Modified page buffer to split entries only where necessary -- specifically

when handling an I/O request on a metadata entry that has been sub-allocated
from a larger file space allocation (i.e. fixed and extensible array), and
that crosses at least one page boundary.  .

This required modifying the metadata cache to provide the type of the
metadata cache entry in the current I/O request.  For now, this is done
with a function call.  Once we are sure this works, it may be appropriate
to convert this to a macro, or to add a flags parameter to the H5F block
read/write calls.

Also updated the metadata cache to report whether a read request is
speculative -- again via a function call.  This allowed me to remove
the last address static variable in the H5PB_read() call, which is
necessary to support multiple files opened in VFD SWMR mode.

Also re-wrote the H5PB_remove_entries() call to handle release
of large metadata file space allocations that have been sub-allocated
into multiple metadata entries.  Also modified the call to
H5PB_remove_entries() in H5MF__xfree_impl() to invoke it whenever
the page buffer is enabled and the size of the space to be freed is
of page size or larger.

Tested serial / debug on charis and Jelly.

Found a bug in H5MF_xfree_impl(), in which the call to H5PB_remove_entries()
is skipped due to HGOTO_DONE calls earlier in the function.  While the
obvious action is to move the call earlier in the function, best to
consult with Vailin first, as there is much going on and it would be
best to avoid making the situation worse.  If nothing else, there are
some error management issues.

1 files changed, 29 insertions, 0 deletions

diff --git a/src/H5Cimage.c b/src/H5Cimage.c
index ee286d9..9a6d667 100644
--- a/src/H5Cimage.c
+++ b/src/H5Cimage.c
@@ -1058,6 +1058,22 @@ H5C__read_cache_image(H5F_t *f, H5C_t *cache_ptr)
 #endif /* H5_HAVE_PARALLEL */
 
 	/* Read the buffer (if serial access, or rank 0 of parallel access) */
+
+        /* No need to set the page buffer hints here, as if paged
+         * allocation is in use, we know that the cache image was allocated
+         * directly from the free space manager, and thus either doesn't 
+         * cross page boundaries, or is page aligned.  Between this,
+         * and the fact that the cache image is never read speculatively,
+         * the page buffer should never request hints in this context.
+         * 
+         * If for some reason it does, the NULL curr_io_type will trigger
+         * an assertion failure.
+         *
+         * Note that we will have to revisit this if we ever use 
+         * cache_ptr->curr_io_type for something other than sanity
+         * checking
+         *                                      JRM -- 3/30/20
+         */
         if(H5F_block_read(f, H5FD_MEM_SUPER, cache_ptr->image_addr,
                 cache_ptr->image_len, cache_ptr->image_buffer) < 0)
             HGOTO_ERROR(H5E_CACHE, H5E_READERROR, FAIL, "Can't read metadata cache image block")
@@ -3554,6 +3570,19 @@ H5C__write_cache_image(H5F_t *f, const H5C_t *cache_ptr)
 #endif /* H5_HAVE_PARALLEL */
 
 	/* Write the buffer (if serial access, or rank 0 for parallel access) */
+
+        /* No need to set the page buffer hints here.
+         *
+         * If paged allocation is in use, we know that the cache image 
+         * was allocated directly from the free space manager, and thus 
+         * either doesn't cross page boundaries, or is page aligned.  
+         * Thus it should never trigger the sanity checks in the page buffer.
+         * 
+         * If for some reason it does, the NULL curr_io_type will trigger
+         * an assertion failure.
+         *
+         *                                      JRM -- 3/30/20
+         */
 	if(H5F_block_write(f, H5FD_MEM_SUPER, cache_ptr->image_addr, cache_ptr->image_len, cache_ptr->image_buffer) < 0)
             HGOTO_ERROR(H5E_CACHE, H5E_CANTFLUSH, FAIL, "can't write metadata cache image block to file")
 #ifdef H5_HAVE_PARALLEL