summaryrefslogtreecommitdiffstats
path: root/test
diff options
context:
space:
mode:
authorRaymond Lu <songyulu@hdfgroup.org>2005-09-12 19:39:16 (GMT)
committerRaymond Lu <songyulu@hdfgroup.org>2005-09-12 19:39:16 (GMT)
commit93e661f16ba03b2acca208fb16518e6ce26e86d8 (patch)
tree65d81102c018e950c8bf30ba07186d10ed4e8b32 /test
parentc78c580f5cb64abd8bf33331f11ef666382741cb (diff)
downloadhdf5-93e661f16ba03b2acca208fb16518e6ce26e86d8.zip
hdf5-93e661f16ba03b2acca208fb16518e6ce26e86d8.tar.gz
hdf5-93e661f16ba03b2acca208fb16518e6ce26e86d8.tar.bz2
[svn-r11395] Purpose: Bug fix
Description: Quincey added the condition branch (else if (*((ST*)S) != (ST)((DT)(*((ST*)S))))) in the definition of the macro H5T_CONV_Fx_CORE. It handles a special situation when the source is "float" and assigned the value of "INT_MAX". Compilers do roundup making this value "INT_MAX+1". This branch is to check that situation and return exception for some compilers, mainly GCC. The branch if (*((ST*)S) > (DT)(D_MAX) || (S_PREC < D_PREC && *((ST*)S) == (DT)(D_MAX))) is for some compilers like Sun, HP, IBM, and SGI where under the same situation the "int" doesn't overflow. A test was added to dt_arith.c to check 2 conversions, from float to int where the float is INT_MAX, and from double to signed char where the double is SCHAR_MAX. Platforms tested: h5committest and fuss.
Diffstat (limited to 'test')
-rw-r--r--test/dt_arith.c226
1 files changed, 226 insertions, 0 deletions
diff --git a/test/dt_arith.c b/test/dt_arith.c
index abdba99..f13bba8 100644
--- a/test/dt_arith.c
+++ b/test/dt_arith.c
@@ -639,6 +639,229 @@ test_hard_query(void)
/*-------------------------------------------------------------------------
+ * Function: expt_handle
+ *
+ * Purpose: Gets called from test_particular_fp_integer() for data type
+ * conversion exceptions.
+ *
+ * Return: H5T_CONV_HANDLED 1
+ *
+ * Programmer: Raymond Lu
+ * Sept 7, 2005
+ *
+ * Modifications:
+ *
+ *-------------------------------------------------------------------------
+ */
+static H5T_conv_ret_t
+expt_handle(H5T_conv_except_t except_type, hid_t UNUSED src_id, hid_t UNUSED dst_id, void UNUSED *src_buf,
+ void *dst_buf, void *user_data)
+{
+ H5T_conv_ret_t ret = H5T_CONV_HANDLED;
+ signed char fill_value1 = 7;
+ int fill_value2 = 13;
+
+ if(except_type == H5T_CONV_EXCEPT_RANGE_HI || except_type == H5T_CONV_EXCEPT_RANGE_LOW ||
+ except_type == H5T_CONV_EXCEPT_TRUNCATE) {
+ if(*(hbool_t*)user_data)
+ *(signed char*)dst_buf = fill_value1;
+ else
+ *(int*)dst_buf = fill_value2;
+ }
+
+ return ret;
+}
+
+
+/*-------------------------------------------------------------------------
+ * Function: test_particular_fp_integer
+ *
+ * Purpose: Tests hard conversions from floating numbers to integers in
+ * a special situation when the source is "float" and assigned
+ * the value of "INT_MAX". A compiler may do roundup making
+ * this value "INT_MAX+1". When this float value is casted to
+ * int, overflow happens. This test makes sure the library
+ * returns exception in this situation.
+ *
+ * Also verifies the library handles conversion from double to
+ * signed char correctly when the value of double is SCHAR_MAX.
+ * The test makes sure the signed char doesn't overflow.
+ *
+ * This test is mainly for netCDF's request.
+ *
+ * Return: Success: 0
+ *
+ * Failure: number of errors
+ *
+ * Programmer: Raymond Lu
+ * Sept 7, 2005
+ *
+ * Modifications:
+ *
+ *-------------------------------------------------------------------------
+ */
+static int test_particular_fp_integer(void)
+{
+ hid_t dxpl_id;
+ hbool_t flag;
+ double src_d = (double)SCHAR_MAX;
+ signed char dst_c;
+ unsigned char *buf1, *buf2;
+ unsigned char *saved_buf1, *saved_buf2;
+ unsigned int src_size1, src_size2;
+ unsigned int dst_size1, dst_size2;
+ float src_f = (float)INT_MAX;
+ int dst_i;
+ int fill_value = 13;
+ int endian; /*endianess */
+ char str[256]; /*message string */
+ unsigned int fails_this_test = 0;
+ int i, j;
+
+ TESTING("hard particular floating number -> integer conversions");
+
+ if((dxpl_id = H5Pcreate(H5P_DATASET_XFER))<0) {
+ H5_FAILED();
+ printf("Can't create data transfer property list\n");
+ goto error;
+ }
+
+ /* Test conversion from double (the value is SCHAR_MAX) to signed char. */
+ endian = H5Tget_order(H5T_NATIVE_DOUBLE);
+ src_size1 = H5Tget_size(H5T_NATIVE_DOUBLE);
+ dst_size1 = H5Tget_size(H5T_NATIVE_SCHAR);
+ buf1 = (unsigned char*)calloc(1, MAX(src_size1, dst_size1));
+ saved_buf1 = (unsigned char*)calloc(1, MAX(src_size1, dst_size1));
+
+ memcpy(buf1, &src_d, src_size1);
+ memcpy(saved_buf1, &src_d, src_size1);
+
+ /* Register exception handling function and signal the destination is "signed char". */
+ flag = 1;
+ if(H5Pset_type_conv_cb(dxpl_id, expt_handle, &flag)<0) {
+ H5_FAILED();
+ printf("Can't register conversion callback\n");
+ goto error;
+ }
+
+ /* Do conversion */
+ if(H5Tconvert(H5T_NATIVE_DOUBLE, H5T_NATIVE_SCHAR, 1, buf1, NULL, dxpl_id)<0) {
+ H5_FAILED();
+ printf("Can't convert data\n");
+ goto error;
+ }
+
+ memcpy(&dst_c, buf1, dst_size1);
+
+ /* Print errors */
+ if(dst_c != SCHAR_MAX) {
+ double x;
+ signed char y;
+
+ if(0 == fails_this_test++)
+ H5_FAILED();
+
+ printf(" test double to signed char:\n");
+ printf(" src = ");
+ for (j=0; j<src_size1; j++)
+ printf(" %02x", saved_buf1[ENDIAN(src_size1, j)]);
+
+ HDmemcpy(&x, saved_buf1, src_size1);
+ printf(" %29.20e\n", x);
+
+ printf(" dst = ");
+ for (j=0; j<dst_size1; j++)
+ printf(" %02x", buf1[ENDIAN(dst_size1, j)]);
+
+ HDmemcpy(&y, buf1, dst_size1);
+ printf(" %29d\n", y);
+ }
+
+ /* Test conversion from float (the value is INT_MAX) to int. */
+ src_size2 = H5Tget_size(H5T_NATIVE_FLOAT);
+ dst_size2 = H5Tget_size(H5T_NATIVE_INT);
+ buf2 = (unsigned char*)calloc(1, MAX(src_size2, dst_size2));
+ saved_buf2 = (unsigned char*)calloc(1, MAX(src_size2, dst_size2));
+ memcpy(buf2, &src_f, src_size2);
+ memcpy(saved_buf2, &src_f, src_size2);
+
+ /* signal exception handling function that the destination is "int". */
+ flag = 0;
+
+ /* Do conversion */
+ if(H5Tconvert(H5T_NATIVE_FLOAT, H5T_NATIVE_INT, 1, buf2, NULL, dxpl_id)<0) {
+ H5_FAILED();
+ printf("Can't convert data\n");
+ goto error;
+ }
+
+ memcpy(&dst_i, buf2, dst_size2);
+
+ /* Print errors */
+ if(dst_i != fill_value) {
+ float x;
+ int y;
+
+ if(0 == fails_this_test++)
+ H5_FAILED();
+
+ printf(" test float to int:\n");
+ printf(" src = ");
+ for (j=0; j<src_size2; j++)
+ printf(" %02x", saved_buf2[ENDIAN(src_size2, j)]);
+
+ HDmemcpy(&x, saved_buf2, src_size2);
+ printf(" %29.20e\n", x);
+
+ printf(" dst = ");
+ for (j=0; j<dst_size2; j++)
+ printf(" %02x", buf2[ENDIAN(dst_size2, j)]);
+
+ HDmemcpy(&y, buf2, dst_size2);
+ printf(" %29d\n", y);
+ }
+
+ if(fails_this_test)
+ goto error;
+
+ if(H5Pclose(dxpl_id)<0) {
+ H5_FAILED();
+ printf("Can't close property list\n");
+ goto error;
+ }
+
+ if(buf1)
+ free(buf1);
+ if(buf2)
+ free(buf2);
+ if(saved_buf1)
+ free(saved_buf1);
+ if(saved_buf2)
+ free(saved_buf2);
+
+ PASSED();
+ return 0;
+
+error:
+ HDfflush(stdout);
+ H5E_BEGIN_TRY {
+ H5Pclose(dxpl_id);
+ } H5E_END_TRY;
+ if(buf1)
+ free(buf1);
+ if(buf2)
+ free(buf2);
+ if(saved_buf1)
+ free(saved_buf1);
+ if(saved_buf2)
+ free(saved_buf2);
+
+ reset_hdf5(); /*print statistics*/
+ return MAX((int)fails_this_test, 1);
+}
+
+
+/*-------------------------------------------------------------------------
* Function: test_derived_flt
*
* Purpose: Tests user-define and query functions of floating-point types.
@@ -4858,6 +5081,9 @@ main(void)
/* Test hardware float-integer conversion functions */
nerrors += run_fp_int_conv("hard");
+ /* Test a few special values for hardware float-integer conversions */
+ nerrors += test_particular_fp_integer();
+
/*----------------------------------------------------------------------
* Software tests
*----------------------------------------------------------------------