ompi/oshmem/spml/ucx: Fix cleanup when disconnected: avoid double free

Signed-off-by: Ilia Yastrebov <[email protected]>
open-mpi · Feb 6, 2024 · 3071892 · 3071892
1 parent bcac86f
commit 3071892
Show file tree

Hide file tree

Showing 2 changed files with 7 additions and 3 deletions.
diff --git a/oshmem/mca/spml/ucx/spml_ucx.h b/oshmem/mca/spml/ucx/spml_ucx.h
@@ -43,6 +43,7 @@ BEGIN_C_DECLS
 
 #define SPML_UCX_ASSERT  MCA_COMMON_UCX_ASSERT
 #define SPML_UCX_ERROR   MCA_COMMON_UCX_ERROR
+#define SPML_UCX_WARN    MCA_COMMON_UCX_WARN
 #define SPML_UCX_VERBOSE MCA_COMMON_UCX_VERBOSE
 #define SPML_UCX_TRANSP_IDX 0
 #define SPML_UCX_TRANSP_CNT 1

diff --git a/oshmem/mca/spml/ucx/spml_ucx_component.c b/oshmem/mca/spml/ucx/spml_ucx_component.c
@@ -458,7 +458,8 @@ static void mca_spml_ucx_ctx_fini(mca_spml_ucx_ctx_t *ctx)
 
 static int mca_spml_ucx_component_fini(void)
 {
-    int fenced = 0, i;
+    volatile int fenced = 0;
+    int i;
     int ret = OSHMEM_SUCCESS;
     mca_spml_ucx_ctx_t *ctx;
 
@@ -491,8 +492,10 @@ static int mca_spml_ucx_component_fini(void)
 
 
     ret = opal_common_ucx_mca_pmix_fence_nb(&fenced);
-    if (OPAL_SUCCESS != ret) {
-        return ret;
+    if (ret != PMIX_SUCCESS) {
+        SPML_UCX_WARN("pmix fence failed: %s", PMIx_Error_string(ret));
+        /* In case of pmix fence failure just continue cleanup */
+        fenced = 1;
     }
 
     while (!fenced) {