X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fptlrpc%2Fniobuf.c;h=4720592e5ebf4477880615f5a2b3f2f2c4638728;hb=917655fc2938b90a9c246dd2d58408c42aa1658d;hp=889216cb3ab43e6c2f1c07c93d6ed3c6344bbb2d;hpb=32c3775bab8902e533fd153a357b46da12076933;p=fs%2Flustre-release.git diff --git a/lustre/ptlrpc/niobuf.c b/lustre/ptlrpc/niobuf.c index 889216c..4720592 100644 --- a/lustre/ptlrpc/niobuf.c +++ b/lustre/ptlrpc/niobuf.c @@ -23,7 +23,7 @@ * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2012, 2016, Intel Corporation. + * Copyright (c) 2012, 2017, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -37,6 +37,7 @@ #include #include #include "ptlrpc_internal.h" +#include /* for CFS_FAIL_PTLRPC_OST_BULK_CB2 */ /** * Helper function. Sends \a len bytes from \a base at offset \a offset @@ -216,13 +217,10 @@ int ptlrpc_start_bulk_transfer(struct ptlrpc_bulk_desc *desc) break; } - /* LU-6441: last md is not sent and desc->bd_md_count == 1 */ - if (OBD_FAIL_CHECK_ORSET(OBD_FAIL_PTLRPC_CLIENT_BULK_CB3, - CFS_FAIL_ONCE) && - total_md > 1 && posted_md == total_md - 1) { - posted_md++; - continue; - } + /* sanity.sh 224c: lets skip last md */ + if (posted_md == desc->bd_md_max_brw - 1) + OBD_FAIL_CHECK_RESET(OBD_FAIL_PTLRPC_CLIENT_BULK_CB3, + CFS_FAIL_PTLRPC_OST_BULK_CB2); /* Network is about to get at the memory */ if (ptlrpc_is_bulk_put_source(desc->bd_type)) @@ -231,7 +229,7 @@ int ptlrpc_start_bulk_transfer(struct ptlrpc_bulk_desc *desc) desc->bd_portal, mbits, 0, 0); else rc = LNetGet(self_nid, desc->bd_mds[posted_md], - peer_id, desc->bd_portal, mbits, 0); + peer_id, desc->bd_portal, mbits, 0, false); posted_md++; if (rc != 0) { @@ -336,8 +334,11 @@ int ptlrpc_register_bulk(struct ptlrpc_request *req) /* cleanup the state of the bulk for it will be reused */ if (req->rq_resend || req->rq_send_state == LUSTRE_IMP_REPLAY) desc->bd_nob_transferred = 0; - else - LASSERT(desc->bd_nob_transferred == 0); + else if (desc->bd_nob_transferred != 0) + /* If the network failed after an RPC was sent, this condition + * could happen. Rather than assert (was here before), return + * an EIO error. */ + RETURN(-EIO); desc->bd_failure = 0; @@ -372,8 +373,13 @@ int ptlrpc_register_bulk(struct ptlrpc_request *req) LNET_MD_OP_GET : LNET_MD_OP_PUT); ptlrpc_fill_bulk_md(&md, desc, posted_md); - rc = LNetMEAttach(desc->bd_portal, peer, mbits, 0, + if (posted_md > 0 && posted_md + 1 == total_md && + OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_BULK_ATTACH)) { + rc = -ENOMEM; + } else { + rc = LNetMEAttach(desc->bd_portal, peer, mbits, 0, LNET_UNLINK, LNET_INS_AFTER, &me_h); + } if (rc != 0) { CERROR("%s: LNetMEAttach failed x%llu/%d: rc = %d\n", desc->bd_import->imp_obd->obd_name, mbits, @@ -402,6 +408,7 @@ int ptlrpc_register_bulk(struct ptlrpc_request *req) LASSERT(desc->bd_md_count >= 0); mdunlink_iterate_helper(desc->bd_mds, desc->bd_md_max_brw); req->rq_status = -ENOMEM; + desc->bd_registered = 0; RETURN(-ENOMEM); } @@ -440,7 +447,7 @@ int ptlrpc_unregister_bulk(struct ptlrpc_request *req, int async) /* Let's setup deadline for reply unlink. */ if (OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_LONG_BULK_UNLINK) && async && req->rq_bulk_deadline == 0 && cfs_fail_val == 0) - req->rq_bulk_deadline = cfs_time_current_sec() + LONG_UNLINK; + req->rq_bulk_deadline = ktime_get_real_seconds() + LONG_UNLINK; if (ptlrpc_client_bulk_active(req) == 0) /* completed or */ RETURN(1); /* never registered */ @@ -468,28 +475,30 @@ int ptlrpc_unregister_bulk(struct ptlrpc_request *req, int async) wait_queue_head_t *wq = (req->rq_set != NULL) ? &req->rq_set->set_waitq : &req->rq_reply_waitq; - /* Network access will complete in finite time but the HUGE - * timeout lets us CWARN for visibility of sluggish NALs */ - lwi = LWI_TIMEOUT_INTERVAL(cfs_time_seconds(LONG_UNLINK), - cfs_time_seconds(1), NULL, NULL); - rc = l_wait_event(*wq, !ptlrpc_client_bulk_active(req), &lwi); - if (rc == 0) { - ptlrpc_rqphase_move(req, req->rq_next_phase); - RETURN(1); - } + /* + * Network access will complete in finite time but the HUGE + * timeout lets us CWARN for visibility of sluggish NALs. + */ + lwi = LWI_TIMEOUT_INTERVAL(cfs_time_seconds(LONG_UNLINK), + cfs_time_seconds(1), NULL, NULL); + rc = l_wait_event(*wq, !ptlrpc_client_bulk_active(req), &lwi); + if (rc == 0) { + ptlrpc_rqphase_move(req, req->rq_next_phase); + RETURN(1); + } - LASSERT(rc == -ETIMEDOUT); - DEBUG_REQ(D_WARNING, req, "Unexpectedly long timeout: desc %p", - desc); - } - RETURN(0); + LASSERT(rc == -ETIMEDOUT); + DEBUG_REQ(D_WARNING, req, "Unexpectedly long timeout: desc %p", + desc); + } + RETURN(0); } static void ptlrpc_at_set_reply(struct ptlrpc_request *req, int flags) { struct ptlrpc_service_part *svcpt = req->rq_rqbd->rqbd_svcpt; struct ptlrpc_service *svc = svcpt->scp_service; - int service_time = max_t(int, cfs_time_current_sec() - + int service_time = max_t(int, ktime_get_real_seconds() - req->rq_arrival_time.tv_sec, 1); if (!(flags & PTLRPC_REPLY_EARLY) && @@ -586,16 +595,6 @@ int ptlrpc_send_reply(struct ptlrpc_request *req, int flags) req->rq_export->exp_obd->obd_minor); } - /* In order to keep interoprability with the client (< 2.3) which - * doesn't have pb_jobid in ptlrpc_body, We have to shrink the - * ptlrpc_body in reply buffer to ptlrpc_body_v2, otherwise, the - * reply buffer on client will be overflow. - * - * XXX Remove this whenver we drop the interoprability with such client. - */ - req->rq_replen = lustre_shrink_msg(req->rq_repmsg, 0, - sizeof(struct ptlrpc_body_v2), 1); - if (req->rq_type != PTL_RPC_MSG_ERR) req->rq_type = PTL_RPC_MSG_REPLY; @@ -624,7 +623,7 @@ int ptlrpc_send_reply(struct ptlrpc_request *req, int flags) if (unlikely(rc)) goto out; - req->rq_sent = cfs_time_current_sec(); + req->rq_sent = ktime_get_real_seconds(); rc = ptl_send_buf(&rs->rs_md_h, rs->rs_repbuf, rs->rs_repdata_len, (rs->rs_difficult && !rs->rs_no_ack) ? @@ -747,8 +746,8 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) spin_unlock(&imp->imp_lock); lustre_msg_set_last_xid(request->rq_reqmsg, min_xid); - DEBUG_REQ(D_RPCTRACE, request, "Allocating new xid for " - "resend on EINPROGRESS"); + DEBUG_REQ(D_RPCTRACE, request, + "Allocating new XID for resend on EINPROGRESS"); } if (request->rq_bulk != NULL) { @@ -758,9 +757,9 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) if (list_empty(&request->rq_unreplied_list) || request->rq_xid <= imp->imp_known_replied_xid) { - DEBUG_REQ(D_ERROR, request, "xid: %llu, replied: %llu, " - "list_empty:%d\n", request->rq_xid, - imp->imp_known_replied_xid, + DEBUG_REQ(D_ERROR, request, + "xid=%llu, replied=%llu, list_empty=%d", + request->rq_xid, imp->imp_known_replied_xid, list_empty(&request->rq_unreplied_list)); LBUG(); } @@ -777,8 +776,8 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) if (request->rq_resend_cb != NULL) request->rq_resend_cb(request, &request->rq_async_args); } - if (request->rq_memalloc) - mpflag = cfs_memory_pressure_get_and_set(); + if (request->rq_memalloc) + mpflag = cfs_memory_pressure_get_and_set(); rc = sptlrpc_cli_wrap_request(request); if (rc) @@ -788,7 +787,7 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) if (request->rq_bulk != NULL) { rc = ptlrpc_register_bulk (request); if (rc != 0) - GOTO(out, rc); + GOTO(cleanup_bulk, rc); /* * All the mds in the request will have the same cpt * encoded in the cookie. So we can just get the first @@ -810,13 +809,13 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) spin_lock(&request->rq_lock); request->rq_err = 1; spin_unlock(&request->rq_lock); - request->rq_status = rc; - GOTO(cleanup_bulk, rc); - } - } else { - request->rq_repdata = NULL; - request->rq_repmsg = NULL; - } + request->rq_status = rc; + GOTO(cleanup_bulk, rc); + } + } else { + request->rq_repdata = NULL; + request->rq_repmsg = NULL; + } rc = LNetMEAttach(request->rq_reply_portal,/*XXX FIXME bug 249*/ connection->c_peer, request->rq_xid, 0, @@ -883,8 +882,8 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) OBD_FAIL_TIMEOUT(OBD_FAIL_PTLRPC_DELAY_SEND, request->rq_timeout + 5); - do_gettimeofday(&request->rq_sent_tv); - request->rq_sent = cfs_time_current_sec(); + request->rq_sent_ns = ktime_get_real(); + request->rq_sent = ktime_get_real_seconds(); /* We give the server rq_timeout secs to process the req, and add the network latency for our local timeout. */ request->rq_deadline = request->rq_sent + request->rq_timeout + @@ -892,7 +891,7 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) ptlrpc_pinger_sending_on_import(imp); - DEBUG_REQ(D_INFO, request, "send flg=%x", + DEBUG_REQ(D_INFO, request, "send flags=%x", lustre_msg_get_flags(request->rq_reqmsg)); rc = ptl_send_buf(&request->rq_req_md_h, request->rq_reqbuf, request->rq_reqdata_len, @@ -909,23 +908,25 @@ int ptl_send_rpc(struct ptlrpc_request *request, int noreply) GOTO(out, rc); cleanup_me: - /* MEUnlink is safe; the PUT didn't even get off the ground, and - * nobody apart from the PUT's target has the right nid+XID to - * access the reply buffer. */ - rc2 = LNetMEUnlink(reply_me_h); - LASSERT (rc2 == 0); - /* UNLINKED callback called synchronously */ - LASSERT(!request->rq_receiving_reply); + /* MEUnlink is safe; the PUT didn't even get off the ground, and + * nobody apart from the PUT's target has the right nid+XID to + * access the reply buffer. */ + rc2 = LNetMEUnlink(reply_me_h); + LASSERT (rc2 == 0); + /* UNLINKED callback called synchronously */ + LASSERT(!request->rq_receiving_reply); cleanup_bulk: - /* We do sync unlink here as there was no real transfer here so - * the chance to have long unlink to sluggish net is smaller here. */ + /* We do sync unlink here as there was no real transfer here so + * the chance to have long unlink to sluggish net is smaller here. */ ptlrpc_unregister_bulk(request, 0); + if (request->rq_bulk != NULL) + request->rq_bulk->bd_registered = 0; out: if (rc == -ENOMEM) { /* set rq_sent so that this request is treated * as a delayed send in the upper layers */ - request->rq_sent = cfs_time_current_sec(); + request->rq_sent = ktime_get_real_seconds(); } if (request->rq_memalloc) @@ -941,7 +942,10 @@ EXPORT_SYMBOL(ptl_send_rpc); int ptlrpc_register_rqbd(struct ptlrpc_request_buffer_desc *rqbd) { struct ptlrpc_service *service = rqbd->rqbd_svcpt->scp_service; - static struct lnet_process_id match_id = {LNET_NID_ANY, LNET_PID_ANY}; + static struct lnet_process_id match_id = { + .nid = LNET_NID_ANY, + .pid = LNET_PID_ANY + }; int rc; struct lnet_md md; struct lnet_handle_me me_h;