AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

2.12.10-1

Browse Source 
Fix bug with CollNet
Fix bug with zero-bytes send/recv operations
Fix NCCL_PARAM implementation to avoid taking a lock on every call
Fix bug when setting NCCL_IB_QPS_PER_CONNECTION to more than one.
Improve error reporting for network errors.
This commit is contained in:
Sylvain Jeaugey
2022-03-30 02:25:49 -07:00
parent 44eb40da0e
commit 353e8ba446
16 changed files with 203 additions and 114 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/misc/socket.cc
+7 -3
View File
@@ -16,12 +16,16 @@
*
* Output: "IPv4/IPv6 address<port>"
*/
const char *ncclSocketToString(union ncclSocketAddress *addr, char *buf) {
const char *ncclSocketToString(union ncclSocketAddress *addr, char *buf, const int numericHostForm /*= 1*/) {
if (buf == NULL || addr == NULL) return NULL;
struct sockaddr *saddr = &addr->sa;
if (saddr->sa_family != AF_INET && saddr->sa_family != AF_INET6) { buf[0]='\0'; return buf; }
char host[NI_MAXHOST], service[NI_MAXSERV];
(void) getnameinfo(saddr, sizeof(union ncclSocketAddress), host, NI_MAXHOST, service, NI_MAXSERV, NI_NUMERICHOST|NI_NUMERICSERV);
/* NI_NUMERICHOST: If set, then the numeric form of the hostname is returned.
* (When not set, this will still happen in case the node's name cannot be determined.)
*/
int flag = NI_NUMERICSERV | (numericHostForm ? NI_NUMERICHOST : 0);
(void) getnameinfo(saddr, sizeof(union ncclSocketAddress), host, NI_MAXHOST, service, NI_MAXSERV, flag);
sprintf(buf, "%s<%s>", host, service);
return buf;
}
@@ -516,7 +520,7 @@ ncclResult_t ncclSocketProgress(int op, struct ncclSocket* sock, void* ptr, int
NCCLCHECK(ncclSocketProgressOpt(op, sock, ptr, size, offset, 0, &closed));
if (closed) {
char line[SOCKET_NAME_MAXLEN+1];
WARN("Net : Connection closed by remote peer %s", ncclSocketToString(&sock->addr, line));
WARN("Net : Connection closed by remote peer %s", ncclSocketToString(&sock->addr, line, 0));
return ncclSystemError;
}
return ncclSuccess;