_wc_send() tracepoint
2021-09-22 17:00 [PATCH v1 1/3] svcrdma: Split the svcrdma_wc_receive() tracepoint Chuck Lever
@ 2021-09-22 17:01 ` Chuck Lever
2021-09-22 17:01 ` [PATCH v1 3/3] svcrdma: Split svcrmda_wc_{read,write} tracepoints Chuck Lever
1 sibling, 0 replies; 3+ messages in thread
From: Chuck Lever @ 2021-09-22 17:01 UTC (permalink / raw)
To: linux-nfs, linux-rdma
There are currently three separate purposes being served by a single
tracepoint here. They need to be split up.
svcrdma_wc_send:
- status is always zero, so there's no value in recording it.
- vendor_err is meaningless unless status is not zero, so
there's no value in recording it.
- This tracepoint is needed only when developing modifications,
so it should be left disabled most of the time.
svcrdma_wc_send_flush:
- As above, needed only rarely, and not an error.
svcrdma_wc_send_err:
- This tracepoint can be left persistently enabled because
completion errors are run-time problems (except for FLUSHED_ERR).
- Tracepoint name now ends in _err to reflect its purpose.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
include/trace/events/rpcrdma.h | 72 +++++++++++++++++++++++++++++++++
net/sunrpc/xprtrdma/svc_rdma_sendto.c | 14 +++++-
2 files changed, 82 insertions(+), 4 deletions(-)
diff --git a/include/trace/events/rpcrdma.h b/include/trace/events/rpcrdma.h
index 342d6d7b5cd9..1d7c12f65f87 100644
--- a/include/trace/events/rpcrdma.h
+++ b/include/trace/events/rpcrdma.h
@@ -60,6 +60,74 @@ DECLARE_EVENT_CLASS(rpcrdma_completion_class,
), \
TP_ARGS(wc, cid))
+DECLARE_EVENT_CLASS(rpcrdma_send_completion_class,
+ TP_PROTO(
+ const struct ib_wc *wc,
+ const struct rpc_rdma_cid *cid
+ ),
+
+ TP_ARGS(wc, cid),
+
+ TP_STRUCT__entry(
+ __field(u32, cq_id)
+ __field(int, completion_id)
+ ),
+
+ TP_fast_assign(
+ __entry->cq_id = cid->ci_queue_id;
+ __entry->completion_id = cid->ci_completion_id;
+ ),
+
+ TP_printk("cq.id=%u cid=%d",
+ __entry->cq_id, __entry->completion_id
+ )
+);
+
+#define DEFINE_SEND_COMPLETION_EVENT(name) \
+ DEFINE_EVENT(rpcrdma_send_completion_class, name, \
+ TP_PROTO( \
+ const struct ib_wc *wc, \
+ const struct rpc_rdma_cid *cid \
+ ), \
+ TP_ARGS(wc, cid))
+
+DECLARE_EVENT_CLASS(rpcrdma_send_flush_class,
+ TP_PROTO(
+ const struct ib_wc *wc,
+ const struct rpc_rdma_cid *cid
+ ),
+
+ TP_ARGS(wc, cid),
+
+ TP_STRUCT__entry(
+ __field(u32, cq_id)
+ __field(int, completion_id)
+ __field(unsigned long, status)
+ __field(unsigned int, vendor_err)
+ ),
+
+ TP_fast_assign(
+ __entry->cq_id = cid->ci_queue_id;
+ __entry->completion_id = cid->ci_completion_id;
+ __entry->status = wc->status;
+ __entry->vendor_err = wc->vendor_err;
+ ),
+
+ TP_printk("cq.id=%u cid=%d status=%s (%lu/0x%x)",
+ __entry->cq_id, __entry->completion_id,
+ rdma_show_wc_status(__entry->status),
+ __entry->status, __entry->vendor_err
+ )
+);
+
+#define DEFINE_SEND_FLUSH_EVENT(name) \
+ DEFINE_EVENT(rpcrdma_send_flush_class, name, \
+ TP_PROTO( \
+ const struct ib_wc *wc, \
+ const struct rpc_rdma_cid *cid \
+ ), \
+ TP_ARGS(wc, cid))
+
DECLARE_EVENT_CLASS(rpcrdma_mr_completion_class,
TP_PROTO(
const struct ib_wc *wc,
@@ -1939,7 +2007,9 @@ TRACE_EVENT(svcrdma_post_send,
)
);
-DEFINE_COMPLETION_EVENT(svcrdma_wc_send);
+DEFINE_SEND_COMPLETION_EVENT(svcrdma_wc_send);
+DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_send_flush);
+DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_send_err);
TRACE_EVENT(svcrdma_post_recv,
TP_PROTO(
diff --git a/net/sunrpc/xprtrdma/svc_rdma_sendto.c b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
index 599021b2391d..22a871e6fe4d 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_sendto.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_sendto.c
@@ -280,13 +280,21 @@ static void svc_rdma_wc_send(struct ib_cq *cq, struct ib_wc *wc)
struct svc_rdma_send_ctxt *ctxt =
container_of(cqe, struct svc_rdma_send_ctxt, sc_cqe);
- trace_svcrdma_wc_send(wc, &ctxt->sc_cid);
-
svc_rdma_wake_send_waiters(rdma, 1);
complete(&ctxt->sc_done);
if (unlikely(wc->status != IB_WC_SUCCESS))
- svc_xprt_deferred_close(&rdma->sc_xprt);
+ goto flushed;
+
+ trace_svcrdma_wc_send(wc, &ctxt->sc_cid);
+ return;
+
+flushed:
+ if (wc->status != IB_WC_WR_FLUSH_ERR)
+ trace_svcrdma_wc_send_err(wc, &ctxt->sc_cid);
+ else
+ trace_svcrdma_wc_send_flush(wc, &ctxt->sc_cid);
+ svc_xprt_deferred_close(&rdma->sc_xprt);
}
/**
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [PATCH v1 3/3] svcrdma: Split svcrmda_wc_{read,write} tracepoints
2021-09-22 17:00 [PATCH v1 1/3] svcrdma: Split the svcrdma_wc_receive() tracepoint Chuck Lever
2021-09-22 17:01 ` [PATCH v1 2/3] svcrdma: Split the svcrdma_wc_send() tracepoint Chuck Lever
@ 2021-09-22 17:01 ` Chuck Lever
1 sibling, 0 replies; 3+ messages in thread
From: Chuck Lever @ 2021-09-22 17:01 UTC (permalink / raw)
To: linux-nfs, linux-rdma
There are currently three separate purposes being served by single
tracepoints. Split them up, as was done with wc_send.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
---
include/trace/events/rpcrdma.h | 38 +++++++++++++++++++++++++++++++++++--
net/sunrpc/xprtrdma/svc_rdma_rw.c | 30 +++++++++++++++++++++++++----
2 files changed, 62 insertions(+), 6 deletions(-)
diff --git a/include/trace/events/rpcrdma.h b/include/trace/events/rpcrdma.h
index 1d7c12f65f87..b5a1388e51a4 100644
--- a/include/trace/events/rpcrdma.h
+++ b/include/trace/events/rpcrdma.h
@@ -2099,8 +2099,42 @@ DEFINE_POST_CHUNK_EVENT(read);
DEFINE_POST_CHUNK_EVENT(write);
DEFINE_POST_CHUNK_EVENT(reply);
-DEFINE_COMPLETION_EVENT(svcrdma_wc_read);
-DEFINE_COMPLETION_EVENT(svcrdma_wc_write);
+TRACE_EVENT(svcrdma_wc_read,
+ TP_PROTO(
+ const struct ib_wc *wc,
+ const struct rpc_rdma_cid *cid,
+ unsigned int totalbytes,
+ const ktime_t posttime
+ ),
+
+ TP_ARGS(wc, cid, totalbytes, posttime),
+
+ TP_STRUCT__entry(
+ __field(u32, cq_id)
+ __field(int, completion_id)
+ __field(s64, read_latency)
+ __field(unsigned int, totalbytes)
+ ),
+
+ TP_fast_assign(
+ __entry->cq_id = cid->ci_queue_id;
+ __entry->completion_id = cid->ci_completion_id;
+ __entry->totalbytes = totalbytes;
+ __entry->read_latency = ktime_us_delta(ktime_get(), posttime);
+ ),
+
+ TP_printk("cq.id=%u cid=%d totalbytes=%u latency-us=%lld",
+ __entry->cq_id, __entry->completion_id,
+ __entry->totalbytes, __entry->read_latency
+ )
+);
+
+DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_read_flush);
+DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_read_err);
+
+DEFINE_SEND_COMPLETION_EVENT(svcrdma_wc_write);
+DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_write_flush);
+DEFINE_SEND_FLUSH_EVENT(svcrdma_wc_write_err);
TRACE_EVENT(svcrdma_qp_error,
TP_PROTO(
diff --git a/net/sunrpc/xprtrdma/svc_rdma_rw.c b/net/sunrpc/xprtrdma/svc_rdma_rw.c
index e27433f08ca7..5f0155fdefc7 100644
--- a/net/sunrpc/xprtrdma/svc_rdma_rw.c
+++ b/net/sunrpc/xprtrdma/svc_rdma_rw.c
@@ -155,6 +155,7 @@ struct svc_rdma_chunk_ctxt {
struct ib_cqe cc_cqe;
struct svcxprt_rdma *cc_rdma;
struct list_head cc_rwctxts;
+ ktime_t cc_posttime;
int cc_sqecount;
enum ib_wc_status cc_status;
struct completion cc_done;
@@ -267,7 +268,16 @@ static void svc_rdma_write_done(struct ib_cq *cq, struct ib_wc *wc)
struct svc_rdma_write_info *info =
container_of(cc, struct svc_rdma_write_info, wi_cc);
- trace_svcrdma_wc_write(wc, &cc->cc_cid);
+ switch (wc->status) {
+ case IB_WC_SUCCESS:
+ trace_svcrdma_wc_write(wc, &cc->cc_cid);
+ break;
+ case IB_WC_WR_FLUSH_ERR:
+ trace_svcrdma_wc_write_flush(wc, &cc->cc_cid);
+ break;
+ default:
+ trace_svcrdma_wc_write_err(wc, &cc->cc_cid);
+ }
svc_rdma_wake_send_waiters(rdma, cc->cc_sqecount);
@@ -320,11 +330,22 @@ static void svc_rdma_wc_read_done(struct ib_cq *cq, struct ib_wc *wc)
struct ib_cqe *cqe = wc->wr_cqe;
struct svc_rdma_chunk_ctxt *cc =
container_of(cqe, struct svc_rdma_chunk_ctxt, cc_cqe);
- struct svcxprt_rdma *rdma = cc->cc_rdma;
+ struct svc_rdma_read_info *info;
- trace_svcrdma_wc_read(wc, &cc->cc_cid);
+ switch (wc->status) {
+ case IB_WC_SUCCESS:
+ info = container_of(cc, struct svc_rdma_read_info, ri_cc);
+ trace_svcrdma_wc_read(wc, &cc->cc_cid, info->ri_totalbytes,
+ cc->cc_posttime);
+ break;
+ case IB_WC_WR_FLUSH_ERR:
+ trace_svcrdma_wc_read_flush(wc, &cc->cc_cid);
+ break;
+ default:
+ trace_svcrdma_wc_read_err(wc, &cc->cc_cid);
+ }
- svc_rdma_wake_send_waiters(rdma, cc->cc_sqecount);
+ svc_rdma_wake_send_waiters(cc->cc_rdma, cc->cc_sqecount);
cc->cc_status = wc->status;
complete(&cc->cc_done);
return;
@@ -363,6 +384,7 @@ static int svc_rdma_post_chunk_ctxt(struct svc_rdma_chunk_ctxt *cc)
do {
if (atomic_sub_return(cc->cc_sqecount,
&rdma->sc_sq_avail) > 0) {
+ cc->cc_posttime = ktime_get();
ret = ib_post_send(rdma->sc_qp, first_wr, &bad_wr);
if (ret)
break;
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2021-09-22 17:01 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-09-22 17:00 [PATCH v1 1/3] svcrdma: Split the svcrdma_wc_receive() tracepoint Chuck Lever
2021-09-22 17:01 ` [PATCH v1 2/3] svcrdma: Split the svcrdma_wc_send() tracepoint Chuck Lever
2021-09-22 17:01 ` [PATCH v1 3/3] svcrdma: Split svcrmda_wc_{read,write} tracepoints Chuck Lever
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.