[svn-r13129] More refactoring. Moved index versions to the index header, so that an

index can be read all at once.  This changes the file format!

Tested on Windows, kagiso, and smirom.

7 files changed, 29 insertions, 45 deletions

diff --git a/src/H5D.c b/src/H5D.c
index ac95d08..090b0c3 100644
--- a/src/H5D.c
+++ b/src/H5D.c
@@ -1195,7 +1195,7 @@ H5D_update_entry_info(H5F_t *file, hid_t dxpl_id, H5D_t *dset)
     /* fill value variables */
     H5D_fill_time_t	fill_time;
     H5O_fill_t		*fill_prop;     /* Pointer to dataset's fill value information */
-    H5O_fill_new_t      fill;
+    H5O_fill_new_t      fill = {NULL, 0, NULL, H5D_ALLOC_TIME_LATE, H5D_FILL_TIME_ALLOC, TRUE};
     H5D_fill_value_t	fill_status;
 
     struct H5O_t       *oh = NULL;      /* Pointer to dataset's object header */
@@ -1218,12 +1218,6 @@ H5D_update_entry_info(H5F_t *file, hid_t dxpl_id, H5D_t *dset)
     /* Get the file's 'use the latest version of the format' flag */
     use_latest_format = H5F_USE_LATEST_FORMAT(file);
 
-    /* Initialize the fill value message */
-    HDmemset(&fill,0,sizeof(H5O_fill_new_t));
-    fill.alloc_time = H5D_ALLOC_TIME_LATE;
-    fill.fill_time = H5D_FILL_TIME_ALLOC;
-    fill.fill_defined = TRUE;
-
     /* Point at dataset's copy, to cache it for later */
     fill_prop = &dset->shared->fill;
     fill_time = dset->shared->fill_time;
@@ -1864,7 +1858,7 @@ done:
 static herr_t
 H5D_open_oid(H5D_t *dataset, hid_t dxpl_id)
 {
-    H5O_fill_new_t  fill;
+    H5O_fill_new_t  fill = {NULL, 0, NULL, H5D_ALLOC_TIME_LATE, H5D_FILL_TIME_IFSET, TRUE};
     unsigned    alloc_time_state;       /* Allocation time state */
     H5O_fill_t     *fill_prop;          /* Pointer to dataset's fill value area */
     H5O_pline_t  pline;                 /* I/O pipeline information */
@@ -1880,12 +1874,6 @@ H5D_open_oid(H5D_t *dataset, hid_t dxpl_id)
     if(NULL == (dataset->shared = H5D_new(H5P_DATASET_CREATE_DEFAULT, FALSE, FALSE)))
         HGOTO_ERROR (H5E_RESOURCE, H5E_NOSPACE, FAIL, "memory allocation failed")
 
-    /* Initialize the fill value message JAMES: initialize above instead? */
-    HDmemset(&fill,0,sizeof(H5O_fill_new_t));
-    fill.alloc_time = H5D_ALLOC_TIME_LATE;
-    fill.fill_time = H5D_FILL_TIME_IFSET;
-    fill.fill_defined = TRUE;
-
     /* Open the dataset object */
     if(H5O_open(&(dataset->oloc)) < 0)
         HGOTO_ERROR(H5E_DATASET, H5E_CANTOPENOBJ, FAIL, "unable to open")
diff --git a/src/H5Oalloc.c b/src/H5Oalloc.c
index 1368e84..b1a81aa 100644
--- a/src/H5Oalloc.c
+++ b/src/H5Oalloc.c
@@ -670,11 +670,11 @@ H5O_alloc_new_chunk(H5F_t *f,
                    (found_attr < 0 ||
                     oh->mesg[u].raw_size < oh->mesg[found_attr].raw_size))
                 found_attr = u;
-/*	} else if(H5O_LINK_ID == msg_id) {
+	} else if(H5O_LINK_ID == msg_id) {
             if(oh->mesg[u].raw_size >= cont_size &&
                    (found_link < 0 ||
                     oh->mesg[u].raw_size < oh->mesg[found_link].raw_size))
-                found_link = u; JAMES */
+                found_link = u;
 	} else {
             if(oh->mesg[u].raw_size >= cont_size &&
                    (found_other < 0 ||
diff --git a/src/H5Oprivate.h b/src/H5Oprivate.h
index 392b223..1a42569 100644
--- a/src/H5Oprivate.h
+++ b/src/H5Oprivate.h
@@ -45,10 +45,9 @@
 typedef struct H5O_msg_class_t H5O_msg_class_t;
 typedef struct H5O_t H5O_t;
 
-/* JAMES: should these be in H5SM_private?  or renamed? */
 /* Fractal heap ID type for shared message heap IDs.  The length of a heap ID
- * depends on how the heap is configured; currently they're always stored in
- * 8-byte fields, although only seven bytes are used.
+ * depends on how the heap is configured; currently they're seven bytes long
+ * but are stored in 8-byte fields in memory.
  */
 #define H5SM_FHEAP_ID_LEN 7
 typedef uint64_t H5SM_fheap_id_t;
diff --git a/src/H5Oshared.c b/src/H5Oshared.c
index 68d3494..fa6a7ef 100644
--- a/src/H5Oshared.c
+++ b/src/H5Oshared.c
@@ -201,6 +201,12 @@ done:
  * Purpose:	Changes the link count for the object referenced by a shared
  *              message.
  *
+ *              This function changes the object header link count and is
+ *              only relevant for committed messages.  Messages shared in
+ *              the heap are re-shared each time they're written, so their
+ *              reference count is stored in the file-wide shared message
+ *              index and is changed in a different place in the code.
+ *
  * Return:	Success:	New link count
  *
  *		Failure:	Negative
@@ -546,7 +552,6 @@ done:
  *
  *-------------------------------------------------------------------------
  */
-/* JAMES: this is where shared messages increment their links */
 static herr_t
 H5O_shared_link(H5F_t *f, hid_t dxpl_id, const void *_mesg)
 {
@@ -727,8 +732,6 @@ H5O_shared_debug (H5F_t UNUSED *f, hid_t UNUSED dxpl_id, const void *_mesg,
 
     FUNC_ENTER_NOAPI_NOINIT_NOFUNC(H5O_shared_debug)
 
-    /* JAMES_HEAP: this oughta change, too, of course. */
-
     /* Check args */
     HDassert(f);
     HDassert(mesg);
diff --git a/src/H5SM.c b/src/H5SM.c
index 4a6e106..e5e28e7 100755
--- a/src/H5SM.c
+++ b/src/H5SM.c
@@ -696,7 +696,6 @@ H5SM_convert_list_to_btree(H5F_t * f, H5SM_index_header_t * header,
         if(list->messages[x].ref_count > 0)
         {
             key.message = list->messages[x];
-            /* JAMES: need ref count! And test, if not having refcount doesn't break any tests. */
 
             if(H5B2_insert(f, dxpl_id, H5SM_INDEX, tree_addr, &key) < 0)
                 HGOTO_ERROR(H5E_BTREE, H5E_CANTINSERT, FAIL, "couldn't add SOHM to B-tree")
@@ -955,7 +954,6 @@ H5SM_write_mesg(H5F_t *f, hid_t dxpl_id, H5SM_index_header_t *header,
 
         /* JAMES: not very efficient (gets hash value twice, searches list twice).  Refactor. */
         /* See if the message is already in the index and get its location */
-        /* JAMES: should return a pointer to the message */
         list_pos = H5SM_find_in_list(list, &key);
         if(list_pos != UFAIL)
         {
@@ -981,16 +979,11 @@ H5SM_write_mesg(H5F_t *f, hid_t dxpl_id, H5SM_index_header_t *header,
     if(!found)
     {
         hsize_t     x;                /* Counter variable */
-        size_t      mesg_size;        /* Size of the message on disk */
 
         /* JAMES: wrap this in a function call? */
 
-        /* Encode the message and get its size */ /* JAMES: already have this */
-        if((mesg_size = H5O_msg_raw_size(f, type_id, mesg)) == 0)
-	    HGOTO_ERROR(H5E_OHDR, H5E_BADMESG, FAIL, "unable to get size of message")
-
         /* Put the message in the heap and record its new heap ID */
-        if(H5HF_insert(fheap, dxpl_id, mesg_size, key.encoding, &shared.u.heap_id) < 0)
+        if(H5HF_insert(fheap, dxpl_id, key.encoding_size, key.encoding, &shared.u.heap_id) < 0)
             HGOTO_ERROR(H5E_HEAP, H5E_CANTINSERT, FAIL, "unable to insert message into fractal heap")
 
         key.message.fheap_id = shared.u.heap_id;
@@ -1044,7 +1037,7 @@ done:
 	HDONE_ERROR(H5E_CACHE, H5E_CANTUNPROTECT, FAIL, "unable to close SOHM index")
 
     if(encoding_buf)
-        H5MM_free(encoding_buf);
+        encoding_buf = H5MM_xfree(encoding_buf);
 
     FUNC_LEAVE_NOAPI(ret_value)
 } /* end H5SM_write_mesg() */
@@ -1129,7 +1122,7 @@ done:
 
     /* Free buf */
     if(mesg_buf)
-        H5MM_xfree(mesg_buf);
+        mesg_buf = H5MM_xfree(mesg_buf);
 
     FUNC_LEAVE_NOAPI(ret_value)
 } /* end H5SM_try_delete() */
@@ -1368,6 +1361,7 @@ done:
     /* Free the serialized message buffer on error */
     if(ret_value < 0 && *encoded_mesg)
         *encoded_mesg = H5MM_xfree(*encoded_mesg);
+
     FUNC_LEAVE_NOAPI(ret_value)
 } /* end H5SM_delete_from_index() */
 
diff --git a/src/H5SMcache.c b/src/H5SMcache.c
index 7f1c282..c79c7d6 100644
--- a/src/H5SMcache.c
+++ b/src/H5SMcache.c
@@ -138,6 +138,7 @@ H5SM_flush_table(H5F_t *f, hid_t dxpl_id, hbool_t destroy, haddr_t addr, H5SM_ma
 
         /* Encode each index header */
         for(x=0; x<table->num_indexes; ++x) {
+            *p++ = H5SM_LIST_VERSION;   /* Encode version for this list. */
             *p++ = table->indexes[x].index_type;      /* Is message index a list or a B-tree? */
 
             UINT16ENCODE(p, table->indexes[x].mesg_types);    /* Type of messages in the index */
@@ -248,8 +249,10 @@ H5SM_load_table(H5F_t *f, hid_t dxpl_id, haddr_t addr, const void UNUSED *udata1
 
     /* Read in the index headers */
     for(x=0; x<table->num_indexes; ++x) {
-        table->indexes[x].index_type= *p++;  /* type of the index (list or B-tree) */
+        if (H5SM_LIST_VERSION != *p++)
+            HGOTO_ERROR(H5E_FILE, H5E_CANTOPENFILE, NULL, "bad shared message list version number")
 
+        table->indexes[x].index_type= *p++;  /* type of the index (list or B-tree) */
         UINT16DECODE(p, table->indexes[x].mesg_types);
         UINT32DECODE(p, table->indexes[x].min_mesg_size);
         UINT16DECODE(p, table->indexes[x].list_max);
@@ -410,6 +413,7 @@ H5SM_flush_list(H5F_t *f, hid_t dxpl_id, hbool_t destroy, haddr_t addr, H5SM_lis
         size_t	size;               /* Header size on disk */
         uint32_t computed_chksum;   /* Computed metadata checksum value */
         hsize_t x;
+        hsize_t mesgs_written;
 
         /* JAMES: consider only writing as many messages as necessary, and then adding a
          * blank "end of list" message or something?
@@ -423,21 +427,21 @@ H5SM_flush_list(H5F_t *f, hid_t dxpl_id, hbool_t destroy, haddr_t addr, H5SM_lis
         HDmemcpy(p, H5SM_LIST_MAGIC, (size_t)H5SM_LIST_SIZEOF_MAGIC);
         p += H5SM_LIST_SIZEOF_MAGIC;
 
-        /* Encode version */
-        *p++ = H5SM_LIST_VERSION;
-
         /* Write messages from the messages array to disk */
-        /* JAMES: we have to search the whole array.  not the best way to do it; could go until we've written
-         * num_messages */
-        for(x=0; x<list->header->list_max; x++) {
+        mesgs_written = 0;
+        for(x=0; x<list->header->list_max && mesgs_written < list->header->num_messages; x++) {
             if(list->messages[x].ref_count > 0) {
               /* JAMES: use H5SM_message_encode here */
               UINT32ENCODE(p, list->messages[x].hash);  /* Read the hash value for this message */
               UINT32ENCODE(p, list->messages[x].ref_count);  /* Read the reference count for this message */
               UINT64ENCODE(p, list->messages[x].fheap_id); /* Get the heap ID for the message */
+
+              ++mesgs_written;
             }
         }
 
+        HDassert(mesgs_written == list->header->num_messages);
+
         /* Compute checksum on buffer */
         computed_chksum = H5_checksum_metadata(buf, (size - H5SM_SIZEOF_CHECKSUM), 0);
         UINT32ENCODE(p, computed_chksum);
@@ -518,10 +522,6 @@ H5SM_load_list(H5F_t *f, hid_t dxpl_id, haddr_t addr, const void UNUSED *udata1,
 	HGOTO_ERROR(H5E_SOHM, H5E_CANTLOAD, NULL, "bad SOHM list signature");
     p += H5SM_LIST_SIZEOF_MAGIC;
 
-    /* Check version JAMES: should be in master table, not list */
-    if (H5SM_LIST_VERSION != *p++)
-        HGOTO_ERROR(H5E_FILE, H5E_CANTOPENFILE, NULL, "wrong shared message list version number")
-
     /* Read messages into the list array */
     for(x=0; x<header->num_messages; x++)
     {
diff --git a/src/H5SMpkg.h b/src/H5SMpkg.h
index 6416a1b..b771893 100755
--- a/src/H5SMpkg.h
+++ b/src/H5SMpkg.h
@@ -41,13 +41,14 @@
 
 #define H5SM_SOHM_ENTRY_SIZE(f) (4  /* Hash value */                         \
          + 4                        /* reference count*/                     \
-         + 8)                       /* JAMES: size of heap ID on disk */
+         + sizeof(H5SM_fheap_id_t)) /* size of heap ID on disk */
 
 #define H5SM_TABLE_SIZE(f) ( H5SM_TABLE_SIZEOF_MAGIC                         \
          + 1                                   /* Table version */           \
          + H5SM_SIZEOF_CHECKSUM)               /* Checksum */
 
 #define H5SM_INDEX_HEADER_SIZE(f) (1 /* Whether index is a list or B-tree */ \
+         + 1         /* Version of index format */                           \
          + 2         /* Type of messages stored in the index */              \
          + 4         /* Minimum size of messages to share */                 \
          + (3 * 2)   /* B-tree cutoff, list cutoff, # of shared messages */  \
@@ -55,7 +56,6 @@
          + H5F_SIZEOF_ADDR(f)) /* Address of heap */
 
 #define H5SM_LIST_SIZE(f, num_mesg) H5SM_LIST_SIZEOF_MAGIC                   \
-         + 1        /* List version */                                       \
          + (H5SM_SOHM_ENTRY_SIZE(f) * num_mesg)                              \
          + H5SM_SIZEOF_CHECKSUM                /* Checksum */