2.19.1-1

Add local user buffer registration for NVLink SHARP.
Add tuning plugin support.
Increase net API to v7 to allow for device-side packet reordering;
remove support for v4 plugins.
Add support for RoCE ECE.
Add support for C2C links.
Better detect SHM allocation failures to avoid crash with Bus Error.
Fix missing thread unlocks in bootstrap (Fixes #936).
Disable network flush by default on H100.
Move device code from src/collectives/device to src/device.

ext-net/example/plugin.c

@@ -4,7 +4,7 @@
  * See LICENSE.txt for license information
  ************************************************************************/
 
-#include <nccl/net.h>
+#include "net.h"
 
 #define __hidden __attribute__ ((visibility("hidden")))
 
@@ -15,14 +15,14 @@ __hidden ncclResult_t pluginDevices(int* ndev) { *ndev = 0; return ncclSuccess;
 
 __hidden ncclResult_t pluginPciPath(int dev, char** path) { return ncclInternalError; }
 __hidden ncclResult_t pluginPtrSupport(int dev, int* supportedTypes) { return ncclInternalError; }
-__hidden ncclResult_t pluginGetProperties(int dev, ncclNetProperties_v6_t* props) {
+__hidden ncclResult_t pluginGetProperties(int dev, ncclNetProperties_v7_t* props) {
   //pluginPciPath(dev, &props.pciPath);
   //pluginPtrSupport(dev, &props.ptrSupport);
   return ncclInternalError;
 }
 __hidden ncclResult_t pluginListen(int dev, void* handle, void** listenComm) { return ncclInternalError; }
-__hidden ncclResult_t pluginConnect(int dev, void* handle, void** sendComm) { return ncclInternalError; }
-__hidden ncclResult_t pluginAccept(void* listenComm, void** recvComm) { return ncclInternalError; }
+__hidden ncclResult_t pluginConnect(int dev, void* handle, void** sendComm, ncclNetDeviceHandle_v7_t** sendDevComm) { return ncclInternalError; }
+__hidden ncclResult_t pluginAccept(void* listenComm, void** recvComm, ncclNetDeviceHandle_v7_t** recvDevComm) { return ncclInternalError; }
 __hidden ncclResult_t pluginRegMr(void* collComm, void* data, int size, int type, void** mhandle) { return ncclInternalError; }
 __hidden ncclResult_t pluginRegMrDmaBuf(void* collComm, void* data, size_t size, int type, uint64_t offset, int fd, void** mhandle) { return ncclInternalError; }
 __hidden ncclResult_t pluginDeregMr(void* collComm, void* mhandle) { return ncclInternalError;}
@@ -33,10 +33,12 @@ __hidden ncclResult_t pluginTest(void* request, int* done, int* size) { return n
 __hidden ncclResult_t pluginCloseSend(void* sendComm) { return ncclInternalError; }
 __hidden ncclResult_t pluginCloseRecv(void* recvComm) { return ncclInternalError; }
 __hidden ncclResult_t pluginCloseListen(void* listenComm) { return ncclInternalError; }
+__hidden ncclResult_t pluginIrecvConsumed(void* recvComm, int n, void* request) { return ncclInternalError; }
+__hidden ncclResult_t pluginGetDeviceMr(void* comm, void* mhandle, void** dptr_mhandle) { return ncclInternalError; }
 
 #define PLUGIN_NAME "Plugin"
 
-const ncclNet_v6_t ncclNetPlugin_v6 = {
+const ncclNet_v7_t ncclNetPlugin_v7 = {
   .name = PLUGIN_NAME,
   .init = pluginInit,
   .devices = pluginDevices,
@@ -54,6 +56,37 @@ const ncclNet_v6_t ncclNetPlugin_v6 = {
   .closeSend = pluginCloseSend,
   .closeRecv = pluginCloseRecv,
   .closeListen = pluginCloseListen,
+  .getDeviceMr = pluginGetDeviceMr,
+  .irecvConsumed = pluginIrecvConsumed,
+};
+
+__hidden ncclResult_t pluginGetProperties_v6(int dev, ncclNetProperties_v6_t* props) {
+  //pluginPciPath(dev, &props.pciPath);
+  //pluginPtrSupport(dev, &props.ptrSupport);
+  return ncclInternalError;
+}
+
+__hidden ncclResult_t pluginConnect_v6(int dev, void* handle, void** sendComm) { return ncclInternalError; }
+__hidden ncclResult_t pluginAccept_v6(void* listenComm, void** recvComm) { return ncclInternalError; }
+
+const ncclNet_v6_t ncclNetPlugin_v6 = {
+  .name = PLUGIN_NAME,
+  .init = pluginInit,
+  .devices = pluginDevices,
+  .getProperties = pluginGetProperties_v6,
+  .listen = pluginListen,
+  .connect = pluginConnect_v6,
+  .accept = pluginAccept_v6,
+  .regMr = pluginRegMr,
+  .regMrDmaBuf = pluginRegMrDmaBuf,
+  .deregMr = pluginDeregMr,
+  .isend = pluginIsend,
+  .irecv = pluginIrecv,
+  .iflush = pluginIflush,
+  .test = pluginTest,
+  .closeSend = pluginCloseSend,
+  .closeRecv = pluginCloseRecv,
+  .closeListen = pluginCloseListen
 };
 
 /* v5 Compat */
@@ -61,10 +94,10 @@ const ncclNet_v5_t ncclNetPlugin_v5 = {
   .name = PLUGIN_NAME,
   .init = pluginInit,
   .devices = pluginDevices,
-  .getProperties = pluginGetProperties,
+  .getProperties = pluginGetProperties_v6,
   .listen = pluginListen,
-  .connect = pluginConnect,
-  .accept = pluginAccept,
+  .connect = pluginConnect_v6,
+  .accept = pluginAccept_v6,
   .regMr = pluginRegMr,
   .deregMr = pluginDeregMr,
   .isend = pluginIsend,
@@ -79,7 +112,7 @@ const ncclNet_v5_t ncclNetPlugin_v5 = {
 /* v4 Compat */
 static ncclResult_t pluginGetProperties_v4(int dev, ncclNetProperties_v4_t* props) {
   ncclNetProperties_v6_t props_v6;
-  ncclResult_t ret = pluginGetProperties(dev, &props_v6);
+  ncclResult_t ret = pluginGetProperties_v6(dev, &props_v6);
   if (ret != ncclSuccess) return ret;
   props->name = props_v6.name;
   props->pciPath = props_v6.pciPath;
@@ -103,14 +136,16 @@ static ncclResult_t pluginIflush_v4(void* recvComm, void* data, int size, void*
 static ncclResult_t pluginConnect_v4(int dev, void* handle, void** sendComm) {
   ncclResult_t ret;
   do {
-    ret = pluginConnect(dev, handle, sendComm);
+    ncclNetDeviceHandle_v7_t* handle = NULL;
+    ret = pluginConnect(dev, handle, sendComm, &handle);
   } while (ret == ncclSuccess && *sendComm == NULL);
   return ret;
 }
 static ncclResult_t pluginAccept_v4(void* listenComm, void** recvComm) {
   ncclResult_t ret;
   do {
-    ret = pluginAccept(listenComm, recvComm);
+    ncclNetDeviceHandle_v7_t* handle = NULL;
+    ret = pluginAccept(listenComm, recvComm, &handle);
   } while (ret == ncclSuccess && *recvComm == NULL);
   return ret;
 }
@@ -151,12 +186,12 @@ static ncclResult_t pluginInit_v3(ncclDebugLogger_t logFunction) {
 static ncclResult_t pluginListen_v3(int dev, void* handle, void** listenComm) {
   char pluginHandle[NCCL_NET_HANDLE_MAXSIZE];
   ncclResult_t ret = pluginListen(dev, &pluginHandle, listenComm);
-  memcpy(handle, &pluginHandle, NCCL_NET_HANDLE_MAXSIZE_V3);
+  memcpy(handle, &pluginHandle, NCCL_NET_HANDLE_MAXSIZE_V4);
   return ret;
 }
 static ncclResult_t pluginConnect_v3(int dev, void* handle, void** sendComm) {
   char pluginHandle[NCCL_NET_HANDLE_MAXSIZE];
-  memcpy(&pluginHandle, handle, NCCL_NET_HANDLE_MAXSIZE_V3);
+  memcpy(&pluginHandle, handle, NCCL_NET_HANDLE_MAXSIZE_V4);
   return pluginConnect_v4(dev, &pluginHandle, sendComm);
 }
 const ncclNet_v3_t ncclNetPlugin_v3 = {