From 2bf122991c69b973ddb2a89e4ccdee4b7661ef8b Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Thu, 5 Jul 2018 18:26:32 +0200
Subject: [PATCH 1/7] Started to convert the RDMA_CM_EVENT loop to a separate
thread and added a new state to the node. This commit is still broken
---
include/villas/common.h | 3 +-
include/villas/nodes/infiniband.h | 1 -
lib/nodes/infiniband.c | 51 ++++++++++++++++++++-----------
3 files changed, 35 insertions(+), 20 deletions(-)
diff --git a/include/villas/common.h b/include/villas/common.h
index 64de9bb8c..a7093ac38 100644
--- a/include/villas/common.h
+++ b/include/villas/common.h
@@ -38,7 +38,8 @@ enum state {
STATE_OPENED = 4, /* alias for STATE_STARTED used by struct io */
STATE_STOPPED = 5,
STATE_UNLOADED = 5, /* alias for STATE_STARTED used by struct plugin */
- STATE_CLOSED = 5 /* alias for STATE_STARTED used by struct io */
+ STATE_CLOSED = 5, /* alias for STATE_STARTED used by struct io */
+ STATE_CONNECTED = 6
};
/** Callback to destroy list elements.
diff --git a/include/villas/nodes/infiniband.h b/include/villas/nodes/infiniband.h
index 0e0b3136c..4e7c056e4 100644
--- a/include/villas/nodes/infiniband.h
+++ b/include/villas/nodes/infiniband.h
@@ -55,7 +55,6 @@ struct infiniband {
struct context_s {
struct rdma_cm_id *listen_id;
struct rdma_cm_id *id;
- struct rdma_event_channel *ec;
struct ibv_pd *pd;
struct ibv_cq *recv_cq;
diff --git a/lib/nodes/infiniband.c b/lib/nodes/infiniband.c
index ba82bb1f7..ce5245986 100644
--- a/lib/nodes/infiniband.c
+++ b/lib/nodes/infiniband.c
@@ -32,7 +32,7 @@
#include <villas/memory.h>
#include <villas/memory/ib.h>
-#include <rdma/rdma_cma.h>
+static struct rdma_event_channel *rdma_event_channel;
int ib_cleanup(struct node *n)
{
@@ -59,7 +59,7 @@ int ib_cleanup(struct node *n)
debug(LOG_IB | 3, "Destroyed rdma_cm_id");
// Destroy event channel
- rdma_destroy_event_channel(ib->ctx.ec);
+ rdma_destroy_event_channel(rdma_event_channel);
debug(LOG_IB | 3, "Destroyed event channel");
return 0;
@@ -587,6 +587,27 @@ int ib_destroy(struct node *n)
return 0;
}
+void * ib_rdma_cm_event_thread(void *n)
+{
+ struct node *node = (struct node *) n;
+ //struct infiniband *ib = (struct infiniband *) node->_vd;
+ struct rdma_cm_event *event;
+
+ debug(LOG_IB | 1, "Started rdma_cm_event thread of node %s", node_name(node));
+
+ while (rdma_get_cm_event(rdma_event_channel, &event) == 0) {
+ struct rdma_cm_event event_copy;
+
+ memcpy(&event_copy, event, sizeof(*event));
+ rdma_ack_cm_event(event);
+
+ if (ib_event(n, &event_copy))
+ break;
+ }
+
+ return NULL;
+}
+
void * ib_disconnect_thread(void *n)
{
struct node *node = (struct node *) n;
@@ -595,7 +616,7 @@ void * ib_disconnect_thread(void *n)
debug(LOG_IB | 1, "Started disconnect thread of node %s", node_name(node));
- while (rdma_get_cm_event(ib->ctx.ec, &event) == 0) {
+ while (rdma_get_cm_event(rdma_event_channel, &event) == 0) {
if (event->event == RDMA_CM_EVENT_DISCONNECTED) {
debug(LOG_IB | 2, "Received RDMA_CM_EVENT_DISCONNECTED");
@@ -618,13 +639,13 @@ int ib_start(struct node *n)
debug(LOG_IB | 1, "Started ib_start");
// Create event channel
- ib->ctx.ec = rdma_create_event_channel();
- if (!ib->ctx.ec)
+ rdma_event_channel = rdma_create_event_channel();
+ if (!rdma_event_channel)
error("Failed to create event channel in node %s!", node_name(n));
debug(LOG_IB | 3, "Created event channel");
- ret = rdma_create_id(ib->ctx.ec, &ib->ctx.id, NULL, ib->conn.port_space);
+ ret = rdma_create_id(rdma_event_channel, &ib->ctx.id, NULL, ib->conn.port_space);
if (ret)
error("Failed to create rdma_cm_id of node %s: %s",
node_name(n), gai_strerror(ret));
@@ -667,22 +688,16 @@ int ib_start(struct node *n)
// sure the nodes are succesfully connected.
debug(LOG_IB | 1, "Starting to monitor events on rdma_cm_id");
- while (rdma_get_cm_event(ib->ctx.ec, &event) == 0) {
- struct rdma_cm_event event_copy;
-
- memcpy(&event_copy, event, sizeof(*event));
-
- rdma_ack_cm_event(event);
-
- if (ib_event(n, &event_copy))
- break;
- }
-
+ //Create thread to monitor rdma_cm_event channel
ret = pthread_create(&ib->conn.stop_thread, NULL, ib_disconnect_thread, n);
if (ret)
error("Failed to create thread to monitor disconnects in node %s: %s",
node_name(n), gai_strerror(ret));
+
+ //Wait until rdma_cm_event thread sets state to STARTED
+ while ( !(n->state == STATE_CONNECTED));
+
return 0;
}
@@ -708,7 +723,7 @@ int ib_stop(struct node *n)
else {
// Else, wait for event to occur
ib->conn.rdma_disconnect_called = 1;
- rdma_get_cm_event(ib->ctx.ec, &event);
+ rdma_get_cm_event(rdma_event_channel, &event);
rdma_ack_cm_event(event);
From e2061e58fc1deacd0a4437772fdc5c42c7a26f15 Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Sat, 7 Jul 2018 12:49:22 +0200
Subject: [PATCH 2/7] Events are now monitored in a separate thread. The
segmentation faults we saw earlier were caused because we exited ib_start
before we created a protection domain, which is used by memory_allocation
---
include/villas/nodes/infiniband.h | 5 +-
lib/nodes/infiniband.c | 378 +++++++++++++++---------------
2 files changed, 187 insertions(+), 196 deletions(-)
diff --git a/include/villas/nodes/infiniband.h b/include/villas/nodes/infiniband.h
index 4e7c056e4..449d8c88c 100644
--- a/include/villas/nodes/infiniband.h
+++ b/include/villas/nodes/infiniband.h
@@ -38,6 +38,7 @@
/* Function pointer typedefs */
typedef void (*ib_on_completion) (struct node*, struct ibv_wc*, int*);
typedef void * (*ib_poll_function) (void*);
+typedef void * (*ib_event_function) (void*);
/* Enums */
enum poll_mode_e {
@@ -55,6 +56,7 @@ struct infiniband {
struct context_s {
struct rdma_cm_id *listen_id;
struct rdma_cm_id *id;
+ struct rdma_event_channel *ec;
struct ibv_pd *pd;
struct ibv_cq *recv_cq;
@@ -87,8 +89,7 @@ struct infiniband {
struct r_addr_key_s *r_addr_key;
- pthread_t stop_thread;
- int rdma_disconnect_called;
+ pthread_t rdma_cm_event_thread;
int available_recv_wrs;
} conn;
diff --git a/lib/nodes/infiniband.c b/lib/nodes/infiniband.c
index ce5245986..18736b088 100644
--- a/lib/nodes/infiniband.c
+++ b/lib/nodes/infiniband.c
@@ -22,6 +22,7 @@
#include <string.h>
#include <math.h>
+#include <unistd.h> //ToDo: remove me.
#include <villas/nodes/infiniband.h>
#include <villas/plugin.h>
@@ -32,8 +33,6 @@
#include <villas/memory.h>
#include <villas/memory/ib.h>
-static struct rdma_event_channel *rdma_event_channel;
-
int ib_cleanup(struct node *n)
{
struct infiniband *ib = (struct infiniband *) n->_vd;
@@ -59,7 +58,7 @@ int ib_cleanup(struct node *n)
debug(LOG_IB | 3, "Destroyed rdma_cm_id");
// Destroy event channel
- rdma_destroy_event_channel(rdma_event_channel);
+ rdma_destroy_event_channel(ib->ctx.ec);
debug(LOG_IB | 3, "Destroyed event channel");
return 0;
@@ -216,13 +215,6 @@ static void ib_build_ibv(struct node *n)
debug(LOG_IB | 1, "Starting to build IBV components");
- //Allocate protection domain
- ib->ctx.pd = ibv_alloc_pd(ib->ctx.id->verbs);
- if (!ib->ctx.pd)
- error("Could not allocate protection domain in node %s", node_name(n));
-
- debug(LOG_IB | 3, "Allocated Protection Domain");
-
// Initiate poll mode
ib_init_wc_poll(n);
@@ -362,51 +354,6 @@ static int ib_connect_request(struct node *n, struct rdma_cm_id *id)
return 0;
}
-static int ib_event(struct node *n, struct rdma_cm_event *event)
-{
- int ret = 0;
-
- switch(event->event) {
- case RDMA_CM_EVENT_ADDR_RESOLVED:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ADDR_RESOLVED");
- ret = ib_addr_resolved(n);
- break;
- case RDMA_CM_EVENT_ADDR_ERROR:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ADDR_ERROR");
- error("Address resolution (rdma_resolve_addr) failed!");
- case RDMA_CM_EVENT_ROUTE_RESOLVED:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ROUTE_RESOLVED");
- ret = ib_route_resolved(n);
- break;
- case RDMA_CM_EVENT_ROUTE_ERROR:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ROUTE_ERROR");
- error("Route resolution (rdma_resovle_route) failed!");
- case RDMA_CM_EVENT_CONNECT_REQUEST:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_CONNECT_REQUEST");
- ret = ib_connect_request(n, event->id);
- break;
- case RDMA_CM_EVENT_CONNECT_ERROR:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_CONNECT_ERROR");
- error("An error has occurred trying to establish a connection!");
- case RDMA_CM_EVENT_REJECTED:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_REJECTED");
- error("Connection request or response was rejected by the remote end point!");
- case RDMA_CM_EVENT_ESTABLISHED:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ESTABLISHED");
- info("Connection established in node %s", node_name(n));
- ret = 1;
- break;
- case RDMA_CM_EVENT_DISCONNECTED:
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_DISCONNECTED");
- ret = ib_cleanup(n);
- break;
- default:
- error("Unknown event occurred: %u", event->event);
- }
-
- return ret;
-}
-
int ib_reverse(struct node *n)
{
return 0;
@@ -590,62 +537,109 @@ int ib_destroy(struct node *n)
void * ib_rdma_cm_event_thread(void *n)
{
struct node *node = (struct node *) n;
- //struct infiniband *ib = (struct infiniband *) node->_vd;
+ struct infiniband *ib = (struct infiniband *) node->_vd;
struct rdma_cm_event *event;
+ int ret = 0;
- debug(LOG_IB | 1, "Started rdma_cm_event thread of node %s", node_name(node));
- while (rdma_get_cm_event(rdma_event_channel, &event) == 0) {
+ //debug(LOG_IB | 1, "Started rdma_cm_event thread of node %s", node_name(node));
+
+ info("EVENT: %p", ib->ctx.ec);
+ info("FD, i: %i", ib->ctx.ec->fd);
+
+ // Wait until node is completely started
+ //while (node->state != STATE_STARTED);
+
+ while (rdma_get_cm_event(ib->ctx.ec, &event) == 0) {
struct rdma_cm_event event_copy;
+ //ToDo: check if copy is really necessary
memcpy(&event_copy, event, sizeof(*event));
rdma_ack_cm_event(event);
- if (ib_event(n, &event_copy))
+ switch(event_copy.event) {
+ case RDMA_CM_EVENT_ADDR_RESOLVED:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ADDR_RESOLVED");
+
+ ret = ib_addr_resolved(n);
+ break;
+
+ case RDMA_CM_EVENT_ADDR_ERROR:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ADDR_ERROR");
+
+ error("Address resolution (rdma_resolve_addr) failed!");
+ break;
+
+ case RDMA_CM_EVENT_ROUTE_RESOLVED:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ROUTE_RESOLVED");
+
+ ret = ib_route_resolved(n);
+ break;
+
+ case RDMA_CM_EVENT_ROUTE_ERROR:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ROUTE_ERROR");
+
+ error("Route resolution (rdma_resovle_route) failed!");
+ break;
+
+ case RDMA_CM_EVENT_CONNECT_REQUEST:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_CONNECT_REQUEST");
+
+ ret = ib_connect_request(n, event_copy.id);
+ break;
+
+ case RDMA_CM_EVENT_CONNECT_ERROR:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_CONNECT_ERROR");
+
+ error("An error has occurred trying to establish a connection!");
+ break;
+
+ case RDMA_CM_EVENT_REJECTED:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_REJECTED");
+
+ error("Connection request or response was rejected by the remote end point!");
+ break;
+ case RDMA_CM_EVENT_ESTABLISHED:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ESTABLISHED");
+
+ // node->state = STATE_CONNECTED;
+
+ info("Connection established in node %s", node_name(n));
+ break;
+
+ case RDMA_CM_EVENT_DISCONNECTED:
+ debug(LOG_IB | 2, "Received RDMA_CM_EVENT_DISCONNECTED");
+
+ ret = ib_cleanup(n);
+ break;
+
+ default:
+ error("Unknown event occurred: %u", event_copy.event);
+ }
+
+
+ if (ret) //ToDo: Fix me
break;
}
return NULL;
}
-void * ib_disconnect_thread(void *n)
-{
- struct node *node = (struct node *) n;
- struct infiniband *ib = (struct infiniband *) node->_vd;
- struct rdma_cm_event *event;
-
- debug(LOG_IB | 1, "Started disconnect thread of node %s", node_name(node));
-
- while (rdma_get_cm_event(rdma_event_channel, &event) == 0) {
- if (event->event == RDMA_CM_EVENT_DISCONNECTED) {
- debug(LOG_IB | 2, "Received RDMA_CM_EVENT_DISCONNECTED");
-
- rdma_ack_cm_event(event);
- ib->conn.rdma_disconnect_called = 1;
-
- node_stop(node);
- return NULL;
- }
- }
- return NULL;
-}
-
int ib_start(struct node *n)
{
struct infiniband *ib = (struct infiniband *) n->_vd;
- struct rdma_cm_event *event = NULL;
int ret;
debug(LOG_IB | 1, "Started ib_start");
// Create event channel
- rdma_event_channel = rdma_create_event_channel();
- if (!rdma_event_channel)
+ ib->ctx.ec = rdma_create_event_channel();
+ if (!ib->ctx.ec)
error("Failed to create event channel in node %s!", node_name(n));
debug(LOG_IB | 3, "Created event channel");
- ret = rdma_create_id(rdma_event_channel, &ib->ctx.id, NULL, ib->conn.port_space);
+ ret = rdma_create_id(ib->ctx.ec, &ib->ctx.id, NULL, ib->conn.port_space);
if (ret)
error("Failed to create rdma_cm_id of node %s: %s",
node_name(n), gai_strerror(ret));
@@ -684,27 +678,30 @@ int ib_start(struct node *n)
debug(LOG_IB | 3, "Started to listen to rdma_cm_id");
}
+ //Allocate protection domain
+ ib->ctx.pd = ibv_alloc_pd(ib->ctx.id->verbs);
+ if (!ib->ctx.pd)
+ error("Could not allocate protection domain in node %s", node_name(n));
+
+ debug(LOG_IB | 3, "Allocated Protection Domain");
+
+
// Several events should occur on the event channel, to make
// sure the nodes are succesfully connected.
debug(LOG_IB | 1, "Starting to monitor events on rdma_cm_id");
//Create thread to monitor rdma_cm_event channel
- ret = pthread_create(&ib->conn.stop_thread, NULL, ib_disconnect_thread, n);
+ ret = pthread_create(&ib->conn.rdma_cm_event_thread, NULL, ib_rdma_cm_event_thread, n);
if (ret)
- error("Failed to create thread to monitor disconnects in node %s: %s",
+ error("Failed to create thread to monitor rdma_cm events in node %s: %s",
node_name(n), gai_strerror(ret));
-
- //Wait until rdma_cm_event thread sets state to STARTED
- while ( !(n->state == STATE_CONNECTED));
-
return 0;
}
int ib_stop(struct node *n)
{
struct infiniband *ib = (struct infiniband *) n->_vd;
- struct rdma_cm_event *event = NULL;
int ret;
// Call RDMA disconnect function
@@ -717,18 +714,6 @@ int ib_stop(struct node *n)
debug(LOG_IB | 3, "Called rdma_disconnect");
- // If disconnected event already occured, directly call cleanup function
- if (ib->conn.rdma_disconnect_called)
- ib_cleanup(n);
- else {
- // Else, wait for event to occur
- ib->conn.rdma_disconnect_called = 1;
- rdma_get_cm_event(rdma_event_channel, &event);
-
- rdma_ack_cm_event(event);
-
- ib_event(n, event);
- }
return 0;
}
@@ -750,78 +735,81 @@ int ib_read(struct node *n, struct sample *smps[], unsigned cnt)
struct ibv_recv_wr wr[cnt], *bad_wr = NULL;
struct ibv_sge sge[cnt];
struct ibv_mr *mr;
- int ret;
+ int ret = 0;
debug(LOG_IB | 15, "ib_read is called");
- if (ib->conn.available_recv_wrs < ib->qp_init.cap.max_recv_wr && cnt==n->in.vectorize) {
- // Get Memory Region
- mr = memory_ib_get_mr(smps[0]);
+ if (n->state == STATE_CONNECTED) {
- for (int i = 0; i < cnt; i++) {
- // Increase refcnt of sample
- sample_get(smps[i]);
+ if (ib->conn.available_recv_wrs < ib->qp_init.cap.max_recv_wr && cnt==n->in.vectorize) {
+ // Get Memory Region
+ mr = memory_ib_get_mr(smps[0]);
- // Prepare receive Scatter/Gather element
- sge[i].addr = (uint64_t) &smps[i]->data;
- sge[i].length = SAMPLE_DATA_LEN(DEFAULT_SAMPLELEN);
- sge[i].lkey = mr->lkey;
+ for (int i = 0; i < cnt; i++) {
+ // Increase refcnt of sample
+ sample_get(smps[i]);
- // Prepare a receive Work Request
- wr[i].wr_id = (uintptr_t) smps[i];
- wr[i].next = &wr[i+1];
- wr[i].sg_list = &sge[i];
- wr[i].num_sge = 1;
+ // Prepare receive Scatter/Gather element
+ sge[i].addr = (uint64_t) &smps[i]->data;
+ sge[i].length = SAMPLE_DATA_LEN(DEFAULT_SAMPLELEN);
+ sge[i].lkey = mr->lkey;
- ib->conn.available_recv_wrs++;
+ // Prepare a receive Work Request
+ wr[i].wr_id = (uintptr_t) smps[i];
+ wr[i].next = &wr[i+1];
+ wr[i].sg_list = &sge[i];
+ wr[i].num_sge = 1;
- if (ib->conn.available_recv_wrs == ib->qp_init.cap.max_recv_wr || i==(cnt-1)) {
- debug(LOG_IB | 10, "Prepared %i new receive Work Requests", (i+1));
+ ib->conn.available_recv_wrs++;
- wr[i].next = NULL;
- break;
+ if (ib->conn.available_recv_wrs == ib->qp_init.cap.max_recv_wr || i==(cnt-1)) {
+ debug(LOG_IB | 10, "Prepared %i new receive Work Requests", (i+1));
+
+ wr[i].next = NULL;
+ break;
+ }
}
+
+ // Post list of Work Requests
+ ret = ibv_post_recv(ib->ctx.id->qp, &wr[0], &bad_wr);
+ if (ret)
+ error("Was unable to post receive WR in node %s: %i, bad WR ID: 0x%lx",
+ node_name(n), ret, bad_wr->wr_id);
+
+ debug(LOG_IB | 10, "Succesfully posted receive Work Requests");
+
}
- // Post list of Work Requests
- ret = ibv_post_recv(ib->ctx.id->qp, &wr[0], &bad_wr);
- if (ret)
- error("Was unable to post receive WR in node %s: %i, bad WR ID: 0x%lx",
- node_name(n), ret, bad_wr->wr_id);
+ // Poll Completion Queue
+ ret = ibv_poll_cq(ib->ctx.recv_cq, n->in.vectorize, wc);
- debug(LOG_IB | 10, "Succesfully posted receive Work Requests");
+ if (ret) {
+ debug(LOG_IB | 10, "Received %i Work Completions", ret);
- }
+ ib->conn.available_recv_wrs -= ret;
- // Poll Completion Queue
- ret = ibv_poll_cq(ib->ctx.recv_cq, n->in.vectorize, wc);
+ for (int i = 0; i < ret; i++) {
+ if (wc[i].status == IBV_WC_WR_FLUSH_ERR) {
+ debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR in ib_read. Ignore it.");
- if (ret) {
- debug(LOG_IB | 10, "Received %i Work Completions", ret);
+ ret = 0;
+ }
+ else if (wc[i].status != IBV_WC_SUCCESS) {
+ warn("Work Completion status was not IBV_WC_SUCCES in node %s: %i",
+ node_name(n), wc[i].status);
+ ret = 0;
+ }
+ else if (wc[i].opcode & IBV_WC_RECV) {
+ smps[i] = (struct sample*)(wc[i].wr_id);
+ smps[i]->length = wc[i].byte_len/sizeof(double);
+ }
+ else
+ ret = 0;
- ib->conn.available_recv_wrs -= ret;
-
- for (int i = 0; i < ret; i++) {
- if (wc[i].status == IBV_WC_WR_FLUSH_ERR) {
- debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR in ib_read. Ignore it.");
-
- ret = 0;
+ //Release sample
+ sample_put((struct sample *) (wc[i].wr_id));
+ debug(LOG_IB | 10, "Releasing sample %p", (struct sample *) (wc[i].wr_id));
}
- else if (wc[i].status != IBV_WC_SUCCESS) {
- warn("Work Completion status was not IBV_WC_SUCCES in node %s: %i",
- node_name(n), wc[i].status);
- ret = 0;
- }
- else if (wc[i].opcode & IBV_WC_RECV) {
- smps[i] = (struct sample*)(wc[i].wr_id);
- smps[i]->length = wc[i].byte_len/sizeof(double);
- }
- else
- ret = 0;
-
- //Release sample
- sample_put((struct sample *) (wc[i].wr_id));
- debug(LOG_IB | 10, "Releasing sample %p", (struct sample *) (wc[i].wr_id));
}
}
@@ -838,53 +826,55 @@ int ib_write(struct node *n, struct sample *smps[], unsigned cnt)
debug(LOG_IB | 10, "ib_write is called");
- memset(&wr, 0, sizeof(wr));
+ if (n->state == STATE_CONNECTED) {
+ memset(&wr, 0, sizeof(wr));
- //ToDo: Place this into configuration and create checks if settings are valid
- int send_inline = 1;
+ //ToDo: Place this into configuration and create checks if settings are valid
+ int send_inline = 1;
- debug(LOG_IB | 10, "Data will be send inline [0/1]: %i", send_inline);
+ debug(LOG_IB | 10, "Data will be send inline [0/1]: %i", send_inline);
- // Get Memory Region
- mr = memory_ib_get_mr(smps[0]);
+ // Get Memory Region
+ mr = memory_ib_get_mr(smps[0]);
- for (int i = 0; i < cnt; i++) {
- // Increase refcnt of sample
- sample_get(smps[i]);
+ for (int i = 0; i < cnt; i++) {
+ // Increase refcnt of sample
+ sample_get(smps[i]);
- //Set Scatter/Gather element to data of sample
- sge[i].addr = (uint64_t)&smps[i]->data;
- sge[i].length = smps[i]->length*sizeof(double);
- sge[i].lkey = mr->lkey;
+ //Set Scatter/Gather element to data of sample
+ sge[i].addr = (uint64_t)&smps[i]->data;
+ sge[i].length = smps[i]->length*sizeof(double);
+ sge[i].lkey = mr->lkey;
- // Set Send Work Request
- wr[i].wr_id = (uintptr_t)smps[i]; //This way the sample can be release in WC
- wr[i].sg_list = &sge[i];
- wr[i].num_sge = 1;
+ // Set Send Work Request
+ wr[i].wr_id = (uintptr_t)smps[i]; //This way the sample can be release in WC
+ wr[i].sg_list = &sge[i];
+ wr[i].num_sge = 1;
- if (i == (cnt-1)) {
- debug(LOG_IB | 10, "Prepared %i send Work Requests", (i+1));
- wr[i].next = NULL;
+ if (i == (cnt-1)) {
+ debug(LOG_IB | 10, "Prepared %i send Work Requests", (i+1));
+ wr[i].next = NULL;
+ }
+ else
+ wr[i].next = &wr[i+1];
+
+ wr[i].send_flags = IBV_SEND_SIGNALED | (send_inline << 3);
+ wr[i].imm_data = htonl(0); //ToDo: set this to a useful value
+ wr[i].opcode = IBV_WR_SEND_WITH_IMM;
}
- else
- wr[i].next = &wr[i+1];
- wr[i].send_flags = IBV_SEND_SIGNALED | (send_inline << 3);
- wr[i].imm_data = htonl(0); //ToDo: set this to a useful value
- wr[i].opcode = IBV_WR_SEND_WITH_IMM;
+ //Send linked list of Work Requests
+ ret = ibv_post_send(ib->ctx.id->qp, wr, &bad_wr);
+ if (ret) {
+ error("Failed to send message in node %s: %i, bad WR ID: 0x%lx",
+ node_name(n), ret, bad_wr->wr_id);
+
+ return -ret;
+ }
+
+ debug(LOG_IB | 4, "Succesfully posted receive Work Requests");
}
- //Send linked list of Work Requests
- ret = ibv_post_send(ib->ctx.id->qp, wr, &bad_wr);
- if (ret) {
- error("Failed to send message in node %s: %i, bad WR ID: 0x%lx",
- node_name(n), ret, bad_wr->wr_id);
-
- return -ret;
- }
-
- debug(LOG_IB | 4, "Succesfully posted receive Work Requests");
-
return cnt;
}
From e16644b0e907ea8650ee474b4590c9832e769dba Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Sat, 7 Jul 2018 12:56:08 +0200
Subject: [PATCH 3/7] Added some error handling in IB memory type
---
lib/memory/ib.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/lib/memory/ib.c b/lib/memory/ib.c
index 6cf0dbc66..5943fd8e9 100644
--- a/lib/memory/ib.c
+++ b/lib/memory/ib.c
@@ -55,6 +55,9 @@ static struct memory_allocation * memory_ib_alloc(struct memory_type *m, size_t
ma->parent = mi->parent->alloc(mi->parent, len + sizeof(struct ibv_mr *), alignment);
ma->address = ma->parent->address;
+ if(!mi->pd)
+ error("Protection domain is not registered!");
+
ma->ib.mr = ibv_reg_mr(mi->pd, ma->address, ma->length, IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
if(!ma->ib.mr) {
mi->parent->free(mi->parent, ma->parent);
From 80da4801e15f8482f30c3c225ef1eba5a77ec79d Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Sat, 7 Jul 2018 13:08:08 +0200
Subject: [PATCH 4/7] Source and target successfully connect and node changes
status from STATE_STARTED to STATE_CONNECTED in this commit. Next step will
be to fix ib_stop and ib_disconnect to make the target able to accept new
connections.
---
lib/nodes/infiniband.c | 22 ++++++++--------------
1 file changed, 8 insertions(+), 14 deletions(-)
diff --git a/lib/nodes/infiniband.c b/lib/nodes/infiniband.c
index 18736b088..6544ae0d9 100644
--- a/lib/nodes/infiniband.c
+++ b/lib/nodes/infiniband.c
@@ -542,22 +542,15 @@ void * ib_rdma_cm_event_thread(void *n)
int ret = 0;
- //debug(LOG_IB | 1, "Started rdma_cm_event thread of node %s", node_name(node));
-
- info("EVENT: %p", ib->ctx.ec);
- info("FD, i: %i", ib->ctx.ec->fd);
+ debug(LOG_IB | 1, "Started rdma_cm_event thread of node %s", node_name(node));
// Wait until node is completely started
- //while (node->state != STATE_STARTED);
+ while (node->state != STATE_STARTED);
+ // Monitor event channel
while (rdma_get_cm_event(ib->ctx.ec, &event) == 0) {
- struct rdma_cm_event event_copy;
- //ToDo: check if copy is really necessary
- memcpy(&event_copy, event, sizeof(*event));
- rdma_ack_cm_event(event);
-
- switch(event_copy.event) {
+ switch(event->event) {
case RDMA_CM_EVENT_ADDR_RESOLVED:
debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ADDR_RESOLVED");
@@ -585,7 +578,7 @@ void * ib_rdma_cm_event_thread(void *n)
case RDMA_CM_EVENT_CONNECT_REQUEST:
debug(LOG_IB | 2, "Received RDMA_CM_EVENT_CONNECT_REQUEST");
- ret = ib_connect_request(n, event_copy.id);
+ ret = ib_connect_request(n, event->id);
break;
case RDMA_CM_EVENT_CONNECT_ERROR:
@@ -602,7 +595,7 @@ void * ib_rdma_cm_event_thread(void *n)
case RDMA_CM_EVENT_ESTABLISHED:
debug(LOG_IB | 2, "Received RDMA_CM_EVENT_ESTABLISHED");
- // node->state = STATE_CONNECTED;
+ node->state = STATE_CONNECTED;
info("Connection established in node %s", node_name(n));
break;
@@ -614,9 +607,10 @@ void * ib_rdma_cm_event_thread(void *n)
break;
default:
- error("Unknown event occurred: %u", event_copy.event);
+ error("Unknown event occurred: %u", event->event);
}
+ rdma_ack_cm_event(event);
if (ret) //ToDo: Fix me
break;
From 836adee4d6ee9cd3477b85052ee6f6d9d3598c44 Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Sat, 7 Jul 2018 14:36:23 +0200
Subject: [PATCH 5/7] Node is able to clean everything up and reconnect. Node
can abort if it is in STARTED and in CONNECTED state
---
include/villas/nodes/infiniband.h | 4 +-
lib/nodes/infiniband.c | 111 +++++++++++++++---------------
2 files changed, 59 insertions(+), 56 deletions(-)
diff --git a/include/villas/nodes/infiniband.h b/include/villas/nodes/infiniband.h
index 449d8c88c..5f5727ea0 100644
--- a/include/villas/nodes/infiniband.h
+++ b/include/villas/nodes/infiniband.h
@@ -77,9 +77,11 @@ struct infiniband {
/* Poll thread */
pthread_t cq_poller_thread;
- int stopThread;
+ int stopThreads;
} poll;
+ int stopThreads;
+
/* Connection specific variables */
struct connection_s {
struct addrinfo *src_addr;
diff --git a/lib/nodes/infiniband.c b/lib/nodes/infiniband.c
index 6544ae0d9..2c326eacb 100644
--- a/lib/nodes/infiniband.c
+++ b/lib/nodes/infiniband.c
@@ -22,7 +22,6 @@
#include <string.h>
#include <math.h>
-#include <unistd.h> //ToDo: remove me.
#include <villas/nodes/infiniband.h>
#include <villas/plugin.h>
@@ -33,7 +32,7 @@
#include <villas/memory.h>
#include <villas/memory/ib.h>
-int ib_cleanup(struct node *n)
+int ib_disconnect(struct node *n)
{
struct infiniband *ib = (struct infiniband *) n->_vd;
debug(LOG_IB | 1, "Starting to clean up");
@@ -53,13 +52,8 @@ int ib_cleanup(struct node *n)
pool_destroy(&ib->mem.p_send);
debug(LOG_IB | 3, "Destroyed memory pools");
- // Destroy RDMA CM ID
- rdma_destroy_id(ib->ctx.id);
- debug(LOG_IB | 3, "Destroyed rdma_cm_id");
-
- // Destroy event channel
- rdma_destroy_event_channel(ib->ctx.ec);
- debug(LOG_IB | 3, "Destroyed event channel");
+ // Set available receive work requests to zero
+ ib->conn.available_recv_wrs = 0;
return 0;
}
@@ -97,9 +91,9 @@ void ib_completion_source(struct node* n, struct ibv_wc* wc, int* size)
for (int i = 0; i < *size; i++) {
//On disconnect, the QP set to error state and will be flushed
if (wc[i].status == IBV_WC_WR_FLUSH_ERR) {
- debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR in ib_completion_source. Stopping thread.");
+ debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR (ib_completion_source). Stopping thread.");
- ib->poll.stopThread = 1;
+ ib->poll.stopThreads = 1;
return;
}
@@ -146,7 +140,7 @@ void * ib_busy_poll_thread(void *n)
while ((size = ibv_poll_cq(ib->ctx.send_cq, ib->cq_size, wc)))
ib->poll.on_compl(n, wc, &size);
- if (ib->poll.stopThread)
+ if (ib->poll.stopThreads)
return NULL;
}
}
@@ -169,21 +163,13 @@ static void ib_init_wc_poll(struct node *n)
}
// Create completion queues and bind to channel (or NULL)
- ib->ctx.recv_cq = ibv_create_cq(ib->ctx.id->verbs,
- ib->cq_size,
- NULL,
- NULL,
- 0);
+ ib->ctx.recv_cq = ibv_create_cq(ib->ctx.id->verbs, ib->cq_size, NULL, NULL, 0);
if (!ib->ctx.recv_cq)
error("Could not create receive completion queue in node %s", node_name(n));
debug(LOG_IB | 3, "Created receive Completion Queue");
- ib->ctx.send_cq = ibv_create_cq(ib->ctx.id->verbs,
- ib->cq_size,
- NULL,
- ib->ctx.comp_channel,
- 0);
+ ib->ctx.send_cq = ibv_create_cq(ib->ctx.id->verbs, ib->cq_size, NULL, ib->ctx.comp_channel, 0);
if (!ib->ctx.send_cq)
error("Could not create send completion queue in node %s", node_name(n));
@@ -237,10 +223,7 @@ static void ib_build_ibv(struct node *n)
ib->mem.p_recv.queue.state = STATE_DESTROYED;
// Set pool size to maximum size of Receive Queue
- pool_init(&ib->mem.p_recv,
- ib->qp_init.cap.max_recv_wr,
- SAMPLE_DATA_LEN(DEFAULT_SAMPLELEN),
- &memory_type_heap);
+ pool_init(&ib->mem.p_recv, ib->qp_init.cap.max_recv_wr, SAMPLE_DATA_LEN(DEFAULT_SAMPLELEN), &memory_type_heap);
if (ret)
error("Failed to init recv memory pool of node %s: %s",
node_name(n), gai_strerror(ret));
@@ -252,8 +235,7 @@ static void ib_build_ibv(struct node *n)
// Register memory for IB Device. Not necessary if data is send
// exclusively inline
- ib->mem.mr_recv = ibv_reg_mr(
- ib->ctx.pd,
+ ib->mem.mr_recv = ibv_reg_mr(ib->ctx.pd,
(char*)&ib->mem.p_recv+ib->mem.p_recv.buffer_off,
ib->mem.p_recv.len,
IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
@@ -268,29 +250,22 @@ static void ib_build_ibv(struct node *n)
ib->mem.p_send.queue.state = STATE_DESTROYED;
// Set pool size to maximum size of Receive Queue
- pool_init(&ib->mem.p_send,
- ib->qp_init.cap.max_send_wr,
- sizeof(double),
- &memory_type_heap);
+ pool_init(&ib->mem.p_send, ib->qp_init.cap.max_send_wr, sizeof(double), &memory_type_heap);
if (ret)
- error("Failed to init send memory of node %s: %s",
- node_name(n), gai_strerror(ret));
+ error("Failed to init send memory of node %s: %s", node_name(n), gai_strerror(ret));
- debug(LOG_IB | 3, "Created internal send pool with %i elements",
- ib->qp_init.cap.max_recv_wr);
+ debug(LOG_IB | 3, "Created internal send pool with %i elements", ib->qp_init.cap.max_recv_wr);
//ToDo: initialize r_addr_key struct if mode is RDMA
// Register memory for IB Device. Not necessary if data is send
// exclusively inline
- ib->mem.mr_send = ibv_reg_mr(
- ib->ctx.pd,
+ ib->mem.mr_send = ibv_reg_mr(ib->ctx.pd,
(char*)&ib->mem.p_send+ib->mem.p_send.buffer_off,
ib->mem.p_send.len,
IBV_ACCESS_LOCAL_WRITE | IBV_ACCESS_REMOTE_WRITE);
if (!ib->mem.mr_send)
- error("Failed to register mr_send with ibv_reg_mr of node %s",
- node_name(n));
+ error("Failed to register mr_send with ibv_reg_mr of node %s", node_name(n));
debug(LOG_IB | 3, "Registered send pool with ibv_reg_mr");
}
@@ -463,7 +438,7 @@ int ib_parse(struct node *n, json_t *cfg)
//Check if node is a source and connect to target
if (remote) {
- debug(LOG_IB | 3, "Node %s is set up to be able to send data (source and target)", node_name(n));
+ debug(LOG_IB | 3, "Node %s is up as source and target", node_name(n));
ib->is_source = 1;
@@ -476,13 +451,13 @@ int ib_parse(struct node *n, json_t *cfg)
error("Failed to resolve remote address '%s' of node %s: %s",
remote, node_name(n), gai_strerror(ret));
- debug(LOG_IB | 4, "Translated %s:%s to a struct addrinfo in node %s", ip_adr, port, node_name(n));
+ debug(LOG_IB | 4, "Translated %s:%s to a struct addrinfo", ip_adr, port);
// Set correct Work Completion function
ib->poll.on_compl = ib_completion_source;
}
else {
- debug(LOG_IB | 3, "Node %s is set up to be able to only receive data (target)", node_name(n));
+ debug(LOG_IB | 3, "Node %s is set up as target", node_name(n));
ib->is_source = 0;
@@ -504,11 +479,11 @@ int ib_check(struct node *n)
int max_recv_pow = (int) pow(2, ceil(log2(ib->qp_init.cap.max_recv_wr)));
if (ib->qp_init.cap.max_send_wr != max_send_pow)
- warn("Max nr. of send WRs (%i) is not a power of 2! The HCA will change it to the next power of 2: %i",
+ warn("Max nr. of send WRs (%i) is not a power of 2! It will be changed to a power of 2: %i",
ib->qp_init.cap.max_send_wr, max_send_pow);
if (ib->qp_init.cap.max_recv_wr != max_recv_pow)
- warn("Max nr. of recv WRs (%i) is not a power of 2! The HCA will change it to the next power of 2: %i",
+ warn("Max nr. of recv WRs (%i) is not a power of 2! It will be changed to a power of 2: %i",
ib->qp_init.cap.max_recv_wr, max_recv_pow);
@@ -603,7 +578,12 @@ void * ib_rdma_cm_event_thread(void *n)
case RDMA_CM_EVENT_DISCONNECTED:
debug(LOG_IB | 2, "Received RDMA_CM_EVENT_DISCONNECTED");
- ret = ib_cleanup(n);
+ node->state = STATE_STARTED;
+ ret = ib_disconnect(n);
+
+ break;
+
+ case RDMA_CM_EVENT_TIMEWAIT_EXIT:
break;
default:
@@ -612,7 +592,7 @@ void * ib_rdma_cm_event_thread(void *n)
rdma_ack_cm_event(event);
- if (ret) //ToDo: Fix me
+ if (ret || ib->stopThreads)
break;
}
@@ -635,8 +615,7 @@ int ib_start(struct node *n)
ret = rdma_create_id(ib->ctx.ec, &ib->ctx.id, NULL, ib->conn.port_space);
if (ret)
- error("Failed to create rdma_cm_id of node %s: %s",
- node_name(n), gai_strerror(ret));
+ error("Failed to create rdma_cm_id of node %s: %s", node_name(n), gai_strerror(ret));
debug(LOG_IB | 3, "Created rdma_cm_id");
@@ -650,10 +629,7 @@ int ib_start(struct node *n)
if (ib->is_source) {
// Resolve address
- ret = rdma_resolve_addr(ib->ctx.id,
- NULL,
- ib->conn.dst_addr->ai_addr,
- ib->conn.timeout);
+ ret = rdma_resolve_addr(ib->ctx.id, NULL, ib->conn.dst_addr->ai_addr, ib->conn.timeout);
if (ret)
error("Failed to resolve remote address after %ims of node %s: %s",
ib->conn.timeout, node_name(n), gai_strerror(ret));
@@ -698,16 +674,41 @@ int ib_stop(struct node *n)
struct infiniband *ib = (struct infiniband *) n->_vd;
int ret;
+ debug(LOG_IB | 1, "Called ib_stop");
+
+ ib->stopThreads = 1;
+
// Call RDMA disconnect function
// Will flush all outstanding WRs to the Completion Queue and
// will call RDMA_CM_EVENT_DISCONNECTED if that is done.
ret = rdma_disconnect(ib->ctx.id);
if (ret)
error("Error while calling rdma_disconnect in node %s: %s",
- node_name(n), gai_strerror(ret));
+ node_name(n), gai_strerror(ret));
debug(LOG_IB | 3, "Called rdma_disconnect");
+ info("Disconnecting... Please give me a few seconds.");
+ // Wait for event thread to join
+ ret = pthread_join(ib->conn.rdma_cm_event_thread, NULL);
+ if (ret)
+ error("Error while joining rdma_cm_event_thread in node %s: %i", node_name(n), ret);
+
+ debug(LOG_IB | 3, "Joined rdma_cm_event_thread");
+
+ // Destroy RDMA CM ID
+ rdma_destroy_id(ib->ctx.id);
+ debug(LOG_IB | 3, "Destroyed rdma_cm_id");
+
+ // Dealloc Protection Domain
+ ibv_dealloc_pd(ib->ctx.pd);
+ debug(LOG_IB | 3, "Destroyed protection domain");
+
+ // Destroy event channel
+ rdma_destroy_event_channel(ib->ctx.ec);
+ debug(LOG_IB | 3, "Destroyed event channel");
+
+ info("Successfully stopped %s", node_name(n));
return 0;
}
@@ -784,7 +785,7 @@ int ib_read(struct node *n, struct sample *smps[], unsigned cnt)
for (int i = 0; i < ret; i++) {
if (wc[i].status == IBV_WC_WR_FLUSH_ERR) {
- debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR in ib_read. Ignore it.");
+ debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR (ib_read). Ignore it.");
ret = 0;
}
From 1bdd0a9e3455c2d0197fc0d45d77ace5551af7ac Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Sat, 7 Jul 2018 15:24:48 +0200
Subject: [PATCH 6/7] node_stop can be called if node is STARTED or CONNECTED
---
lib/node.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/node.c b/lib/node.c
index 5580bc399..c7d93b84b 100644
--- a/lib/node.c
+++ b/lib/node.c
@@ -351,7 +351,7 @@ int node_stop(struct node *n)
{
int ret;
- if (n->state != STATE_STARTED)
+ if (n->state != STATE_STARTED && n->state != STATE_CONNECTED)
return 0;
info("Stopping node %s", node_name(n));
From 06e7434d6c23f102b6b37d8c842809d05f10aeec Mon Sep 17 00:00:00 2001
From: Dennis Potter <dennis@dennispotter.eu>
Date: Sat, 7 Jul 2018 15:34:07 +0200
Subject: [PATCH 7/7] Solved some state problems. This commit also solves #154,
which was caused by a non-terminated thread. (This thread will be removed in
a later commit anyway
---
include/villas/nodes/infiniband.h | 2 --
lib/nodes/infiniband.c | 41 ++++++++++++++++++-------------
2 files changed, 24 insertions(+), 19 deletions(-)
diff --git a/include/villas/nodes/infiniband.h b/include/villas/nodes/infiniband.h
index 5f5727ea0..cf2cb286d 100644
--- a/include/villas/nodes/infiniband.h
+++ b/include/villas/nodes/infiniband.h
@@ -76,8 +76,6 @@ struct infiniband {
/* Poll thread */
pthread_t cq_poller_thread;
-
- int stopThreads;
} poll;
int stopThreads;
diff --git a/lib/nodes/infiniband.c b/lib/nodes/infiniband.c
index 2c326eacb..5b103b71e 100644
--- a/lib/nodes/infiniband.c
+++ b/lib/nodes/infiniband.c
@@ -37,6 +37,8 @@ int ib_disconnect(struct node *n)
struct infiniband *ib = (struct infiniband *) n->_vd;
debug(LOG_IB | 1, "Starting to clean up");
+ rdma_disconnect(ib->ctx.id);
+
// Destroy QP
rdma_destroy_qp(ib->ctx.id);
debug(LOG_IB | 3, "Destroyed QP");
@@ -86,17 +88,7 @@ void ib_completion_target(struct node* n, struct ibv_wc* wc, int* size){}
void ib_completion_source(struct node* n, struct ibv_wc* wc, int* size)
{
- struct infiniband *ib = (struct infiniband *) ((struct node *) n)->_vd;
-
for (int i = 0; i < *size; i++) {
- //On disconnect, the QP set to error state and will be flushed
- if (wc[i].status == IBV_WC_WR_FLUSH_ERR) {
- debug(LOG_IB | 5, "Received IBV_WC_WR_FLUSH_ERR (ib_completion_source). Stopping thread.");
-
- ib->poll.stopThreads = 1;
- return;
- }
-
if (wc[i].status != IBV_WC_SUCCESS)
warn("Work Completion status was not IBV_WC_SUCCES in node %s: %i",
node_name(n), wc[i].status);
@@ -140,7 +132,7 @@ void * ib_busy_poll_thread(void *n)
while ((size = ibv_poll_cq(ib->ctx.send_cq, ib->cq_size, wc)))
ib->poll.on_compl(n, wc, &size);
- if (ib->poll.stopThreads)
+ if (ib->stopThreads)
return NULL;
}
}
@@ -581,6 +573,8 @@ void * ib_rdma_cm_event_thread(void *n)
node->state = STATE_STARTED;
ret = ib_disconnect(n);
+ info("Host disconnected. Ready to accept new connections.");
+
break;
case RDMA_CM_EVENT_TIMEWAIT_EXIT:
@@ -627,6 +621,12 @@ int ib_start(struct node *n)
debug(LOG_IB | 3, "Bound rdma_cm_id to Infiniband device");
+ // The ID will be overwritten for the target. If the event type is
+ // RDMA_CM_EVENT_CONNECT_REQUEST, >then this references a new id for
+ // that communication.
+ ib->ctx.listen_id = ib->ctx.id;
+
+
if (ib->is_source) {
// Resolve address
ret = rdma_resolve_addr(ib->ctx.id, NULL, ib->conn.dst_addr->ai_addr, ib->conn.timeout);
@@ -635,11 +635,6 @@ int ib_start(struct node *n)
ib->conn.timeout, node_name(n), gai_strerror(ret));
}
else {
- // The ID will be overwritten for the target. If the event type is
- // RDMA_CM_EVENT_CONNECT_REQUEST, >then this references a new id for
- // that communication.
- ib->ctx.listen_id = ib->ctx.id;
-
// Listen on rdma_cm_id for events
ret = rdma_listen(ib->ctx.listen_id, 10);
if (ret)
@@ -681,7 +676,11 @@ int ib_stop(struct node *n)
// Call RDMA disconnect function
// Will flush all outstanding WRs to the Completion Queue and
// will call RDMA_CM_EVENT_DISCONNECTED if that is done.
- ret = rdma_disconnect(ib->ctx.id);
+ if(! ib->is_source && n->state == STATE_CONNECTED)
+ ret = rdma_disconnect(ib->ctx.id);
+ else
+ ret = rdma_disconnect(ib->ctx.listen_id);
+
if (ret)
error("Error while calling rdma_disconnect in node %s: %s",
node_name(n), gai_strerror(ret));
@@ -696,6 +695,14 @@ int ib_stop(struct node *n)
debug(LOG_IB | 3, "Joined rdma_cm_event_thread");
+ // Wait for polling thread to join
+ if (ib->is_source) {
+ ret = pthread_join(ib->poll.cq_poller_thread, NULL);
+ if (ret)
+ error("Error while joining cq_poller_thread in node %s: %i", node_name(n), ret);
+ }
+
+
// Destroy RDMA CM ID
rdma_destroy_id(ib->ctx.id);
debug(LOG_IB | 3, "Destroyed rdma_cm_id");