Merge pull request #2063 in HDFFV/hdf5 from ~NFORTNE2/hdf5_naf:vol_file_cmp to develop

* commit '40b2658b4b266b4dc65ec3db2897e0aaf54f8e42':
  Fix missing free in H5T__ref_mem_read()
  Fix bugs in H5VL file comparison code.  Add short circuit success to H5VL_cmp_connector_cls().
  Implement file comparison VOL callback.  Other changes to allow references to work with non-native connectors.  There is a bug somewhere.

5 files changed, 163 insertions, 34 deletions

diff --git a/src/H5Tref.c b/src/H5Tref.c
index 823def9..6f21363 100644
--- a/src/H5Tref.c
+++ b/src/H5Tref.c
@@ -345,10 +345,12 @@ static size_t
 H5T__ref_mem_getsize(H5VL_object_t H5_ATTR_UNUSED *src_file, const void *src_buf,
     size_t H5_ATTR_UNUSED src_size, H5VL_object_t *dst_file, hbool_t *dst_copy)
 {
-    H5F_t *src_f;
-    H5F_t *dst_f;
     H5VL_object_t *vol_obj = NULL;
     const H5R_ref_priv_t *src_ref = (const H5R_ref_priv_t *)src_buf;
+    hbool_t files_equal = FALSE;
+    char file_name_buf_static[256];
+    char *file_name_buf_dyn = NULL;
+    ssize_t file_name_len;
     unsigned flags = 0;
     size_t ret_value = 0;
 
@@ -361,27 +363,43 @@ H5T__ref_mem_getsize(H5VL_object_t H5_ATTR_UNUSED *src_file, const void *src_buf
     if(NULL == (vol_obj = H5VL_vol_object(src_ref->loc_id)))
         HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid location identifier")
 
-    /* We should assert here that the terminal connector is H5VL_NATIVE once
-     * there is a facility to do so -NAF 2019/10/30 */
-
-    /* Retrieve files from VOL objects */
-    if(NULL == (src_f = (H5F_t *)H5VL_object_data(vol_obj)))
-        HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid VOL object")
-    if(NULL == (dst_f = (H5F_t *)H5VL_object_data(dst_file)))
-        HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid VOL object")
-
     /* Set external flag if referenced file is not destination file */
-    flags |= (src_f->shared != dst_f->shared) ? H5R_IS_EXTERNAL : 0;
+    if(H5VL_file_specific(vol_obj, H5VL_FILE_IS_EQUAL, H5P_DATASET_XFER_DEFAULT, NULL, dst_file, &files_equal) < 0)
+        HGOTO_ERROR(H5E_REFERENCE, H5E_CANTCOMPARE, 0, "can't check if files are equal")
+    flags |= !files_equal ? H5R_IS_EXTERNAL : 0;
 
     /* Force re-calculating encoding size if any flags are set */
     if(flags || !src_ref->encode_size) {
         /* Pass the correct encoding version for the selection depending on the
          * file libver bounds, this is later retrieved in H5S hyper encode */
-        if(src_ref->type == (int8_t)H5R_DATASET_REGION2)
-            H5CX_set_libver_bounds(dst_f);
+        if(src_ref->type == (int8_t)H5R_DATASET_REGION2) {
+            /* Temporary hack to check if this is the native connector.  We need to
+             * add a way to check if the terminal connector is native.  For now this
+             * will break passthroughs, but it's needed for other VOL connectors to
+             * work.  -NAF */
+            if(dst_file->connector->cls->value == H5VL_NATIVE_VALUE) {
+                H5F_t *dst_f;
+
+                if(NULL == (dst_f = (H5F_t *)H5VL_object_data(dst_file)))
+                    HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid VOL object")
+                H5CX_set_libver_bounds(dst_f);
+            } /* end if */
+            else
+                H5CX_set_libver_bounds(NULL);
+        } /* end if */
+
+        /* Get file name */
+        if(H5VL_file_get(vol_obj, H5VL_FILE_GET_NAME, H5P_DATASET_XFER_DEFAULT, NULL, H5I_FILE, sizeof(file_name_buf_static), file_name_buf_static, &file_name_len) < 0)
+            HGOTO_ERROR(H5E_REFERENCE, H5E_CANTGET, 0, "can't get file name")
+        if(file_name_len >= (ssize_t)sizeof(file_name_buf_static)) {
+            if(NULL == (file_name_buf_dyn = (char *)H5MM_malloc((size_t)file_name_len + 1)))
+                HGOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, 0, "can't allocate space for file name")
+            if(H5VL_file_get(vol_obj, H5VL_FILE_GET_NAME, H5P_DATASET_XFER_DEFAULT, NULL, H5I_FILE, (size_t)file_name_len + 1, file_name_buf_dyn, &file_name_len) < 0)
+                HGOTO_ERROR(H5E_REFERENCE, H5E_CANTGET, 0, "can't get file name")
+        } /* end if */
 
         /* Determine encoding size */
-        if(H5R__encode(H5F_ACTUAL_NAME(src_f), src_ref, NULL, &ret_value, flags) < 0)
+        if(H5R__encode(file_name_buf_dyn ? file_name_buf_dyn : file_name_buf_static, src_ref, NULL, &ret_value, flags) < 0)
             HGOTO_ERROR(H5E_REFERENCE, H5E_CANTENCODE, 0, "unable to determine encoding size")
     } else {
         /* Can do a direct copy and skip blob decoding */
@@ -393,6 +411,8 @@ H5T__ref_mem_getsize(H5VL_object_t H5_ATTR_UNUSED *src_file, const void *src_buf
     }
 
 done:
+    H5MM_xfree(file_name_buf_dyn);
+
     FUNC_LEAVE_NOAPI(ret_value)
 }   /* end H5T__ref_mem_getsize() */
 
@@ -411,10 +431,12 @@ H5T__ref_mem_read(H5VL_object_t H5_ATTR_UNUSED *src_file, const void *src_buf,
     size_t H5_ATTR_UNUSED src_size, H5VL_object_t *dst_file, void *dst_buf,
     size_t dst_size)
 {
-    H5F_t *src_f;
-    H5F_t *dst_f;
     H5VL_object_t *vol_obj = NULL;
     const H5R_ref_priv_t *src_ref = (const H5R_ref_priv_t *)src_buf;
+    hbool_t files_equal = FALSE;
+    char file_name_buf_static[256];
+    char *file_name_buf_dyn = NULL;
+    ssize_t file_name_len;
     unsigned flags = 0;
     herr_t ret_value = SUCCEED;
 
@@ -430,28 +452,46 @@ H5T__ref_mem_read(H5VL_object_t H5_ATTR_UNUSED *src_file, const void *src_buf,
     if(NULL == (vol_obj = H5VL_vol_object(src_ref->loc_id)))
         HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid location identifier")
 
-    /* We should assert here that the terminal connector is H5VL_NATIVE once
-     * there is a facility to do so -NAF 2019/10/30 */
-
-    /* Retrieve files from VOL objects */
-    if(NULL == (src_f = (H5F_t *)H5VL_object_data(vol_obj)))
-        HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid VOL object")
-    if(NULL == (dst_f = (H5F_t *)H5VL_object_data(dst_file)))
-        HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid VOL object")
-
     /* Set external flag if referenced file is not destination file */
-    flags |= (src_f->shared != dst_f->shared) ? H5R_IS_EXTERNAL : 0;
+    if(H5VL_file_specific(vol_obj, H5VL_FILE_IS_EQUAL, H5P_DATASET_XFER_DEFAULT, NULL, dst_file, &files_equal) < 0)
+        HGOTO_ERROR(H5E_REFERENCE, H5E_CANTCOMPARE, FAIL, "can't check if files are equal")
+    flags |= !files_equal ? H5R_IS_EXTERNAL : 0;
 
     /* Pass the correct encoding version for the selection depending on the
      * file libver bounds, this is later retrieved in H5S hyper encode */
-    if(src_ref->type == (int8_t)H5R_DATASET_REGION2)
-        H5CX_set_libver_bounds(dst_f);
+    if(src_ref->type == (int8_t)H5R_DATASET_REGION2) {
+        /* Temporary hack to check if this is the native connector.  We need to
+         * add a way to check if the terminal connector is native.  For now this
+         * will break passthroughs, but it's needed for other VOL connectors to
+         * work.  -NAF */
+        if(dst_file->connector->cls->value == H5VL_NATIVE_VALUE) {
+            H5F_t *dst_f;
+
+            if(NULL == (dst_f = (H5F_t *)H5VL_object_data(dst_file)))
+                HGOTO_ERROR(H5E_ARGS, H5E_BADTYPE, 0, "invalid VOL object")
+            H5CX_set_libver_bounds(dst_f);
+        } /* end if */
+        else
+            H5CX_set_libver_bounds(NULL);
+    } /* end if */
+
+    /* Get file name */
+    if(H5VL_file_get(vol_obj, H5VL_FILE_GET_NAME, H5P_DATASET_XFER_DEFAULT, NULL, H5I_FILE, sizeof(file_name_buf_static), file_name_buf_static, &file_name_len) < 0)
+        HGOTO_ERROR(H5E_REFERENCE, H5E_CANTGET, 0, "can't get file name")
+    if(file_name_len >= (ssize_t)sizeof(file_name_buf_static)) {
+        if(NULL == (file_name_buf_dyn = (char *)H5MM_malloc((size_t)file_name_len + 1)))
+            HGOTO_ERROR(H5E_RESOURCE, H5E_CANTALLOC, 0, "can't allocate space for file name")
+        if(H5VL_file_get(vol_obj, H5VL_FILE_GET_NAME, H5P_DATASET_XFER_DEFAULT, NULL, H5I_FILE, (size_t)file_name_len + 1, file_name_buf_dyn, &file_name_len) < 0)
+            HGOTO_ERROR(H5E_REFERENCE, H5E_CANTGET, 0, "can't get file name")
+    } /* end if */
 
     /* Encode reference */
-    if(H5R__encode(H5F_ACTUAL_NAME(src_f), src_ref, (unsigned char *)dst_buf, &dst_size, flags) < 0)
+    if(H5R__encode(file_name_buf_dyn ? file_name_buf_dyn : file_name_buf_static, src_ref, (unsigned char *)dst_buf, &dst_size, flags) < 0)
         HGOTO_ERROR(H5E_REFERENCE, H5E_CANTENCODE, FAIL, "Cannot encode reference")
 
 done:
+    H5MM_xfree(file_name_buf_dyn);
+
     FUNC_LEAVE_NOAPI(ret_value)
 }   /* end H5T__ref_mem_read() */
 
@@ -1067,3 +1107,4 @@ H5T_ref_reclaim(void *elem, const H5T_t *dt)
 done:
     FUNC_LEAVE_NOAPI(ret_value)
 } /* end H5T_ref_reclaim() */
+
diff --git a/src/H5VLcallback.c b/src/H5VLcallback.c
index 8d7368b..77df207 100644
--- a/src/H5VLcallback.c
+++ b/src/H5VLcallback.c
@@ -101,6 +101,8 @@ static void * H5VL__file_open(const H5VL_class_t *cls, const char *name,
     unsigned flags, hid_t fapl_id, hid_t dxpl_id, void **req);
 static herr_t H5VL__file_get(void *obj, const H5VL_class_t *cls, H5VL_file_get_t get_type, 
     hid_t dxpl_id, void **req, va_list arguments);
+static herr_t H5VL__file_specific_wrap_va_list(void *obj, const H5VL_class_t *cls,
+    H5VL_file_specific_t specific_type,  hid_t dxpl_id, void **req, ...);
 static herr_t H5VL__file_specific(void *obj, const H5VL_class_t *cls, H5VL_file_specific_t specific_type, 
     hid_t dxpl_id, void **req, va_list arguments);
 static herr_t H5VL__file_optional(void *obj, const H5VL_class_t *cls, hid_t dxpl_id,
@@ -2989,6 +2991,43 @@ done:
 
 
 /*-------------------------------------------------------------------------
+ * Function:	H5VL__file_specific_wrap_va_list
+ *
+ * Purpose:     Perform File specific operations through the VOL.  Just
+ *              starts a va_list and passes it to the connector's
+ *              callback.  Needed when the VOL layer needs to replace one
+ *              of the variable arguments.
+ *
+ * Return:      Success:    Non-negative
+ *              Failure:    Negative
+ *
+ *-------------------------------------------------------------------------
+ */
+static herr_t
+H5VL__file_specific_wrap_va_list(void *obj, const H5VL_class_t *cls,
+    H5VL_file_specific_t specific_type,  hid_t dxpl_id, void **req, ...)
+{
+    va_list arguments;                  /* Argument list passed from the API call */
+    herr_t ret_value = SUCCEED;         /* Return value */
+
+    FUNC_ENTER_STATIC
+
+    /* Start access to the varargs, so they are available in all situations below */
+    HDva_start(arguments, req);
+
+    /* Call the corresponding VOL callback */
+    if((cls->file_cls.specific)(obj, specific_type, dxpl_id, req, arguments) < 0)
+        HGOTO_ERROR(H5E_VOL, H5E_CANTOPERATE, FAIL, "file specific failed")
+
+done:
+    /* End access to the va_list */
+    HDva_end(arguments);
+
+    FUNC_LEAVE_NOAPI(ret_value)
+} /* end H5VL__file_specific_wrap_va_list() */
+
+
+/*-------------------------------------------------------------------------
  * Function:	H5VL__file_specific
  *
  * Purpose:	Perform File specific operations through the VOL
@@ -3010,9 +3049,38 @@ H5VL__file_specific(void *obj, const H5VL_class_t *cls, H5VL_file_specific_t spe
     if(NULL == cls->file_cls.specific)
         HGOTO_ERROR(H5E_VOL, H5E_UNSUPPORTED, FAIL, "VOL connector has no 'file specific' method")
 
-    /* Call the corresponding VOL callback */
-    if((cls->file_cls.specific)(obj, specific_type, dxpl_id, req, arguments) < 0)
-        HGOTO_ERROR(H5E_VOL, H5E_CANTOPERATE, FAIL, "file specific failed")
+    /* Special handling for file is equal */
+    if(specific_type == H5VL_FILE_IS_EQUAL) {
+        va_list             tmp_args;       /* Argument list passed from the API call */
+        H5VL_object_t       *vol_obj2;      /* Second VOL object */
+        hbool_t             *is_equal;      /* Output variable */
+        int                 cmp_value;      /* Comparison result */
+
+        /* Get parameters */
+        HDva_copy(tmp_args, arguments);
+        vol_obj2 = HDva_arg(tmp_args, H5VL_object_t *);
+        is_equal = HDva_arg(tmp_args, hbool_t *);
+        HDva_end(tmp_args);
+
+        HDassert(vol_obj2);
+
+        /* Compare connector classes */
+        if(H5VL_cmp_connector_cls(&cmp_value, cls, vol_obj2->connector->cls) < 0)
+            HGOTO_ERROR(H5E_VOL, H5E_CANTCOMPARE, FAIL, "can't compare connector classes")
+
+        /* If the classes are different the files are different */
+        if(cmp_value)
+            *is_equal = FALSE;
+        else
+            /* Make callback (need to extract data from vol_obj2 and redo the
+             * va_list) */
+            if(H5VL__file_specific_wrap_va_list(obj, cls, specific_type, dxpl_id, req, vol_obj2->data, is_equal) < 0)
+                HGOTO_ERROR(H5E_VOL, H5E_CANTOPERATE, FAIL, "file specific failed")
+    } /* end if */
+    else
+        /* Call the corresponding VOL callback */
+        if((cls->file_cls.specific)(obj, specific_type, dxpl_id, req, arguments) < 0)
+            HGOTO_ERROR(H5E_VOL, H5E_CANTOPERATE, FAIL, "file specific failed")
 
 done:
     FUNC_LEAVE_NOAPI(ret_value)
diff --git a/src/H5VLconnector.h b/src/H5VLconnector.h
index 0012eeb..f0925bf 100644
--- a/src/H5VLconnector.h
+++ b/src/H5VLconnector.h
@@ -122,7 +122,8 @@ typedef enum H5VL_file_specific_t {
     H5VL_FILE_MOUNT,                        /* Mount a file                     */
     H5VL_FILE_UNMOUNT,                      /* Unmount a file                   */
     H5VL_FILE_IS_ACCESSIBLE,                /* Check if a file is accessible    */
-    H5VL_FILE_DELETE                        /* Delete a file                    */
+    H5VL_FILE_DELETE,                       /* Delete a file                    */
+    H5VL_FILE_IS_EQUAL                      /* Check if two files are the same  */
 } H5VL_file_specific_t;
 
 /* types for group GET callback */
diff --git a/src/H5VLint.c b/src/H5VLint.c
index 733a2b5..284b266 100644
--- a/src/H5VLint.c
+++ b/src/H5VLint.c
@@ -1639,6 +1639,12 @@ H5VL_cmp_connector_cls(int *cmp_value, const H5VL_class_t *cls1, const H5VL_clas
     HDassert(cls1);
     HDassert(cls2);
 
+    /* If the pointers are the same the classes are the same */
+    if(cls1 == cls2) {
+        *cmp_value = 0;
+        HGOTO_DONE(SUCCEED);
+    } /* end if */
+
     /* Compare connector "values" */
     if(cls1->value < cls2->value) {
         *cmp_value = -1;
diff --git a/src/H5VLnative_file.c b/src/H5VLnative_file.c
index 3bda0b8..588d59f 100644
--- a/src/H5VLnative_file.c
+++ b/src/H5VLnative_file.c
@@ -418,6 +418,19 @@ H5VL__native_file_specific(void *obj, H5VL_file_specific_t specific_type,
                 break;
             }
 
+        /* Check if two files are the same */
+        case H5VL_FILE_IS_EQUAL:
+            {
+                H5F_t *file2 = (H5F_t *)HDva_arg(arguments, void *);
+                hbool_t *is_equal = HDva_arg(arguments, hbool_t *);
+
+                if(!obj || !file2)
+                    *is_equal = FALSE;
+                else
+                    *is_equal = (((H5F_t *)obj)->shared == file2->shared);
+                break;
+            }
+
         default:
             HGOTO_ERROR(H5E_VOL, H5E_UNSUPPORTED, FAIL, "invalid specific operation")
     } /* end switch */