philipp/hdf5
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[svn-r27733] Merge of r26602 from trunk

Browse Source 
Remove the H5_SW_ULONG_TO_FP_BOTTOM_BIT_WORKS and
H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS macros/defines, as they are no longer used by
supported platforms.

Tested on: jam (has baked in the trunk for over 6 months)
This commit is contained in:
Dana Robinson
2015-09-09 21:52:47 -05:00
parent 67bc86740e
commit ad8c586c59
2 changed files with 15 additions and 43 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
release_docs/RELEASE.txt
View File

@@ -103,6 +103,7 @@ New Features
nec-superux14.1
sv1-cray
x86_64-redstorm-linux-gnu
powerpc-ibm-aix5.x
As a part of this work, a few lines that deal with locating the
sys/fpu.h header on SGI machines and some OSF/1 configure lines
@@ -119,6 +120,20 @@ New Features
(DER - 2015/09/09 HDFFV-9191)
- Removed the H5_SW_ULONG_TO_FP_BOTTOM_BIT_WORKS and
H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS symbols and associated code.
H5_SW_ULONG_TO_FP_BOTTOM_BIT_WORKS was a work-around on old 64-bit
SGI and Solaris systems.
H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS was a work-around for old PGI
compilers on Linux.
Neither of these were used in any current library code and only appeared
in the dt_arith test.
(DER - 2015/09/09 HDFFV-9187)
Library
-------

43
test/dt_arith.c
View File

@@ -4343,49 +4343,6 @@ test_conv_int_fp(const char *name, int run_test, hid_t src, hid_t dst)
}
}
}
/* On some machines (notably the SGI and Solaris 64-bit machines) unsigned long
* values are not converted to float or double values correctly, they are
* consistently off by the lowest bit being rounded oppositely to our
* software conversion routines output. So, on those machines, we allow
* the converted value to be +/- 1 from the machine's value. -QAK
*/
#ifndef H5_SW_ULONG_TO_FP_BOTTOM_BIT_WORKS
if(dst_size==sizeof(unsigned)) {
unsigned tmp_s, tmp_h;
HDmemcpy(&tmp_s,&buf[j*dst_size],sizeof(unsigned));
HDmemcpy(&tmp_h,&hw[0],sizeof(unsigned));
if((tmp_s+1)==tmp_h || (tmp_s-1)==tmp_h)
continue; /*no error*/
} /* end if */
else if (dst_size==sizeof(unsigned long)) {
unsigned long tmp_s, tmp_h;
HDmemcpy(&tmp_s,&buf[j*dst_size],sizeof(unsigned long));
HDmemcpy(&tmp_h,&hw[0],sizeof(unsigned long));
if((tmp_s+1)==tmp_h || (tmp_s-1)==tmp_h)
continue; /*no error*/
} /* end if */
else if (dst_size==sizeof(unsigned long long)) {
unsigned long long tmp_s, tmp_h;
HDmemcpy(&tmp_s,&buf[j*dst_size],sizeof(unsigned long long));
HDmemcpy(&tmp_h,&hw[0],sizeof(unsigned long long));
if((tmp_s+1)==tmp_h || (tmp_s-1)==tmp_h)
continue; /*no error*/
} /* end if */
#endif /* end H5_ULONG_FP_BOTTOM_BIT_WORKS */
/* For PGI compiler on Linux, during conversion from 'float' or 'double' to
* 'unsigned long long', round-up happens when the fraction of float-point
* value is greater than 0.5. So we allow the converted value to be off by 1.
*/
#ifndef H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS
if((src_type==FLT_FLOAT || src_type==FLT_DOUBLE) && dst_type==INT_ULLONG) {
unsigned long long tmp_s, tmp_h;
HDmemcpy(&tmp_s,&buf[j*dst_size],sizeof(unsigned long long));
HDmemcpy(&tmp_h,&hw[0],sizeof(unsigned long long));
if((tmp_s+1)==tmp_h)
continue; /*no error*/
}
#endif /*end H5_FP_TO_ULLONG_BOTTOM_BIT_WORKS*/
/* Print errors */
if (0==fails_this_test++) {