[svn-r22281] Purpose:

  Fix for HDFFV-7993 - h5repack fails with error "chunk size must be <= maximum dimension size for fixed-sized dimensions"

Description:
  Fixed a failure when change the chunk size of a specified chunked dataset with unlimited max dims.
  Also took care of converting to contiguous and compact from the dataset.
  Test cases were added and tagged with jira#.
  Merged from HDF5 trunk r22277.

Tested:
  jam (linux32-LE), koala (linux64-LE), ostrich (linuxppc64-BE), tejeda (mac32-LE), Windows (32-LE cmake), Cmake (jam)

9 files changed, 232 insertions, 23 deletions

diff --git a/tools/h5repack/CMakeLists.txt b/tools/h5repack/CMakeLists.txt
index ae7a658..98d5c5a 100644
--- a/tools/h5repack/CMakeLists.txt
+++ b/tools/h5repack/CMakeLists.txt
@@ -96,6 +96,7 @@ IF (BUILD_TESTING)
       ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_layout.h5
       ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_layouto.h5
       ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_layout2.h5
+      ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_layout3.h5
       ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_named_dtypes.h5
       ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_nbit.h5
       ${HDF5_TOOLS_H5REPACK_SOURCE_DIR}/testfiles/h5repack_objs.h5
@@ -322,10 +323,12 @@ IF (BUILD_TESTING)
           -E remove 
          ./testfiles/h5repack_filters.h5.out
          ./testfiles/h5repack_filters.h5.out.err
-         ./testfiles/h5repack_layout2.h5-v.out
-         ./testfiles/h5repack_layout2.h5-v.out.err
          ./testfiles/h5repack_layout.h5-v.out
          ./testfiles/h5repack_layout.h5-v.out.err
+         ./testfiles/h5repack_layout2.h5-v.out
+         ./testfiles/h5repack_layout2.h5-v.out.err
+         ./testfiles/h5repack_layout3.h5-v.out
+         ./testfiles/h5repack_layout3.h5-v.out.err
          ./testfiles/out.tfamily%05d.h5
          ./testfiles/out.h5diff_attr1.h5
          ./testfiles/out.h5repack_attr.h5
@@ -336,9 +339,10 @@ IF (BUILD_TESTING)
          ./testfiles/out.h5repack_filters.h5
          ./testfiles/out.h5repack_fletcher.h5
          ./testfiles/out.h5repack_hlink.h5
-         ./testfiles/out.h5repack_layout2.h5
          ./testfiles/out.h5repack_layout.h5
          ./testfiles/out.h5repack_layouto.h5
+         ./testfiles/out.h5repack_layout2.h5
+         ./testfiles/out.h5repack_layout3.h5
          ./testfiles/out.h5repack_named_dtypes.h5
          ./testfiles/out.h5repack_nbit.h5
          ./testfiles/out.h5repack_objs.h5
@@ -369,9 +373,10 @@ IF (BUILD_TESTING)
          h5repack_fletcher_out.h5
          h5repack_hlink.h5
          h5repack_hlink_out.h5
-         h5repack_layout2.h5
          h5repack_layout.h5
          h5repack_layout_out.h5
+         h5repack_layout2.h5
+         h5repack_layout3.h5
          h5repack_named_dtypes.h5
          h5repack_named_dtypes_out.h5
          h5repack_nbit.h5
@@ -777,6 +782,24 @@ IF (BUILD_TESTING)
   ADD_H5_TEST (contig_small_fixed_compa "TEST" ${FILE18} -l chunked_small_fixed:COMPA)
   ADD_H5_VERIFY_TEST (contig_small_fixed_compa "TEST" 0 ${FILE18} chunked_small_fixed COMPACT)
 
+#--------------------------------------------------------------------------- 
+# Test file contains chunked datasets (need multiple dsets) with 
+# unlimited max dims.   (HDFFV-7933)
+# Use first dset to test.
+#---------------------------------------------------------------------------
+# chunk to chunk - specify chunk dim bigger than any current dim
+ADD_H5_TEST (chunk2chunk "TEST" h5repack_layout3.h5 -l chunk_unlimit1:CHUNK=100x300)
+ADD_H5_VERIFY_TEST (chunk2chunk "TEST" 0 h5repack_layout3.h5 chunk_unlimit1 CHUNK)
+
+# chunk to contiguous 
+ADD_H5_TEST (chunk2conti "TEST" h5repack_layout3.h5 -l chunk_unlimit1:CONTI)
+ADD_H5_VERIFY_TEST (chunk2conti "TEST" 0 h5repack_layout3.h5 chunk_unlimit1 CONTI)
+
+# chunk to compact - convert big dataset (should be > 64k) for this purpose, 
+# should remain as original layout (chunk)
+ADD_H5_TEST (chunk2compa "TEST" h5repack_layout3.h5 -l chunk_unlimit1:COMPA)
+ADD_H5_VERIFY_TEST (chunk2compa "TEST" 0 h5repack_layout3.h5 chunk_unlimit1 CHUNK)
+
 # Native option
 # Do not use FILE1, as the named dtype will be converted to native, and h5diff will
 # report a difference.
diff --git a/tools/h5repack/h5repack.c b/tools/h5repack/h5repack.c
index 27662e9..6517ae1 100644
--- a/tools/h5repack/h5repack.c
+++ b/tools/h5repack/h5repack.c
@@ -201,10 +201,12 @@ int h5repack_addlayout(const char* str,
     if (obj_list==NULL)
         return -1;
 
-    /* set global layout option */
+    /* set layout option */
+    options->layout_g = pack.layout;
+
+    /* no individual dataset specified */
     if (options->all_layout==1 )
     {
-        options->layout_g=pack.layout;
         if (pack.layout==H5D_CHUNKED)
         {
             /* -2 means the NONE option, remove chunking
@@ -223,6 +225,7 @@ int h5repack_addlayout(const char* str,
         }
     }
 
+    /* individual dataset specified */
     if (options->all_layout==0)
         options_add_layout(obj_list,
         n_objs,
@@ -499,7 +502,7 @@ int copy_attr(hid_t loc_in,
         if (type_class == H5T_COMPOUND) {
         	int nmembers = H5Tget_nmembers(wtype_id) ;
         	for (j=0; j<nmembers; j++) {
-        		hid_t mtid = H5Tget_member_type( wtype_id, j );
+        		hid_t mtid = H5Tget_member_type( wtype_id, (unsigned) j );
         		H5T_class_t mtclass = H5Tget_class(mtid);
         		H5Tclose(mtid);
 
diff --git a/tools/h5repack/h5repack.h b/tools/h5repack/h5repack.h
index 4c43bd4..81a883d 100644
--- a/tools/h5repack/h5repack.h
+++ b/tools/h5repack/h5repack.h
@@ -28,6 +28,7 @@
 #define MAX_VAR_DIMS 32               /* max per variable dimensions */
 #define FORMAT_OBJ      " %-27s %s\n"   /* obj type, name */
 #define FORMAT_OBJ_ATTR "  %-27s %s\n"  /* obj type, name */
+#define MAX_COMPACT_DSIZE  64512  /* max data size for compact layout. -1k for header size */
 
 /*-------------------------------------------------------------------------
  * data structures for command line options
diff --git a/tools/h5repack/h5repack.sh.in b/tools/h5repack/h5repack.sh.in
index 236c555..e05b03f 100644
--- a/tools/h5repack/h5repack.sh.in
+++ b/tools/h5repack/h5repack.sh.in
@@ -94,6 +94,7 @@ $SRC_H5REPACK_TESTFILES/h5repack_hlink.h5
 $SRC_H5REPACK_TESTFILES/h5repack_layout.h5
 $SRC_H5REPACK_TESTFILES/h5repack_layouto.h5
 $SRC_H5REPACK_TESTFILES/h5repack_layout2.h5
+$SRC_H5REPACK_TESTFILES/h5repack_layout3.h5
 $SRC_H5REPACK_TESTFILES/h5repack_named_dtypes.h5
 $SRC_H5REPACK_TESTFILES/h5repack_nbit.h5
 $SRC_H5REPACK_TESTFILES/h5repack_objs.h5
@@ -165,7 +166,7 @@ TESTING() {
 # beginning with the word "Verifying".
 #
 VERIFY() {
-    MESSAGE "Verifying h5diff output $*"
+    MESSAGE "Verifying $*"
 }
 
 # Print a message that a test has been skipped (because a required filter
@@ -179,7 +180,7 @@ SKIP() {
 #
 DIFFTEST() 
 {
-    VERIFY  $@
+    VERIFY h5diff output $@
     $RUNSERIAL $H5DIFF_BIN -q  "$@" 
     RET=$?
     if [ $RET != 0 ] ; then
@@ -235,7 +236,7 @@ VERIFY_LAYOUT_DSET()
     
     #---------------------------------
     # check the layout from a dataset
-    VERIFY  "Layout"
+    VERIFY  "a dataset layout"
     $RUNSERIAL $H5DUMP_BIN -d $dset -pH $outfile > $layoutfile
     $GREP $expectlayout $layoutfile > /dev/null
     if [ $? -eq 0 ]; then
@@ -260,7 +261,7 @@ VERIFY_LAYOUT_ALL()
     #---------------------------------
     # check the layout from a dataset
     # check if the other layouts still exsit
-    VERIFY  "Layout "
+    VERIFY  "layouts"
         # if CONTIGUOUS
         if [ $expectlayout = "CONTIGUOUS" ]; then
             $RUNSERIAL $H5DUMP_BIN -pH $outfile > $layoutfile
@@ -360,7 +361,8 @@ TOOLTEST1()
     rm -f $outfile
 }
           
-# Call h5repack and compare output to a text file for -v option
+# This is same as TOOLTEST() with comparing display output
+# from -v option
 #
 TOOLTESTV() 
 {
@@ -374,10 +376,20 @@ TOOLTESTV()
     outfile=$TESTDIR/out.$1
     shift
     $RUNSERIAL $H5REPACK_BIN "$@" $infile $outfile >$actual 2>$actual_err
-    cp $actual $actual_sav
+    RET=$?
+    if [ $RET != 0 ] ; then
+	echo "*FAILED*"
+	nerrors="`expr $nerrors + 1`"
+    else
+	echo " PASSED"
+	DIFFTEST $infile $outfile
+    fi
+
+    # display output compare
     STDOUT_FILTER $actual
     cat $actual_err >> $actual
 
+    VERIFY output from $H5REPACK $@
     if cmp -s $expect $actual; then
      echo " PASSED"
     else
@@ -387,7 +399,8 @@ TOOLTESTV()
      test yes = "$verbose" && diff -c $expect $actual |sed 's/^/    /'
     fi
     
-   rm -f $actual $actual_err $actual_sav
+   rm -f $actual $actual_err
+   rm -f $outfile
 }
 
 # TOOLTEST_META:
@@ -559,7 +572,8 @@ arg="h5repack_filters.h5 -v -f /dset_deflate:GZIP=9"
 if test  $USE_FILTER_DEFLATE != "yes" ; then
  SKIP $arg
 else
- TOOLTEST $arg
+ # compare output
+ TOOLTESTV $arg
 fi
   
 ###########################################################
@@ -778,6 +792,24 @@ VERIFY_LAYOUT_DSET h5repack_layout2.h5 contig_small COMPACT
 TOOLTEST_MAIN h5repack_layout2.h5 -l chunked_small_fixed:COMPA
 VERIFY_LAYOUT_DSET h5repack_layout2.h5 chunked_small_fixed COMPACT
 
+#--------------------------------------------------------------------------- 
+# Test file contains chunked datasets (need multiple dsets) with 
+# unlimited max dims.   (HDFFV-7933)
+# Use first dset to test.
+#---------------------------------------------------------------------------
+# chunk to chunk - specify chunk dim bigger than any current dim
+TOOLTEST_MAIN h5repack_layout3.h5 -l chunk_unlimit1:CHUNK=100x300
+VERIFY_LAYOUT_DSET h5repack_layout3.h5 chunk_unlimit1 CHUNK
+
+# chunk to contiguous 
+TOOLTEST_MAIN h5repack_layout3.h5 -l chunk_unlimit1:CONTI
+VERIFY_LAYOUT_DSET h5repack_layout3.h5 chunk_unlimit1 CONTI
+
+# chunk to compact - convert big dataset (should be > 64k) for this purpose, 
+# should remain as original layout (chunk)
+TOOLTEST_MAIN h5repack_layout3.h5 -l chunk_unlimit1:COMPA
+VERIFY_LAYOUT_DSET h5repack_layout3.h5 chunk_unlimit1 CHUNK
+
 # Native option
 # Do not use FILE1, as the named dtype will be converted to native, and h5diff will
 # report a difference.
@@ -856,8 +888,7 @@ TOOLTEST h5repack_attr_refs.h5
 # Note: this test is experimental for sharing test file among tools
 TOOLTEST h5diff_attr1.h5
 
-# tests for metadata block size option ('-M')
-TOOLTEST_META h5repack_layout.h5 -M 8192
+# tests for metadata block size option 
 TOOLTEST_META h5repack_layout.h5 --metadata_block_size=8192
 
 if test $nerrors -eq 0 ; then
diff --git a/tools/h5repack/h5repack_copy.c b/tools/h5repack/h5repack_copy.c
index 03a0043..c936155 100644
--- a/tools/h5repack/h5repack_copy.c
+++ b/tools/h5repack/h5repack_copy.c
@@ -737,6 +737,8 @@ int do_copy_objects(hid_t fidin,
     unsigned u;
     int      is_ref=0;
     htri_t   is_named;
+    hbool_t  limit_maxdims;
+    hsize_t size_dset;
 
 
     /*-------------------------------------------------------------------------
@@ -753,8 +755,10 @@ int do_copy_objects(hid_t fidin,
 
     for ( i = 0; i < travt->nobjs; i++)
     {
-
+        /* init variables per obj */
         buf = NULL;
+        limit_maxdims = FALSE;
+
         switch ( travt->objs[i].type )
         {
 
@@ -929,6 +933,9 @@ int do_copy_objects(hid_t fidin,
                 if((msize = H5Tget_size(wtype_id)) == 0)
                     goto error;
 
+                /* size of current dset */
+                size_dset = nelmts * msize;
+
                 /*-------------------------------------------------------------------------
                 * check if the dataset creation property list has filters that
                 * are not registered in the current configuration
@@ -958,7 +965,7 @@ int do_copy_objects(hid_t fidin,
                          */
                         if (options->layout_g != H5D_COMPACT)
                         {
-                            if ( nelmts*msize < options->min_comp )
+                            if ( size_dset < options->min_comp )
                                 apply_s=0;
                         }
 
@@ -975,9 +982,38 @@ int do_copy_objects(hid_t fidin,
                                 goto error;
                         }
 
-                        /* unset the unlimimted dimensions, which cannot be applied to layout other than chunked. */ 
-                        if (options->layout_g != H5D_CHUNKED) {
-                             H5Sset_extent_simple( f_space_id, rank, dims, NULL );
+                        /*------------------------------------------------- 
+                         * Unset the unlimited max dims if convert to other
+                         * than chunk layouts, because unlimited max dims
+                         * only can be applied to chunk layout. 
+                         * Also perform only for targeted dataset
+                         * Also check for size limit to convert to compact
+                         *-------------------------------------------------*/
+                        if (options->layout_g != H5D_CHUNKED) 
+                        {
+                            /* any dataset is specified */
+                            if (options->op_tbl->nelems > 0)
+                            {
+                                /* if current obj match specified obj */
+                                if (options_get_object (travt->objs[i].name, options->op_tbl))
+                                    limit_maxdims = TRUE;
+                            }
+                            else /* no dataset is specified */
+                            {
+                                limit_maxdims = TRUE;
+                            }
+                        
+                            /* if convert to COMPACT */
+                            if (options->layout_g == H5D_COMPACT)
+                            {
+                                /* should be smaller than 64K */
+                                if ( size_dset > MAX_COMPACT_DSIZE )
+                                    limit_maxdims = FALSE;
+                            }
+                        
+                            /* unset unlimited max dims */
+                            if (limit_maxdims)
+                                H5Sset_extent_simple( f_space_id, rank, dims, NULL );
                         }
 
                         /*-------------------------------------------------------------------------
diff --git a/tools/h5repack/h5repack_main.c b/tools/h5repack/h5repack_main.c
index 2df59c3..605d50a 100644
--- a/tools/h5repack/h5repack_main.c
+++ b/tools/h5repack/h5repack_main.c
@@ -425,6 +425,8 @@ void parse_command_line(int argc, const char **argv, pack_opt_t* options)
                 HDexit(EXIT_FAILURE);
             }
             break;
+        default:
+            break;
         } /* switch */
 
 
diff --git a/tools/h5repack/h5repack_opttable.c b/tools/h5repack/h5repack_opttable.c
index d0fcff3..4ec7fcf 100644
--- a/tools/h5repack/h5repack_opttable.c
+++ b/tools/h5repack/h5repack_opttable.c
@@ -388,11 +388,22 @@ pack_info_t* options_get_object( const char *path,
                                  pack_opttbl_t *table )
 {
     unsigned int i;
+    const char tbl_path[MAX_NC_NAME];
+
 
     for ( i = 0; i < table->nelems; i++)
     {
+        /* make full path (start with "/") to compare correctly  */
+        if (HDstrncmp(table->objs[i].path, "/", 1))
+        {
+            HDstrcpy(tbl_path, "/");
+            HDstrcat(tbl_path, table->objs[i].path);
+        }
+        else
+            HDstrcpy(tbl_path, table->objs[i].path);
+
         /* found it */
-        if (HDstrcmp(table->objs[i].path,path)==0)
+        if (HDstrcmp(tbl_path, path)==0)
         {
             return (&table->objs[i]);
         }
diff --git a/tools/h5repack/h5repacktst.c b/tools/h5repack/h5repacktst.c
index a30b671..20dbcd4 100644
--- a/tools/h5repack/h5repacktst.c
+++ b/tools/h5repack/h5repacktst.c
@@ -121,6 +121,7 @@ static int make_hlinks(hid_t loc_id);
 static int make_early(void);
 static int make_layout(hid_t loc_id);
 static int make_layout2(hid_t loc_id);
+static int make_layout3(hid_t loc_id);
 #ifdef H5_HAVE_FILTER_SZIP
 static int make_szip(hid_t loc_id);
 #endif /* H5_HAVE_FILTER_SZIP */
@@ -1697,6 +1698,19 @@ int make_testfiles(void)
         return -1;
 
     /*-------------------------------------------------------------------------
+    * for test layout conversions form chunk with unlimited max dims
+    *-------------------------------------------------------------------------
+    */
+    if((fid = H5Fcreate("h5repack_layout3.h5", H5F_ACC_TRUNC, H5P_DEFAULT, H5P_DEFAULT)) < 0)
+        return -1;
+
+    if(make_layout3(fid) < 0)
+        goto out;
+
+    if(H5Fclose(fid) < 0)
+        return -1;
+
+    /*-------------------------------------------------------------------------
     * create a file for the H5D_ALLOC_TIME_EARLY test
     *-------------------------------------------------------------------------
     */
@@ -3049,6 +3063,94 @@ out:
 } /* make_layout2() */
 
 /*-------------------------------------------------------------------------
+* Function: make_layout3
+*
+* Purpose: make chunked datasets with unlimited max dim and chunk dim is
+*          bigger than current dim. (HDFFV-7933)
+*          Test for converting chunk to chunk , chunk to conti and chunk
+*          to compact.  
+*          - The chunk to chunk changes layout bigger than any current dim 
+*            again. 
+*          - The chunk to compact test dataset bigger than 64K, should
+*            remain original layout.*
+*
+*-------------------------------------------------------------------------
+*/
+#define DIM1_L3 300
+#define DIM2_L3 200
+static
+int make_layout3(hid_t loc_id)
+{
+    hid_t    dcpl=-1; /* dataset creation property list */
+    hid_t    sid=-1;  /* dataspace ID */
+    hsize_t  dims[RANK]={DIM1_L3,DIM2_L3};
+    hsize_t  maxdims[RANK]={H5S_UNLIMITED, H5S_UNLIMITED};
+    hsize_t  chunk_dims[RANK]={DIM1_L3*2,5};
+    int      buf[DIM1_L3][DIM2_L3];
+    int      i, j, n;
+
+    for (i=n=0; i<DIM1_L3; i++)
+    {
+        for (j=0; j<DIM2_L3; j++)
+        {
+            buf[i][j]=n++;
+        }
+    }
+
+    /*-------------------------------------------------------------------------
+    * make several dataset with several layout options
+    *-------------------------------------------------------------------------
+    */
+    /* create a space */
+    if((sid = H5Screate_simple(RANK, dims, maxdims)) < 0)
+        return -1;
+    /* create a dataset creation property list; the same DCPL is used for all dsets */
+    if ((dcpl = H5Pcreate(H5P_DATASET_CREATE)) < 0)
+    {
+        goto out;
+    }
+
+
+    /*-------------------------------------------------------------------------
+    * H5D_CHUNKED
+    *-------------------------------------------------------------------------
+    */
+    if(H5Pset_chunk(dcpl, RANK, chunk_dims) < 0)
+        goto out;
+    if (make_dset(loc_id,"chunk_unlimit1",sid,dcpl,buf) < 0)
+    {
+        goto out;
+    }
+
+    if(H5Pset_chunk(dcpl, RANK, chunk_dims) < 0)
+        goto out;
+
+    if (make_dset(loc_id,"chunk_unlimit2",sid,dcpl,buf) < 0)
+    {
+        goto out;
+    }
+
+    /*-------------------------------------------------------------------------
+    * close space and dcpl
+    *-------------------------------------------------------------------------
+    */
+    if(H5Sclose(sid) < 0)
+        goto out;
+    if(H5Pclose(dcpl) < 0)
+        goto out;
+
+    return 0;
+
+out:
+    H5E_BEGIN_TRY
+    {
+        H5Pclose(dcpl);
+        H5Sclose(sid);
+    } H5E_END_TRY;
+    return -1;
+}
+
+/*-------------------------------------------------------------------------
 * Function: make a file with an integer dataset with a fill value
 *
 * Purpose: test copy of fill values
diff --git a/tools/h5repack/testfiles/h5repack_layout3.h5 b/tools/h5repack/testfiles/h5repack_layout3.h5
new file mode 100644
index 0000000..5319031
--- /dev/null
+++ b/tools/h5repack/testfiles/h5repack_layout3.h5