From 05eb06d86685e7d9dac60e6bbb46d7f4c30b056e Mon Sep 17 00:00:00 2001 From: Chuck Lever Date: Wed, 19 Jun 2019 10:32:48 -0400 Subject: xprtrdma: Fix occasional transport deadlock Under high I/O workloads, I've noticed that an RPC/RDMA transport occasionally deadlocks (IOPS goes to zero, and doesn't recover). Diagnosis shows that the sendctx queue is empty, but when sendctxs are returned to the queue, the xprt_write_space wake-up never occurs. The wake-up logic in rpcrdma_sendctx_put_locked is racy. I noticed that both EMPTY_SCQ and XPRT_WRITE_SPACE are implemented via an atomic bit. Just one of those is sufficient. Removing EMPTY_SCQ in favor of the generic bit mechanism makes the deadlock un-reproducible. Without EMPTY_SCQ, rpcrdma_buffer::rb_flags is no longer used and is therefore removed. Unfortunately this patch does not apply cleanly to stable. If needed, someone will have to port it and test it. Fixes: 2fad659209d5 ("xprtrdma: Wait on empty sendctx queue") Signed-off-by: Chuck Lever Signed-off-by: Anna Schumaker --- net/sunrpc/xprtrdma/frwr_ops.c | 6 +++++- net/sunrpc/xprtrdma/rpc_rdma.c | 26 ++++++++++++-------------- net/sunrpc/xprtrdma/verbs.c | 11 +++-------- net/sunrpc/xprtrdma/xprt_rdma.h | 6 ------ 4 files changed, 20 insertions(+), 29 deletions(-) (limited to 'net/sunrpc') diff --git a/net/sunrpc/xprtrdma/frwr_ops.c b/net/sunrpc/xprtrdma/frwr_ops.c index 794ba4ca0994..ac47314fb751 100644 --- a/net/sunrpc/xprtrdma/frwr_ops.c +++ b/net/sunrpc/xprtrdma/frwr_ops.c @@ -391,7 +391,7 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt, rpcrdma_mr_recycle(mr); mr = rpcrdma_mr_get(r_xprt); if (!mr) - return ERR_PTR(-EAGAIN); + goto out_getmr_err; } while (mr->frwr.fr_state != FRWR_IS_INVALID); frwr = &mr->frwr; frwr->fr_state = FRWR_IS_VALID; @@ -448,6 +448,10 @@ struct rpcrdma_mr_seg *frwr_map(struct rpcrdma_xprt *r_xprt, *out = mr; return seg; +out_getmr_err: + xprt_wait_for_buffer_space(&r_xprt->rx_xprt); + return ERR_PTR(-EAGAIN); + out_dmamap_err: mr->mr_dir = DMA_NONE; trace_xprtrdma_frwr_sgerr(mr, i); diff --git a/net/sunrpc/xprtrdma/rpc_rdma.c b/net/sunrpc/xprtrdma/rpc_rdma.c index 97bfb804b6c6..59b214ba8813 100644 --- a/net/sunrpc/xprtrdma/rpc_rdma.c +++ b/net/sunrpc/xprtrdma/rpc_rdma.c @@ -699,22 +699,28 @@ rpcrdma_prepare_send_sges(struct rpcrdma_xprt *r_xprt, struct rpcrdma_req *req, u32 hdrlen, struct xdr_buf *xdr, enum rpcrdma_chunktype rtype) { + int ret; + + ret = -EAGAIN; req->rl_sendctx = rpcrdma_sendctx_get_locked(r_xprt); if (!req->rl_sendctx) - return -EAGAIN; + goto err; req->rl_sendctx->sc_wr.num_sge = 0; req->rl_sendctx->sc_unmap_count = 0; req->rl_sendctx->sc_req = req; __clear_bit(RPCRDMA_REQ_F_TX_RESOURCES, &req->rl_flags); + ret = -EIO; if (!rpcrdma_prepare_hdr_sge(r_xprt, req, hdrlen)) - return -EIO; - + goto err; if (rtype != rpcrdma_areadch) if (!rpcrdma_prepare_msg_sges(r_xprt, req, xdr, rtype)) - return -EIO; - + goto err; return 0; + +err: + trace_xprtrdma_prepsend_failed(&req->rl_slot, ret); + return ret; } /** @@ -877,15 +883,7 @@ rpcrdma_marshal_req(struct rpcrdma_xprt *r_xprt, struct rpc_rqst *rqst) out_err: trace_xprtrdma_marshal_failed(rqst, ret); - switch (ret) { - case -EAGAIN: - xprt_wait_for_buffer_space(rqst->rq_xprt); - break; - case -ENOBUFS: - break; - default: - r_xprt->rx_stats.failed_marshal_count++; - } + r_xprt->rx_stats.failed_marshal_count++; return ret; } diff --git a/net/sunrpc/xprtrdma/verbs.c b/net/sunrpc/xprtrdma/verbs.c index e71315e9dca2..0be5a36cacb6 100644 --- a/net/sunrpc/xprtrdma/verbs.c +++ b/net/sunrpc/xprtrdma/verbs.c @@ -901,7 +901,7 @@ out_emptyq: * completions recently. This is a sign the Send Queue is * backing up. Cause the caller to pause and try again. */ - set_bit(RPCRDMA_BUF_F_EMPTY_SCQ, &buf->rb_flags); + xprt_wait_for_buffer_space(&r_xprt->rx_xprt); r_xprt->rx_stats.empty_sendctx_q++; return NULL; } @@ -936,10 +936,7 @@ rpcrdma_sendctx_put_locked(struct rpcrdma_sendctx *sc) /* Paired with READ_ONCE */ smp_store_release(&buf->rb_sc_tail, next_tail); - if (test_and_clear_bit(RPCRDMA_BUF_F_EMPTY_SCQ, &buf->rb_flags)) { - smp_mb__after_atomic(); - xprt_write_space(&sc->sc_xprt->rx_xprt); - } + xprt_write_space(&sc->sc_xprt->rx_xprt); } static void @@ -977,8 +974,6 @@ rpcrdma_mrs_create(struct rpcrdma_xprt *r_xprt) r_xprt->rx_stats.mrs_allocated += count; spin_unlock(&buf->rb_mrlock); trace_xprtrdma_createmrs(r_xprt, count); - - xprt_write_space(&r_xprt->rx_xprt); } static void @@ -990,6 +985,7 @@ rpcrdma_mr_refresh_worker(struct work_struct *work) rx_buf); rpcrdma_mrs_create(r_xprt); + xprt_write_space(&r_xprt->rx_xprt); } /** @@ -1089,7 +1085,6 @@ int rpcrdma_buffer_create(struct rpcrdma_xprt *r_xprt) struct rpcrdma_buffer *buf = &r_xprt->rx_buf; int i, rc; - buf->rb_flags = 0; buf->rb_max_requests = r_xprt->rx_ep.rep_max_requests; buf->rb_bc_srv_max_requests = 0; spin_lock_init(&buf->rb_mrlock); diff --git a/net/sunrpc/xprtrdma/xprt_rdma.h b/net/sunrpc/xprtrdma/xprt_rdma.h index d1e0749bcbc4..2c6c5d8c3de1 100644 --- a/net/sunrpc/xprtrdma/xprt_rdma.h +++ b/net/sunrpc/xprtrdma/xprt_rdma.h @@ -391,7 +391,6 @@ struct rpcrdma_buffer { struct list_head rb_recv_bufs; struct list_head rb_allreqs; - unsigned long rb_flags; u32 rb_max_requests; u32 rb_credits; /* most recent credit grant */ @@ -402,11 +401,6 @@ struct rpcrdma_buffer { struct delayed_work rb_refresh_worker; }; -/* rb_flags */ -enum { - RPCRDMA_BUF_F_EMPTY_SCQ = 0, -}; - /* * Statistics for RPCRDMA */ -- cgit v1.2.3-70-g09d2