Whamcloud - gitweb
LU-7760 ptlrpc: remove incorrect pid printing
[fs/lustre-release.git] / lustre / ptlrpc / client.c
index 13a3ece..bac3301 100644 (file)
@@ -357,7 +357,7 @@ void ptlrpc_at_adj_net_latency(struct ptlrpc_request *req,
 {
         unsigned int nl, oldnl;
         struct imp_at *at;
-        time_t now = cfs_time_current_sec();
+       time64_t now = ktime_get_real_seconds();
 
         LASSERT(req->rq_import);
 
@@ -371,9 +371,8 @@ void ptlrpc_at_adj_net_latency(struct ptlrpc_request *req,
                 */
                CDEBUG((lustre_msg_get_flags(req->rq_reqmsg) & MSG_RESENT) ?
                       D_ADAPTTO : D_WARNING,
-                      "Reported service time %u > total measured time "
-                      CFS_DURATION_T"\n", service_time,
-                      cfs_time_sub(now, req->rq_sent));
+                      "Reported service time %u > total measured time %lld\n",
+                      service_time, now - req->rq_sent);
                return;
        }
 
@@ -419,11 +418,11 @@ static int unpack_reply(struct ptlrpc_request *req)
 static int ptlrpc_at_recv_early_reply(struct ptlrpc_request *req)
 __must_hold(&req->rq_lock)
 {
-        struct ptlrpc_request *early_req;
-        time_t                 olddl;
-        int                    rc;
-        ENTRY;
+       struct ptlrpc_request *early_req;
+       time64_t olddl;
+       int rc;
 
+       ENTRY;
         req->rq_early = 0;
        spin_unlock(&req->rq_lock);
 
@@ -464,10 +463,10 @@ __must_hold(&req->rq_lock)
                           ptlrpc_at_get_net_latency(req);
 
        DEBUG_REQ(D_ADAPTTO, req,
-                 "Early reply #%d, new deadline in "CFS_DURATION_T"s "
-                 "("CFS_DURATION_T"s)", req->rq_early_count,
-                 cfs_time_sub(req->rq_deadline, cfs_time_current_sec()),
-                 cfs_time_sub(req->rq_deadline, olddl));
+                 "Early reply #%d, new deadline in %llds (%llds)",
+                 req->rq_early_count,
+                 req->rq_deadline - ktime_get_real_seconds(),
+                 req->rq_deadline - olddl);
 
        RETURN(rc);
 }
@@ -744,7 +743,7 @@ int ptlrpc_request_bufs_pack(struct ptlrpc_request *request,
 
        /* Let's setup deadline for req/reply/bulk unlink for opcode. */
        if (cfs_fail_val == opcode) {
-               time_t *fail_t = NULL, *fail2_t = NULL;
+               time64_t *fail_t = NULL, *fail2_t = NULL;
 
                if (CFS_FAIL_CHECK(OBD_FAIL_PTLRPC_LONG_BULK_UNLINK))
                        fail_t = &request->rq_bulk_deadline;
@@ -758,16 +757,17 @@ int ptlrpc_request_bufs_pack(struct ptlrpc_request *request,
                }
 
                if (fail_t) {
-                       *fail_t = cfs_time_current_sec() + LONG_UNLINK;
+                       *fail_t = ktime_get_real_seconds() + LONG_UNLINK;
 
                        if (fail2_t)
-                               *fail2_t = cfs_time_current_sec() + LONG_UNLINK;
+                               *fail2_t = ktime_get_real_seconds() +
+                                          LONG_UNLINK;
 
-                       /* The RPC is infected, let the test to change the
-                        * fail_loc */
-                       set_current_state(TASK_UNINTERRUPTIBLE);
-                       schedule_timeout(cfs_time_seconds(2));
-                       set_current_state(TASK_RUNNING);
+                       /*
+                        * The RPC is infected, let the test to change the
+                        * fail_loc
+                        */
+                       msleep(4 * MSEC_PER_SEC);
                }
        }
 
@@ -1327,14 +1327,14 @@ __u64 ptlrpc_known_replied_xid(struct obd_import *imp)
  */
 static int after_reply(struct ptlrpc_request *req)
 {
-        struct obd_import *imp = req->rq_import;
-        struct obd_device *obd = req->rq_import->imp_obd;
-        int rc;
-        struct timeval work_start;
-       __u64 committed;
-        long timediff;
-        ENTRY;
+       struct obd_import *imp = req->rq_import;
+       struct obd_device *obd = req->rq_import->imp_obd;
+       ktime_t work_start;
+       u64 committed;
+       s64 timediff;
+       int rc;
 
+       ENTRY;
         LASSERT(obd != NULL);
         /* repbuf must be unlinked */
        LASSERT(!req->rq_receiving_reply && req->rq_reply_unlinked);
@@ -1360,8 +1360,8 @@ static int after_reply(struct ptlrpc_request *req)
                 RETURN(0);
         }
 
-       do_gettimeofday(&work_start);
-       timediff = cfs_timeval_sub(&work_start, &req->rq_sent_tv, NULL);
+       work_start = ktime_get_real();
+       timediff = ktime_us_delta(req->rq_sent_ns, work_start);
 
         /*
          * NB Until this point, the whole of the incoming message,
@@ -1386,7 +1386,7 @@ static int after_reply(struct ptlrpc_request *req)
        /* retry indefinitely on EINPROGRESS */
        if (lustre_msg_get_status(req->rq_repmsg) == -EINPROGRESS &&
            ptlrpc_no_resend(req) == 0 && !req->rq_no_retry_einprogress) {
-               time_t  now = cfs_time_current_sec();
+               time64_t now = ktime_get_real_seconds();
 
                DEBUG_REQ(D_RPCTRACE, req, "Resending request on EINPROGRESS");
                spin_lock(&req->rq_lock);
@@ -1537,7 +1537,7 @@ static int ptlrpc_send_new_req(struct ptlrpc_request *req)
                    pool_is_at_full_capacity())
                        RETURN(-ENOMEM);
 
-        if (req->rq_sent && (req->rq_sent > cfs_time_current_sec()) &&
+       if (req->rq_sent && (req->rq_sent > ktime_get_real_seconds()) &&
             (!req->rq_generation_set ||
              req->rq_import_generation == imp->imp_generation))
                 RETURN (0);
@@ -1557,8 +1557,7 @@ static int ptlrpc_send_new_req(struct ptlrpc_request *req)
                req->rq_waiting = 1;
                spin_unlock(&req->rq_lock);
 
-               DEBUG_REQ(D_HA, req, "req from PID %d waiting for recovery: "
-                         "(%s != %s)", lustre_msg_get_status(req->rq_reqmsg),
+               DEBUG_REQ(D_HA, req, "req waiting for recovery: (%s != %s)",
                          ptlrpc_import_state_name(req->rq_send_state),
                          ptlrpc_import_state_name(imp->imp_state));
                LASSERT(list_empty(&req->rq_list));
@@ -1729,7 +1728,7 @@ int ptlrpc_check_set(const struct lu_env *env, struct ptlrpc_request_set *set)
 
                /* delayed resend - skip */
                if (req->rq_phase == RQ_PHASE_RPC && req->rq_resend &&
-                   req->rq_sent > cfs_time_current_sec())
+                   req->rq_sent > ktime_get_real_seconds())
                        continue;
 
                if (!(req->rq_phase == RQ_PHASE_RPC ||
@@ -2105,14 +2104,13 @@ int ptlrpc_expire_one_request(struct ptlrpc_request *req, int async_unlink)
        req->rq_timedout = 1;
        spin_unlock(&req->rq_lock);
 
-       DEBUG_REQ(D_WARNING, req, "Request sent has %s: [sent "CFS_DURATION_T
-                 "/real "CFS_DURATION_T"]",
+       DEBUG_REQ(D_WARNING, req, "Request sent has %s: [sent %lld/real %lld]",
                   req->rq_net_err ? "failed due to network error" :
                      ((req->rq_real_sent == 0 ||
-                       cfs_time_before(req->rq_real_sent, req->rq_sent) ||
-                       cfs_time_aftereq(req->rq_real_sent, req->rq_deadline)) ?
+                      req->rq_real_sent < req->rq_sent ||
+                      req->rq_real_sent >= req->rq_deadline) ?
                       "timed out for sent delay" : "timed out for slow reply"),
-                  req->rq_sent, req->rq_real_sent);
+                 (s64)req->rq_sent, (s64)req->rq_real_sent);
 
        if (imp != NULL && obd_debug_peer_on_timeout)
                LNetDebugPeer(imp->imp_connection->c_peer);
@@ -2168,11 +2166,11 @@ int ptlrpc_expire_one_request(struct ptlrpc_request *req, int async_unlink)
  */
 int ptlrpc_expired_set(void *data)
 {
-       struct ptlrpc_request_set       *set = data;
-       struct list_head                *tmp;
-       time_t                          now = cfs_time_current_sec();
-       ENTRY;
+       struct ptlrpc_request_set *set = data;
+       struct list_head *tmp;
+       time64_t now = ktime_get_real_seconds();
 
+       ENTRY;
        LASSERT(set != NULL);
 
        /*
@@ -2254,13 +2252,13 @@ static void ptlrpc_interrupted_set(void *data)
  */
 int ptlrpc_set_next_timeout(struct ptlrpc_request_set *set)
 {
-       struct list_head        *tmp;
-       time_t                   now = cfs_time_current_sec();
-       int                      timeout = 0;
-       struct ptlrpc_request   *req;
-       int                      deadline;
-       ENTRY;
+       struct list_head *tmp;
+       time64_t now = ktime_get_real_seconds();
+       int timeout = 0;
+       struct ptlrpc_request *req;
+       time64_t deadline;
 
+       ENTRY;
        list_for_each(tmp, &set->set_requests) {
                req = list_entry(tmp, struct ptlrpc_request, rq_set_chain);
 
@@ -2507,29 +2505,54 @@ void ptlrpc_req_finished_with_imp_lock(struct ptlrpc_request *request)
  * Drops one reference count for request \a request.
  * \a locked set indicates that caller holds import imp_lock.
  * Frees the request whe reference count reaches zero.
+ *
+ * \retval 1   the request is freed
+ * \retval 0   some others still hold references on the request
  */
 static int __ptlrpc_req_finished(struct ptlrpc_request *request, int locked)
 {
-        ENTRY;
-        if (request == NULL)
-                RETURN(1);
+       int count;
+       ENTRY;
 
-        if (request == LP_POISON ||
-            request->rq_reqmsg == LP_POISON) {
-                CERROR("dereferencing freed request (bug 575)\n");
-                LBUG();
-                RETURN(1);
-        }
+       if (!request)
+               RETURN(1);
 
-        DEBUG_REQ(D_INFO, request, "refcount now %u",
+       LASSERT(request != LP_POISON);
+       LASSERT(request->rq_reqmsg != LP_POISON);
+
+       DEBUG_REQ(D_INFO, request, "refcount now %u",
                  atomic_read(&request->rq_refcount) - 1);
 
-       if (atomic_dec_and_test(&request->rq_refcount)) {
-                __ptlrpc_free_req(request, locked);
-                RETURN(1);
-        }
+       spin_lock(&request->rq_lock);
+       count = atomic_dec_return(&request->rq_refcount);
+       LASSERTF(count >= 0, "Invalid ref count %d\n", count);
 
-        RETURN(0);
+       /* For open RPC, the client does not know the EA size (LOV, ACL, and
+        * so on) before replied, then the client has to reserve very large
+        * reply buffer. Such buffer will not be released until the RPC freed.
+        * Since The open RPC is replayable, we need to keep it in the replay
+        * list until close. If there are a lot of files opened concurrently,
+        * then the client may be OOM.
+        *
+        * If fact, it is unnecessary to keep reply buffer for open replay,
+        * related EAs have already been saved via mdc_save_lovea() before
+        * coming here. So it is safe to free the reply buffer some earlier
+        * before releasing the RPC to avoid client OOM. LU-9514 */
+       if (count == 1 && request->rq_early_free_repbuf && request->rq_repbuf) {
+               spin_lock(&request->rq_early_free_lock);
+               sptlrpc_cli_free_repbuf(request);
+               request->rq_repbuf = NULL;
+               request->rq_repbuf_len = 0;
+               request->rq_repdata = NULL;
+               request->rq_reqdata_len = 0;
+               spin_unlock(&request->rq_early_free_lock);
+       }
+       spin_unlock(&request->rq_lock);
+
+       if (!count)
+               __ptlrpc_free_req(request, locked);
+
+       RETURN(!count);
 }
 
 /**
@@ -2570,8 +2593,8 @@ static int ptlrpc_unregister_reply(struct ptlrpc_request *request, int async)
        /* Let's setup deadline for reply unlink. */
        if (OBD_FAIL_CHECK(OBD_FAIL_PTLRPC_LONG_REPL_UNLINK) &&
            async && request->rq_reply_deadline == 0 && cfs_fail_val == 0)
-               request->rq_reply_deadline =
-                       cfs_time_current_sec() + LONG_UNLINK;
+               request->rq_reply_deadline = ktime_get_real_seconds() +
+                                            LONG_UNLINK;
 
         /*
          * Nothing left to do.
@@ -3083,6 +3106,9 @@ int ptlrpc_replay_req(struct ptlrpc_request *req)
         DEBUG_REQ(D_HA, req, "REPLAY");
 
        atomic_inc(&req->rq_import->imp_replay_inflight);
+       spin_lock(&req->rq_lock);
+       req->rq_early_free_repbuf = 0;
+       spin_unlock(&req->rq_lock);
        ptlrpc_request_addref(req);     /* ptlrpcd needs a ref */
 
        ptlrpcd_add_req(req);
@@ -3196,7 +3222,7 @@ static spinlock_t ptlrpc_last_xid_lock;
 #define YEAR_2004 (1ULL << 30)
 void ptlrpc_init_xid(void)
 {
-       time_t now = cfs_time_current_sec();
+       time64_t now = ktime_get_real_seconds();
 
        spin_lock_init(&ptlrpc_last_xid_lock);
        if (now < YEAR_2004) {
@@ -3344,7 +3370,7 @@ static void ptlrpcd_add_work_req(struct ptlrpc_request *req)
 {
        /* re-initialize the req */
        req->rq_timeout         = obd_timeout;
-       req->rq_sent            = cfs_time_current_sec();
+       req->rq_sent            = ktime_get_real_seconds();
        req->rq_deadline        = req->rq_sent + req->rq_timeout;
        req->rq_phase           = RQ_PHASE_INTERPRET;
        req->rq_next_phase      = RQ_PHASE_COMPLETE;