Commit edffb84c authored by Trond Myklebust's avatar Trond Myklebust
Browse files

Merge tag 'nfs-rdma-for-5.11-1' of git://git.linux-nfs.org/projects/anna/linux-nfs into linux-next 



NFSoRDmA Client updates for Linux 5.11

Cleanups and improvements:
  - Remove use of raw kernel memory addresses in tracepoints
  - Replace dprintk() call sites in ERR_CHUNK path
  - Trace unmap sync calls
  - Optimize MR DMA-unmapping

Signed-off-by: default avatarTrond Myklebust <trond.myklebust@hammerspace.com>
parents 5c3485bb 7a03aeb6

include/trace/events/rpcrdma.h

+212 −238
Original line number Diff line number Diff line
@@ -60,7 +60,7 @@ DECLARE_EVENT_CLASS(rpcrdma_completion_class, 
				),					\

				TP_ARGS(wc, cid))



DECLARE_EVENT_CLASS(xprtrdma_reply_event,

DECLARE_EVENT_CLASS(xprtrdma_reply_class,

	TP_PROTO(

		const struct rpcrdma_rep *rep

	),
@@ -68,29 +68,30 @@ DECLARE_EVENT_CLASS(xprtrdma_reply_event, 
	TP_ARGS(rep),



	TP_STRUCT__entry(

		__field(const void *, rep)

		__field(const void *, r_xprt)

		__field(u32, xid)

		__field(u32, version)

		__field(u32, proc)

		__string(addr, rpcrdma_addrstr(rep->rr_rxprt))

		__string(port, rpcrdma_portstr(rep->rr_rxprt))

	),



	TP_fast_assign(

		__entry->rep = rep;

		__entry->r_xprt = rep->rr_rxprt;

		__entry->xid = be32_to_cpu(rep->rr_xid);

		__entry->version = be32_to_cpu(rep->rr_vers);

		__entry->proc = be32_to_cpu(rep->rr_proc);

		__assign_str(addr, rpcrdma_addrstr(rep->rr_rxprt));

		__assign_str(port, rpcrdma_portstr(rep->rr_rxprt));

	),



	TP_printk("rxprt %p xid=0x%08x rep=%p: version %u proc %u",

		__entry->r_xprt, __entry->xid, __entry->rep,

		__entry->version, __entry->proc

	TP_printk("peer=[%s]:%s xid=0x%08x version=%u proc=%u",

		__get_str(addr), __get_str(port),

		__entry->xid, __entry->version, __entry->proc

	)

);



#define DEFINE_REPLY_EVENT(name)					\

		DEFINE_EVENT(xprtrdma_reply_event, name,		\

		DEFINE_EVENT(xprtrdma_reply_class,			\

				xprtrdma_reply_##name##_err,		\

				TP_PROTO(				\

					const struct rpcrdma_rep *rep	\

				),					\
@@ -261,54 +262,67 @@ DECLARE_EVENT_CLASS(xprtrdma_wrch_event, 
				),					\

				TP_ARGS(task, mr, nsegs))



DECLARE_EVENT_CLASS(xprtrdma_frwr_done,

TRACE_DEFINE_ENUM(DMA_BIDIRECTIONAL);

TRACE_DEFINE_ENUM(DMA_TO_DEVICE);

TRACE_DEFINE_ENUM(DMA_FROM_DEVICE);

TRACE_DEFINE_ENUM(DMA_NONE);



#define xprtrdma_show_direction(x)					\

		__print_symbolic(x,					\

				{ DMA_BIDIRECTIONAL, "BIDIR" },		\

				{ DMA_TO_DEVICE, "TO_DEVICE" },		\

				{ DMA_FROM_DEVICE, "FROM_DEVICE" },	\

				{ DMA_NONE, "NONE" })



DECLARE_EVENT_CLASS(xprtrdma_mr_class,

	TP_PROTO(

		const struct ib_wc *wc,

		const struct rpcrdma_frwr *frwr

		const struct rpcrdma_mr *mr

	),



	TP_ARGS(wc, frwr),

	TP_ARGS(mr),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(u32, mr_id)

		__field(unsigned int, status)

		__field(unsigned int, vendor_err)

		__field(int, nents)

		__field(u32, handle)

		__field(u32, length)

		__field(u64, offset)

		__field(u32, dir)

	),



	TP_fast_assign(

		__entry->mr_id = frwr->fr_mr->res.id;

		__entry->status = wc->status;

		__entry->vendor_err = __entry->status ? wc->vendor_err : 0;

		const struct rpcrdma_req *req = mr->mr_req;

		const struct rpc_task *task = req->rl_slot.rq_task;



		__entry->task_id = task->tk_pid;

		__entry->client_id = task->tk_client->cl_clid;

		__entry->mr_id  = mr->frwr.fr_mr->res.id;

		__entry->nents  = mr->mr_nents;

		__entry->handle = mr->mr_handle;

		__entry->length = mr->mr_length;

		__entry->offset = mr->mr_offset;

		__entry->dir    = mr->mr_dir;

	),



	TP_printk(

		"mr.id=%u: %s (%u/0x%x)",

		__entry->mr_id, rdma_show_wc_status(__entry->status),

		__entry->status, __entry->vendor_err

	TP_printk("task:%u@%u mr.id=%u nents=%d %u@0x%016llx:0x%08x (%s)",

		__entry->task_id, __entry->client_id,

		__entry->mr_id, __entry->nents, __entry->length,

		(unsigned long long)__entry->offset, __entry->handle,

		xprtrdma_show_direction(__entry->dir)

	)

);



#define DEFINE_FRWR_DONE_EVENT(name)					\

		DEFINE_EVENT(xprtrdma_frwr_done, name,			\

#define DEFINE_MR_EVENT(name)						\

		DEFINE_EVENT(xprtrdma_mr_class,				\

				xprtrdma_mr_##name,			\

				TP_PROTO(				\

					const struct ib_wc *wc,		\

					const struct rpcrdma_frwr *frwr	\

					const struct rpcrdma_mr *mr	\

				),					\

				TP_ARGS(wc, frwr))



TRACE_DEFINE_ENUM(DMA_BIDIRECTIONAL);

TRACE_DEFINE_ENUM(DMA_TO_DEVICE);

TRACE_DEFINE_ENUM(DMA_FROM_DEVICE);

TRACE_DEFINE_ENUM(DMA_NONE);



#define xprtrdma_show_direction(x)					\

		__print_symbolic(x,					\

				{ DMA_BIDIRECTIONAL, "BIDIR" },		\

				{ DMA_TO_DEVICE, "TO_DEVICE" },		\

				{ DMA_FROM_DEVICE, "FROM_DEVICE" },	\

				{ DMA_NONE, "NONE" })

				TP_ARGS(mr))



DECLARE_EVENT_CLASS(xprtrdma_mr,

DECLARE_EVENT_CLASS(xprtrdma_anonymous_mr_class,

	TP_PROTO(

		const struct rpcrdma_mr *mr

	),
@@ -340,45 +354,47 @@ DECLARE_EVENT_CLASS(xprtrdma_mr, 
	)

);



#define DEFINE_MR_EVENT(name) \

		DEFINE_EVENT(xprtrdma_mr, xprtrdma_mr_##name, \

#define DEFINE_ANON_MR_EVENT(name)					\

		DEFINE_EVENT(xprtrdma_anonymous_mr_class,		\

				xprtrdma_mr_##name,			\

				TP_PROTO(				\

					const struct rpcrdma_mr *mr	\

				),					\

				TP_ARGS(mr))



DECLARE_EVENT_CLASS(xprtrdma_cb_event,

DECLARE_EVENT_CLASS(xprtrdma_callback_class,

	TP_PROTO(

		const struct rpcrdma_xprt *r_xprt,

		const struct rpc_rqst *rqst

	),



	TP_ARGS(rqst),

	TP_ARGS(r_xprt, rqst),



	TP_STRUCT__entry(

		__field(const void *, rqst)

		__field(const void *, rep)

		__field(const void *, req)

		__field(u32, xid)

		__string(addr, rpcrdma_addrstr(r_xprt))

		__string(port, rpcrdma_portstr(r_xprt))

	),



	TP_fast_assign(

		__entry->rqst = rqst;

		__entry->req = rpcr_to_rdmar(rqst);

		__entry->rep = rpcr_to_rdmar(rqst)->rl_reply;

		__entry->xid = be32_to_cpu(rqst->rq_xid);

		__assign_str(addr, rpcrdma_addrstr(r_xprt));

		__assign_str(port, rpcrdma_portstr(r_xprt));

	),



	TP_printk("xid=0x%08x, rqst=%p req=%p rep=%p",

		__entry->xid, __entry->rqst, __entry->req, __entry->rep

	TP_printk("peer=[%s]:%s xid=0x%08x",

		__get_str(addr), __get_str(port), __entry->xid

	)

);



#define DEFINE_CB_EVENT(name)						\

		DEFINE_EVENT(xprtrdma_cb_event, name,			\

#define DEFINE_CALLBACK_EVENT(name)					\

		DEFINE_EVENT(xprtrdma_callback_class,			\

				xprtrdma_cb_##name,			\

				TP_PROTO(				\

					const struct rpcrdma_xprt *r_xprt, \

					const struct rpc_rqst *rqst	\

				),					\

				TP_ARGS(rqst))

				TP_ARGS(r_xprt, rqst))



/**

 ** Connection events
@@ -549,61 +565,33 @@ TRACE_EVENT(xprtrdma_createmrs, 
	)

);



TRACE_EVENT(xprtrdma_mr_get,

TRACE_EVENT(xprtrdma_nomrs_err,

	TP_PROTO(

		const struct rpcrdma_xprt *r_xprt,

		const struct rpcrdma_req *req

	),



	TP_ARGS(req),

	TP_ARGS(r_xprt, req),



	TP_STRUCT__entry(

		__field(const void *, req)

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(u32, xid)

	),



	TP_fast_assign(

		const struct rpc_rqst *rqst = &req->rl_slot;



		__entry->req = req;

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client->cl_clid;

		__entry->xid = be32_to_cpu(rqst->rq_xid);

	),



	TP_printk("task:%u@%u xid=0x%08x req=%p",

		__entry->task_id, __entry->client_id, __entry->xid,

		__entry->req

	)

);



TRACE_EVENT(xprtrdma_nomrs,

	TP_PROTO(

		const struct rpcrdma_req *req

	),



	TP_ARGS(req),



	TP_STRUCT__entry(

		__field(const void *, req)

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(u32, xid)

		__string(addr, rpcrdma_addrstr(r_xprt))

		__string(port, rpcrdma_portstr(r_xprt))

	),



	TP_fast_assign(

		const struct rpc_rqst *rqst = &req->rl_slot;



		__entry->req = req;

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client->cl_clid;

		__entry->xid = be32_to_cpu(rqst->rq_xid);

		__assign_str(addr, rpcrdma_addrstr(r_xprt));

		__assign_str(port, rpcrdma_portstr(r_xprt));

	),



	TP_printk("task:%u@%u xid=0x%08x req=%p",

		__entry->task_id, __entry->client_id, __entry->xid,

		__entry->req

	TP_printk("peer=[%s]:%s task:%u@%u",

		__get_str(addr), __get_str(port),

		__entry->task_id, __entry->client_id

	)

);
@@ -735,8 +723,8 @@ TRACE_EVENT(xprtrdma_post_send, 
	TP_ARGS(req),



	TP_STRUCT__entry(

		__field(const void *, req)

		__field(const void *, sc)

		__field(u32, cq_id)

		__field(int, completion_id)

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(int, num_sge)
@@ -745,20 +733,21 @@ TRACE_EVENT(xprtrdma_post_send, 


	TP_fast_assign(

		const struct rpc_rqst *rqst = &req->rl_slot;

		const struct rpcrdma_sendctx *sc = req->rl_sendctx;



		__entry->cq_id = sc->sc_cid.ci_queue_id;

		__entry->completion_id = sc->sc_cid.ci_completion_id;

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client ?

				     rqst->rq_task->tk_client->cl_clid : -1;

		__entry->req = req;

		__entry->sc = req->rl_sendctx;

		__entry->num_sge = req->rl_wr.num_sge;

		__entry->signaled = req->rl_wr.send_flags & IB_SEND_SIGNALED;

	),



	TP_printk("task:%u@%u req=%p sc=%p (%d SGE%s) %s",

	TP_printk("task:%u@%u cq.id=%u cid=%d (%d SGE%s) %s",

		__entry->task_id, __entry->client_id,

		__entry->req, __entry->sc, __entry->num_sge,

		(__entry->num_sge == 1 ? "" : "s"),

		__entry->cq_id, __entry->completion_id,

		__entry->num_sge, (__entry->num_sge == 1 ? "" : "s"),

		(__entry->signaled ? "signaled" : "")

	)

);
@@ -771,15 +760,17 @@ TRACE_EVENT(xprtrdma_post_recv, 
	TP_ARGS(rep),



	TP_STRUCT__entry(

		__field(const void *, rep)

		__field(u32, cq_id)

		__field(int, completion_id)

	),



	TP_fast_assign(

		__entry->rep = rep;

		__entry->cq_id = rep->rr_cid.ci_queue_id;

		__entry->completion_id = rep->rr_cid.ci_completion_id;

	),



	TP_printk("rep=%p",

		__entry->rep

	TP_printk("cq.id=%d cid=%d",

		__entry->cq_id, __entry->completion_id

	)

);
@@ -816,7 +807,7 @@ TRACE_EVENT(xprtrdma_post_recvs, 
	)

);



TRACE_EVENT(xprtrdma_post_linv,

TRACE_EVENT(xprtrdma_post_linv_err,

	TP_PROTO(

		const struct rpcrdma_req *req,

		int status
@@ -825,19 +816,21 @@ TRACE_EVENT(xprtrdma_post_linv, 
	TP_ARGS(req, status),



	TP_STRUCT__entry(

		__field(const void *, req)

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(int, status)

		__field(u32, xid)

	),



	TP_fast_assign(

		__entry->req = req;

		const struct rpc_task *task = req->rl_slot.rq_task;



		__entry->task_id = task->tk_pid;

		__entry->client_id = task->tk_client->cl_clid;

		__entry->status = status;

		__entry->xid = be32_to_cpu(req->rl_slot.rq_xid);

	),



	TP_printk("req=%p xid=0x%08x status=%d",

		__entry->req, __entry->xid, __entry->status

	TP_printk("task:%u@%u status=%d",

		__entry->task_id, __entry->client_id, __entry->status

	)

);
@@ -845,75 +838,12 @@ TRACE_EVENT(xprtrdma_post_linv, 
 ** Completion events

 **/



TRACE_EVENT(xprtrdma_wc_send,

	TP_PROTO(

		const struct rpcrdma_sendctx *sc,

		const struct ib_wc *wc

	),



	TP_ARGS(sc, wc),



	TP_STRUCT__entry(

		__field(const void *, req)

		__field(const void *, sc)

		__field(unsigned int, unmap_count)

		__field(unsigned int, status)

		__field(unsigned int, vendor_err)

	),



	TP_fast_assign(

		__entry->req = sc->sc_req;

		__entry->sc = sc;

		__entry->unmap_count = sc->sc_unmap_count;

		__entry->status = wc->status;

		__entry->vendor_err = __entry->status ? wc->vendor_err : 0;

	),



	TP_printk("req=%p sc=%p unmapped=%u: %s (%u/0x%x)",

		__entry->req, __entry->sc, __entry->unmap_count,

		rdma_show_wc_status(__entry->status),

		__entry->status, __entry->vendor_err

	)

);



TRACE_EVENT(xprtrdma_wc_receive,

	TP_PROTO(

		const struct ib_wc *wc

	),



	TP_ARGS(wc),



	TP_STRUCT__entry(

		__field(const void *, rep)

		__field(u32, byte_len)

		__field(unsigned int, status)

		__field(u32, vendor_err)

	),



	TP_fast_assign(

		__entry->rep = container_of(wc->wr_cqe, struct rpcrdma_rep,

					    rr_cqe);

		__entry->status = wc->status;

		if (wc->status) {

			__entry->byte_len = 0;

			__entry->vendor_err = wc->vendor_err;

		} else {

			__entry->byte_len = wc->byte_len;

			__entry->vendor_err = 0;

		}

	),



	TP_printk("rep=%p %u bytes: %s (%u/0x%x)",

		__entry->rep, __entry->byte_len,

		rdma_show_wc_status(__entry->status),

		__entry->status, __entry->vendor_err

	)

);



DEFINE_FRWR_DONE_EVENT(xprtrdma_wc_fastreg);

DEFINE_FRWR_DONE_EVENT(xprtrdma_wc_li);

DEFINE_FRWR_DONE_EVENT(xprtrdma_wc_li_wake);

DEFINE_FRWR_DONE_EVENT(xprtrdma_wc_li_done);

DEFINE_COMPLETION_EVENT(xprtrdma_wc_receive);

DEFINE_COMPLETION_EVENT(xprtrdma_wc_send);

DEFINE_COMPLETION_EVENT(xprtrdma_wc_fastreg);

DEFINE_COMPLETION_EVENT(xprtrdma_wc_li);

DEFINE_COMPLETION_EVENT(xprtrdma_wc_li_wake);

DEFINE_COMPLETION_EVENT(xprtrdma_wc_li_done);



TRACE_EVENT(xprtrdma_frwr_alloc,

	TP_PROTO(
@@ -1036,9 +966,9 @@ TRACE_EVENT(xprtrdma_frwr_maperr, 


DEFINE_MR_EVENT(localinv);

DEFINE_MR_EVENT(map);

DEFINE_MR_EVENT(unmap);

DEFINE_MR_EVENT(reminv);

DEFINE_MR_EVENT(recycle);



DEFINE_ANON_MR_EVENT(unmap);

DEFINE_ANON_MR_EVENT(recycle);



TRACE_EVENT(xprtrdma_dma_maperr,

	TP_PROTO(
@@ -1066,17 +996,14 @@ TRACE_EVENT(xprtrdma_reply, 
	TP_PROTO(

		const struct rpc_task *task,

		const struct rpcrdma_rep *rep,

		const struct rpcrdma_req *req,

		unsigned int credits

	),



	TP_ARGS(task, rep, req, credits),

	TP_ARGS(task, rep, credits),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(const void *, rep)

		__field(const void *, req)

		__field(u32, xid)

		__field(unsigned int, credits)

	),
@@ -1084,49 +1011,102 @@ TRACE_EVENT(xprtrdma_reply, 
	TP_fast_assign(

		__entry->task_id = task->tk_pid;

		__entry->client_id = task->tk_client->cl_clid;

		__entry->rep = rep;

		__entry->req = req;

		__entry->xid = be32_to_cpu(rep->rr_xid);

		__entry->credits = credits;

	),



	TP_printk("task:%u@%u xid=0x%08x, %u credits, rep=%p -> req=%p",

	TP_printk("task:%u@%u xid=0x%08x credits=%u",

		__entry->task_id, __entry->client_id, __entry->xid,

		__entry->credits, __entry->rep, __entry->req

		__entry->credits

	)

);



TRACE_EVENT(xprtrdma_defer_cmp,

DEFINE_REPLY_EVENT(vers);

DEFINE_REPLY_EVENT(rqst);

DEFINE_REPLY_EVENT(short);

DEFINE_REPLY_EVENT(hdr);



TRACE_EVENT(xprtrdma_err_vers,

	TP_PROTO(

		const struct rpcrdma_rep *rep

		const struct rpc_rqst *rqst,

		__be32 *min,

		__be32 *max

	),



	TP_ARGS(rep),

	TP_ARGS(rqst, min, max),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(const void *, rep)

		__field(u32, xid)

		__field(u32, min)

		__field(u32, max)

	),



	TP_fast_assign(

		__entry->task_id = rep->rr_rqst->rq_task->tk_pid;

		__entry->client_id = rep->rr_rqst->rq_task->tk_client->cl_clid;

		__entry->rep = rep;

		__entry->xid = be32_to_cpu(rep->rr_xid);

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client->cl_clid;

		__entry->xid = be32_to_cpu(rqst->rq_xid);

		__entry->min = be32_to_cpup(min);

		__entry->max = be32_to_cpup(max);

	),



	TP_printk("task:%u@%u xid=0x%08x rep=%p",

	TP_printk("task:%u@%u xid=0x%08x versions=[%u, %u]",

		__entry->task_id, __entry->client_id, __entry->xid,

		__entry->rep

		__entry->min, __entry->max

	)

);



DEFINE_REPLY_EVENT(xprtrdma_reply_vers);

DEFINE_REPLY_EVENT(xprtrdma_reply_rqst);

DEFINE_REPLY_EVENT(xprtrdma_reply_short);

DEFINE_REPLY_EVENT(xprtrdma_reply_hdr);

TRACE_EVENT(xprtrdma_err_chunk,

	TP_PROTO(

		const struct rpc_rqst *rqst

	),



	TP_ARGS(rqst),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(u32, xid)

	),



	TP_fast_assign(

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client->cl_clid;

		__entry->xid = be32_to_cpu(rqst->rq_xid);

	),



	TP_printk("task:%u@%u xid=0x%08x",

		__entry->task_id, __entry->client_id, __entry->xid

	)

);



TRACE_EVENT(xprtrdma_err_unrecognized,

	TP_PROTO(

		const struct rpc_rqst *rqst,

		__be32 *procedure

	),



	TP_ARGS(rqst, procedure),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(u32, xid)

		__field(u32, procedure)

	),



	TP_fast_assign(

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client->cl_clid;

		__entry->procedure = be32_to_cpup(procedure);

	),



	TP_printk("task:%u@%u xid=0x%08x procedure=%u",

		__entry->task_id, __entry->client_id, __entry->xid,

		__entry->procedure

	)

);



TRACE_EVENT(xprtrdma_fixup,

	TP_PROTO(
@@ -1187,6 +1167,28 @@ TRACE_EVENT(xprtrdma_decode_seg, 
	)

);



TRACE_EVENT(xprtrdma_mrs_zap,

	TP_PROTO(

		const struct rpc_task *task

	),



	TP_ARGS(task),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

	),



	TP_fast_assign(

		__entry->task_id = task->tk_pid;

		__entry->client_id = task->tk_client->cl_clid;

	),



	TP_printk("task:%u@%u",

		__entry->task_id, __entry->client_id

	)

);



/**

 ** Callback events

 **/
@@ -1219,36 +1221,8 @@ TRACE_EVENT(xprtrdma_cb_setup, 
	)

);



DEFINE_CB_EVENT(xprtrdma_cb_call);

DEFINE_CB_EVENT(xprtrdma_cb_reply);



TRACE_EVENT(xprtrdma_leaked_rep,

	TP_PROTO(

		const struct rpc_rqst *rqst,

		const struct rpcrdma_rep *rep

	),



	TP_ARGS(rqst, rep),



	TP_STRUCT__entry(

		__field(unsigned int, task_id)

		__field(unsigned int, client_id)

		__field(u32, xid)

		__field(const void *, rep)

	),



	TP_fast_assign(

		__entry->task_id = rqst->rq_task->tk_pid;

		__entry->client_id = rqst->rq_task->tk_client->cl_clid;

		__entry->xid = be32_to_cpu(rqst->rq_xid);

		__entry->rep = rep;

	),



	TP_printk("task:%u@%u xid=0x%08x rep=%p",

		__entry->task_id, __entry->client_id, __entry->xid,

		__entry->rep

	)

);

DEFINE_CALLBACK_EVENT(call);

DEFINE_CALLBACK_EVENT(reply);



/**

 ** Server-side RPC/RDMA events

net/sunrpc/xprtrdma/backchannel.c

+3 −3
Original line number Diff line number Diff line 
// SPDX-License-Identifier: GPL-2.0

/*

 * Copyright (c) 2015 Oracle.  All rights reserved.

 * Copyright (c) 2015-2020, Oracle and/or its affiliates.

 *

 * Support for backward direction RPCs on RPC/RDMA.

 */
@@ -82,7 +82,7 @@ static int rpcrdma_bc_marshal_reply(struct rpc_rqst *rqst) 
				      &rqst->rq_snd_buf, rpcrdma_noch_pullup))

		return -EIO;



	trace_xprtrdma_cb_reply(rqst);

	trace_xprtrdma_cb_reply(r_xprt, rqst);

	return 0;

}
@@ -260,7 +260,7 @@ void rpcrdma_bc_receive_call(struct rpcrdma_xprt *r_xprt, 
	 */

	req = rpcr_to_rdmar(rqst);

	req->rl_reply = rep;

	trace_xprtrdma_cb_call(rqst);

	trace_xprtrdma_cb_call(r_xprt, rqst);



	/* Queue rqst for ULP's callback service */

	bc_serv = xprt->bc_serv;

net/sunrpc/xprtrdma/frwr_ops.c

+55 −26
Original line number Diff line number Diff line
@@ -65,18 +65,23 @@ void frwr_release_mr(struct rpcrdma_mr *mr) 
	kfree(mr);

}



static void frwr_mr_unmap(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr *mr)

{

	if (mr->mr_device) {

		trace_xprtrdma_mr_unmap(mr);

		ib_dma_unmap_sg(mr->mr_device, mr->mr_sg, mr->mr_nents,

				mr->mr_dir);

		mr->mr_device = NULL;

	}

}



static void frwr_mr_recycle(struct rpcrdma_mr *mr)

{

	struct rpcrdma_xprt *r_xprt = mr->mr_xprt;



	trace_xprtrdma_mr_recycle(mr);



	if (mr->mr_dir != DMA_NONE) {

		trace_xprtrdma_mr_unmap(mr);

		ib_dma_unmap_sg(r_xprt->rx_ep->re_id->device,

				mr->mr_sg, mr->mr_nents, mr->mr_dir);

		mr->mr_dir = DMA_NONE;

	}

	frwr_mr_unmap(r_xprt, mr);



	spin_lock(&r_xprt->rx_buf.rb_lock);

	list_del(&mr->mr_all);
@@ -86,6 +91,16 @@ static void frwr_mr_recycle(struct rpcrdma_mr *mr) 
	frwr_release_mr(mr);

}



static void frwr_mr_put(struct rpcrdma_mr *mr)

{

	frwr_mr_unmap(mr->mr_xprt, mr);



	/* The MR is returned to the req's MR free list instead

	 * of to the xprt's MR free list. No spinlock is needed.

	 */

	rpcrdma_mr_push(mr, &mr->mr_req->rl_free_mrs);

}



/* frwr_reset - Place MRs back on the free list

 * @req: request to reset

 *
@@ -101,7 +116,7 @@ void frwr_reset(struct rpcrdma_req *req) 
	struct rpcrdma_mr *mr;



	while ((mr = rpcrdma_mr_pop(&req->rl_registered)))

		rpcrdma_mr_put(mr);

		frwr_mr_put(mr);

}



/**
@@ -130,7 +145,7 @@ int frwr_mr_init(struct rpcrdma_xprt *r_xprt, struct rpcrdma_mr *mr) 


	mr->mr_xprt = r_xprt;

	mr->frwr.fr_mr = frmr;

	mr->mr_dir = DMA_NONE;

	mr->mr_device = NULL;

	INIT_LIST_HEAD(&mr->mr_list);

	init_completion(&mr->frwr.fr_linv_done);
@@ -315,6 +330,7 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt, 
				  mr->mr_dir);

	if (!dma_nents)

		goto out_dmamap_err;

	mr->mr_device = ep->re_id->device;



	ibmr = mr->frwr.fr_mr;

	n = ib_map_mr_sg(ibmr, mr->mr_sg, dma_nents, NULL, PAGE_SIZE);
@@ -341,7 +357,6 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt, 
	return seg;



out_dmamap_err:

	mr->mr_dir = DMA_NONE;

	trace_xprtrdma_frwr_sgerr(mr, i);

	return ERR_PTR(-EIO);
@@ -363,12 +378,21 @@ static void frwr_wc_fastreg(struct ib_cq *cq, struct ib_wc *wc) 
		container_of(cqe, struct rpcrdma_frwr, fr_cqe);



	/* WARNING: Only wr_cqe and status are reliable at this point */

	trace_xprtrdma_wc_fastreg(wc, frwr);

	trace_xprtrdma_wc_fastreg(wc, &frwr->fr_cid);

	/* The MR will get recycled when the associated req is retransmitted */



	rpcrdma_flush_disconnect(cq->cq_context, wc);

}



static void frwr_cid_init(struct rpcrdma_ep *ep,

			  struct rpcrdma_frwr *frwr)

{

	struct rpc_rdma_cid *cid = &frwr->fr_cid;



	cid->ci_queue_id = ep->re_attr.send_cq->res.id;

	cid->ci_completion_id = frwr->fr_mr->res.id;

}



/**

 * frwr_send - post Send WRs containing the RPC Call message

 * @r_xprt: controlling transport instance
@@ -385,6 +409,7 @@ static void frwr_wc_fastreg(struct ib_cq *cq, struct ib_wc *wc) 
 */

int frwr_send(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)

{

	struct rpcrdma_ep *ep = r_xprt->rx_ep;

	struct ib_send_wr *post_wr;

	struct rpcrdma_mr *mr;
@@ -395,6 +420,7 @@ int frwr_send(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 
		frwr = &mr->frwr;



		frwr->fr_cqe.done = frwr_wc_fastreg;

		frwr_cid_init(ep, frwr);

		frwr->fr_regwr.wr.next = post_wr;

		frwr->fr_regwr.wr.wr_cqe = &frwr->fr_cqe;

		frwr->fr_regwr.wr.num_sge = 0;
@@ -404,7 +430,7 @@ int frwr_send(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 
		post_wr = &frwr->fr_regwr.wr;

	}



	return ib_post_send(r_xprt->rx_ep->re_id->qp, post_wr, NULL);

	return ib_post_send(ep->re_id->qp, post_wr, NULL);

}



/**
@@ -420,18 +446,17 @@ void frwr_reminv(struct rpcrdma_rep *rep, struct list_head *mrs) 
	list_for_each_entry(mr, mrs, mr_list)

		if (mr->mr_handle == rep->rr_inv_rkey) {

			list_del_init(&mr->mr_list);

			trace_xprtrdma_mr_reminv(mr);

			rpcrdma_mr_put(mr);

			frwr_mr_put(mr);

			break;	/* only one invalidated MR per RPC */

		}

}



static void __frwr_release_mr(struct ib_wc *wc, struct rpcrdma_mr *mr)

static void frwr_mr_done(struct ib_wc *wc, struct rpcrdma_mr *mr)

{

	if (wc->status != IB_WC_SUCCESS)

		frwr_mr_recycle(mr);

	else

		rpcrdma_mr_put(mr);

		frwr_mr_put(mr);

}



/**
@@ -448,8 +473,8 @@ static void frwr_wc_localinv(struct ib_cq *cq, struct ib_wc *wc) 
	struct rpcrdma_mr *mr = container_of(frwr, struct rpcrdma_mr, frwr);



	/* WARNING: Only wr_cqe and status are reliable at this point */

	trace_xprtrdma_wc_li(wc, frwr);

	__frwr_release_mr(wc, mr);

	trace_xprtrdma_wc_li(wc, &frwr->fr_cid);

	frwr_mr_done(wc, mr);



	rpcrdma_flush_disconnect(cq->cq_context, wc);

}
@@ -469,8 +494,8 @@ static void frwr_wc_localinv_wake(struct ib_cq *cq, struct ib_wc *wc) 
	struct rpcrdma_mr *mr = container_of(frwr, struct rpcrdma_mr, frwr);



	/* WARNING: Only wr_cqe and status are reliable at this point */

	trace_xprtrdma_wc_li_wake(wc, frwr);

	__frwr_release_mr(wc, mr);

	trace_xprtrdma_wc_li_wake(wc, &frwr->fr_cid);

	frwr_mr_done(wc, mr);

	complete(&frwr->fr_linv_done);



	rpcrdma_flush_disconnect(cq->cq_context, wc);
@@ -490,6 +515,7 @@ static void frwr_wc_localinv_wake(struct ib_cq *cq, struct ib_wc *wc) 
void frwr_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)

{

	struct ib_send_wr *first, **prev, *last;

	struct rpcrdma_ep *ep = r_xprt->rx_ep;

	const struct ib_send_wr *bad_wr;

	struct rpcrdma_frwr *frwr;

	struct rpcrdma_mr *mr;
@@ -509,6 +535,7 @@ void frwr_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 


		frwr = &mr->frwr;

		frwr->fr_cqe.done = frwr_wc_localinv;

		frwr_cid_init(ep, frwr);

		last = &frwr->fr_invwr;

		last->next = NULL;

		last->wr_cqe = &frwr->fr_cqe;
@@ -534,7 +561,7 @@ void frwr_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 
	 * unless re_id->qp is a valid pointer.

	 */

	bad_wr = NULL;

	rc = ib_post_send(r_xprt->rx_ep->re_id->qp, first, &bad_wr);

	rc = ib_post_send(ep->re_id->qp, first, &bad_wr);



	/* The final LOCAL_INV WR in the chain is supposed to

	 * do the wake. If it was never posted, the wake will
@@ -547,7 +574,7 @@ void frwr_unmap_sync(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 


	/* Recycle MRs in the LOCAL_INV chain that did not get posted.

	 */

	trace_xprtrdma_post_linv(req, rc);

	trace_xprtrdma_post_linv_err(req, rc);

	while (bad_wr) {

		frwr = container_of(bad_wr, struct rpcrdma_frwr,

				    fr_invwr);
@@ -574,10 +601,10 @@ static void frwr_wc_localinv_done(struct ib_cq *cq, struct ib_wc *wc) 
	struct rpcrdma_rep *rep = mr->mr_req->rl_reply;



	/* WARNING: Only wr_cqe and status are reliable at this point */

	trace_xprtrdma_wc_li_done(wc, frwr);

	__frwr_release_mr(wc, mr);

	trace_xprtrdma_wc_li_done(wc, &frwr->fr_cid);

	frwr_mr_done(wc, mr);



	/* Ensure @rep is generated before __frwr_release_mr */

	/* Ensure @rep is generated before frwr_mr_done */

	smp_rmb();

	rpcrdma_complete_rqst(rep);
@@ -597,6 +624,7 @@ static void frwr_wc_localinv_done(struct ib_cq *cq, struct ib_wc *wc) 
void frwr_unmap_async(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req)

{

	struct ib_send_wr *first, *last, **prev;

	struct rpcrdma_ep *ep = r_xprt->rx_ep;

	const struct ib_send_wr *bad_wr;

	struct rpcrdma_frwr *frwr;

	struct rpcrdma_mr *mr;
@@ -614,6 +642,7 @@ void frwr_unmap_async(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 


		frwr = &mr->frwr;

		frwr->fr_cqe.done = frwr_wc_localinv;

		frwr_cid_init(ep, frwr);

		last = &frwr->fr_invwr;

		last->next = NULL;

		last->wr_cqe = &frwr->fr_cqe;
@@ -639,13 +668,13 @@ void frwr_unmap_async(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req) 
	 * unless re_id->qp is a valid pointer.

	 */

	bad_wr = NULL;

	rc = ib_post_send(r_xprt->rx_ep->re_id->qp, first, &bad_wr);

	rc = ib_post_send(ep->re_id->qp, first, &bad_wr);

	if (!rc)

		return;



	/* Recycle MRs in the LOCAL_INV chain that did not get posted.

	 */

	trace_xprtrdma_post_linv(req, rc);

	trace_xprtrdma_post_linv_err(req, rc);

	while (bad_wr) {

		frwr = container_of(bad_wr, struct rpcrdma_frwr, fr_invwr);

		mr = container_of(frwr, struct rpcrdma_mr, frwr);

net/sunrpc/xprtrdma/rpc_rdma.c

+11 −21

File changed.

Preview size limit exceeded, changes collapsed.

net/sunrpc/xprtrdma/transport.c

+4 −3
Original line number Diff line number Diff line
@@ -599,11 +599,12 @@ static void 
xprt_rdma_free(struct rpc_task *task)

{

	struct rpc_rqst *rqst = task->tk_rqstp;

	struct rpcrdma_xprt *r_xprt = rpcx_to_rdmax(rqst->rq_xprt);

	struct rpcrdma_req *req = rpcr_to_rdmar(rqst);



	if (!list_empty(&req->rl_registered))

		frwr_unmap_sync(r_xprt, req);

	if (unlikely(!list_empty(&req->rl_registered))) {

		trace_xprtrdma_mrs_zap(task);

		frwr_unmap_sync(rpcx_to_rdmax(rqst->rq_xprt), req);

	}



	/* XXX: If the RPC is completing because of a signal and

	 * not because a reply was received, we ought to ensure