[netCDF #GTY-840307]: valgrind test fails on Fedora

> I'm getting the following test failure of Fedora 13, 14, and rawhide with
> today's netcdf-4.1.2-beta1.
> 
> Testing programs with valgrind...
> Memory testing with tst_h_dimscales:
> *** Checking HDF5 dimension scales.
> *** Creating simple dimension scales file...ok.
> *** Checking that simple dimscale file can be read...ok.
> *** Creating simple dimension scales file with lots of datasets...ok.
> *** Creating a file with an unlimited dimension scale...ok.
> *** Checking that unlimited dimscale file can be read...ok.
> *** Creating some 3D datasets using shared dimscales...ok.
> *** Checking 3D datasets created with shared dimscales...ok.
> *** Creating 3D datasets using shared dimscales in groups...ok.
> *** Checking 3D datasets in groups created with shared dimscales...ok.
> *** Tests successful!
> Memory testing with tst_h_files:
> *** Checking HDF5 file functions.
> *** Checking HDF5 file creates and opens...ok.
> *** Checking HDF5 file creates and opens some more...ok.
> *** Creating file...ok.
> *** Creating HDF5 file in the canonical netCDF-4 way...ok.
> *** Tests successful!
> Memory testing with tst_h_atts2:
> *** Checking HDF5 attribute functions some more.
> *** Tests successful!
> Memory testing with tst_h_dimscales2:
> *** Checking HDF5 dimscales some more.
> *** Creating a file with one var with one dimension scale...ok.
> *** Checking that one var, one dimscale file can be read...ok.
> *** Creating a file with one var with two dimension scales...ok.
> *** Checking that one var, two dimscales file can be read...ok.
> *** Creating a file with one var with two unlimited dimension scales...ok.
> *** Checking dimension scales with attached dimension scales...ok.
> *** Checking cration ordering of datasets which are also dimension 
> scales...ok.
> *** Tests successful!
> Memory testing with tst_h_enums:
> *** Checking HDF5 enum types.
> *** Checking simple HDF5 enum type...ok.
> *** Checking HDF5 enum type missing values...ok.
> *** Tests successful!
> Memory testing with tst_h_grps:
> *** Checking HDF5 group functions.
> *** Checking out root group...ok.
> *** Checking out H5Gmove...ok.
> *** Checking out sub-groups...ok.
> *** Checking out UTF8 named sub-group...ok.
> *** Checking out UTF8 named sub-group with group creation ordering...ok.
> *** Tests successful!
> Memory testing with tst_h_wrt_cmp:
> *** Checking HDF5 compound types (even more so).
> *** Checking packing of HDF5 compound types...ok.
> *** Tests successful!
> Memory testing with tst_h_rd_cmp:
> *** Checking HDF5 compound types (we're almost there kids).
> *** Checking packing of HDF5 compound types...ok.
> *** Tests successful!
> Memory testing with tst_h_vl:
> *** Checking HDF5 VLEN types.
> *** Checking simple HDF5 variable length types...ok.
> *** Tests successful!
> Memory testing with tst_h_atts:
> *** Checking HDF5 attribute functions.
> *** Checking HDF5 attribute ordering...ok.
> *** Checking HDF5 attribute deletes...ok.
> *** Checking HDF5 attributes attached to the fileid...ok.
> *** Checking HDF5 attributes in a group...ok.
> *** Checking HDF5 zero length attributes...ok.
> *** Checking how HDF5 handles a delete and recreation of dataset with 
> atts...ok.
> *** Checking some more simple atts...ok.
> *** Checking some more simple atts even more...ok.
> *** Checking HDF5 attribute ordering some more...ok.
> *** Checking HDF5 attribute ordering with 9 attributes...(skipping for HDF5
> 1.8.0 beta1)ok.
> *** Tests successful!
> Memory testing with tst_h_dimscales3:
> *** Checking HDF5 phony, secret, and underhanded dimscales. Shhh! Don't tell
> anyone!
> *** Creating a phony dimscale...ok.
> *** Tests successful!
> Memory testing with tst_h_files2:
> *** Checking HDF5 file functions some more.
> *** Checking opaque attribute creation...ok.
> *** Tests successful!
> Memory testing with tst_h_compounds:
> ==7915== Syscall param write(buf) points to uninitialised byte(s)
> ==7915==    at 0x5E1D2A0: __write_nocancel (in /lib64/libc-2.12.90.so)
> ==7915==    by 0x563E17D: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x563436A: H5FD_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x561EC67: H5F_block_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55F705B: H5D_flush_sieve_buf (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55E9510: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55F7374: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55F4F4C: H5D_close (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x56B6DA7: H5I_dec_ref (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55CFF8B: H5Dclose (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x406670: main (tst_h_compounds.c:600)
> ==7915==  Address 0x9fdb48c is 28 bytes inside a block of size 65,544 alloc'd
> ==7915==    at 0x4C2847D: malloc (vg_replace_malloc.c:236)
> ==7915==    by 0x5640ABE: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x56419D2: H5FL_blk_malloc (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55E868C: H5D_contig_writevv (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55FFADE: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55FFEB6: H5D_select_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55E72D3: H5D_contig_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55F941E: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55F8134: H5Dwrite (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x404F31: main (tst_h_compounds.c:142)
> ==7915==
> ==7915== Syscall param write(buf) points to uninitialised byte(s)
> ==7915==    at 0x5E1D2A0: __write_nocancel (in /lib64/libc-2.12.90.so)
> ==7915==    by 0x563E17D: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x563436A: H5FD_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x561DC14: H5F_accum_flush (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x5618223: H5F_flush (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x56156A5: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x5618A13: H5F_try_close (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x561846D: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x56B6DA7: H5I_dec_ref (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x5618E11: H5Fclose (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x406BB6: main (tst_h_compounds.c:711)
> ==7915==  Address 0x9ed3727 is 1,671 bytes inside a block of size 4,104 
> alloc'd
> ==7915==    at 0x4C2847D: malloc (vg_replace_malloc.c:236)
> ==7915==    by 0x5640ABE: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x56419D2: H5FL_blk_malloc (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x5641EC5: H5FL_blk_realloc (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x561C6C9: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x561C9FF: H5F_accum_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x561EBB7: H5F_block_write (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x5657910: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55CDE06: ??? (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x55C3543: H5C_flush_cache (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x559EB1D: H5AC_flush (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==    by 0x56181AA: H5F_flush (in /usr/lib64/libhdf5.so.6.0.4)
> ==7915==
> *** Checking HDF5 compound types.
> *** Checking simple HDF5 compound types...ok.
> *** Checking HDF5 compound types and groups...ok.
> *** Checking HDF5 compound type which contains an array...ok.
> *** Checking HDF5 compound type 6 different types...ok.
> *** Checking HDF5 compound variable which contains a compound type...ok.
> *** Checking HDF5 variable which contains a compound type with different
> string handling...ok.
> *** Checking HDF5 compound attribute which contains a compound type...ok.
> *** Checking HDF5 variable of compound type which contains a string...ok.
> *** Checking HDF5 compound attribute which contains an array of char...ok.
> *** Checking HDF5 compound attribute which contains an array of unsigned
> byte...ok.
> *** Checking simple read of HDF5 compound attribute which contains a simple
> compound type...ok.
> *** Checking simple read of HDF5 compound attribute which contains a simple
> compound type...ok.
> *** Tests successful!
> FAIL: run_valgrind_tests.sh
> 
> 
> --
> Orion Poplawski
> Technical Manager                     303-415-9701 x222
> NWRA/CoRA Division                    FAX: 303-415-9702
> 3380 Mitchell Lane                  address@hidden
> Boulder, CO 80301              http://www.cora.nwra.com
> 
> 

Howdy Orion!

There have been many, many changes since the beta1 release to improve valgrind 
testing. As soon as I get current portability problems (mostly caused by the 
recent melt-down of my main test machine). If you want to see the improved 
code, get the daily snapshot, but there are still some valgrind failures in the 
nctest directory.

Thanks,

Ed

Ticket Details
===================
Ticket ID: GTY-840307
Department: Support netCDF
Priority: Normal
Status: Closed