Add ability for user to empty the CUDA IPC registration cache when it is full

opal/mca/mpool/rgpusm/mpool_rgpusm.h

@@ -11,7 +11,7 @@
  * Copyright (c) 2004-2005 The Regents of the University of California.
  *                         All rights reserved.
  * Copyright (c) 2006      Voltaire. All rights reserved.
- * Copyright (c) 2012      NVIDIA Corporation.  All rights reserved.
+ * Copyright (c) 2012-2015 NVIDIA Corporation.  All rights reserved.
  * Copyright (c) 2015      Los Alamos National Security, LLC.  All rights
  *                         reserved.
  *
@@ -41,6 +41,7 @@ struct mca_mpool_rgpusm_component_t {
     bool print_stats;
     int leave_pinned;
     int output;
+    bool empty_cache;
 };
 typedef struct mca_mpool_rgpusm_component_t mca_mpool_rgpusm_component_t;
 

opal/mca/mpool/rgpusm/mpool_rgpusm_component.c

@@ -12,7 +12,7 @@
  *                         All rights reserved.
  * Copyright (c) 2006      Voltaire. All rights reserved.
  * Copyright (c) 2007-2009 Cisco Systems, Inc.  All rights reserved.
- * Copyright (c) 2012      NVIDIA Corporation.  All rights reserved.
+ * Copyright (c) 2012-2015 NVIDIA Corporation.  All rights reserved.
  * Copyright (c) 2015      Los Alamos National Security, LLC.  All rights
  *                         reserved.
  *
@@ -127,6 +127,15 @@ static int rgpusm_register(void)
                                            MCA_BASE_VAR_SCOPE_READONLY,
                                            &opal_mpool_rgpusm_verbose);
 
+    /* Force emptying of entire registration cache when it gets full */
+    mca_mpool_rgpusm_component.empty_cache = false;
+    (void) mca_base_component_var_register(&mca_mpool_rgpusm_component.super.mpool_version,
+                                           "empty_cache", "When set, empty entire registration cache when it is full",
+                                           MCA_BASE_VAR_TYPE_BOOL, NULL, 0, 0,
+                                           OPAL_INFO_LVL_5,
+                                           MCA_BASE_VAR_SCOPE_READONLY,
+                                           &mca_mpool_rgpusm_component.empty_cache);
+
     return OPAL_SUCCESS;
 }
 

opal/mca/mpool/rgpusm/mpool_rgpusm_module.c

@@ -14,7 +14,7 @@
  * Copyright (c) 2006      Voltaire. All rights reserved.
  * Copyright (c) 2007      Mellanox Technologies. All rights reserved.
  * Copyright (c) 2010      IBM Corporation.  All rights reserved.
- * Copyright (c) 2012-2014 NVIDIA Corporation.  All rights reserved.
+ * Copyright (c) 2012-2015 NVIDIA Corporation.  All rights reserved.
  * Copyright (c) 2015      Los Alamos National Security, LLC.  All rights
  *                         reserved.
  *
@@ -406,12 +406,35 @@ int mca_mpool_rgpusm_register(mca_mpool_base_module_t *mpool, void *addr,
 
     opal_output_verbose(80, mca_mpool_rgpusm_component.output,
                         "RGPUSM: About to insert in rgpusm cache addr=%p, size=%d", addr, (int)size);
-    while((rc = mpool->rcache->rcache_insert(mpool->rcache, (mca_mpool_base_registration_t *)rgpusm_reg,
-             mca_mpool_rgpusm_component.rcache_size_limit)) ==
-            OPAL_ERR_TEMP_OUT_OF_RESOURCE) {
-        opal_output(-1, "No room in the cache - boot one out");
-        if (!mca_mpool_rgpusm_deregister_lru(mpool)) {
-            break;
+    rc = mpool->rcache->rcache_insert(mpool->rcache, (mca_mpool_base_registration_t *)rgpusm_reg,
+                                      mca_mpool_rgpusm_component.rcache_size_limit);
+    if (OPAL_ERR_TEMP_OUT_OF_RESOURCE == rc) {
+        opal_output_verbose(40, mca_mpool_rgpusm_component.output,
+                            "RGPUSM: No room in the cache - boot the first one out");
+        (void)mca_mpool_rgpusm_deregister_lru(mpool);
+        if (mca_mpool_rgpusm_component.empty_cache) {
+            int remNum = 1;
+            /* Empty out every registration from LRU until it is empty */
+            opal_output_verbose(40, mca_mpool_rgpusm_component.output,
+                                "RGPUSM: About to delete all the unused entries in the cache");
+            while (mca_mpool_rgpusm_deregister_lru(mpool)) {
+                remNum++;
+            }
+            opal_output_verbose(40, mca_mpool_rgpusm_component.output,
+                                "RGPUSM: Deleted and deregistered %d entries", remNum);
+            rc = mpool->rcache->rcache_insert(mpool->rcache, (mca_mpool_base_registration_t *)rgpusm_reg,
+                                              mca_mpool_rgpusm_component.rcache_size_limit);
+        } else {
+            /* Check for room after one removal. If not, remove another one until there is space */
+            while((rc = mpool->rcache->rcache_insert(mpool->rcache, (mca_mpool_base_registration_t *)rgpusm_reg,
+                                                     mca_mpool_rgpusm_component.rcache_size_limit)) ==
+                  OPAL_ERR_TEMP_OUT_OF_RESOURCE) {
+                opal_output_verbose(40, mca_mpool_rgpusm_component.output,
+                                    "RGPUSM: No room in the cache - boot one out");
+                if (!mca_mpool_rgpusm_deregister_lru(mpool)) {
+                    break;
+                }
+            }
         }
     }