Fix uninitialized bytes in cmpd_dset test #4072
Conversation
Compound fill values were set to the integer -1, causing valgrind to flag 'uninitialized bytes' errors. This is just a problem with the cmpd_dset test and not a core library problem.
derobins
added
Priority - 0. Blocker ⛔
derobins
requested review from
lrknox,
byrnHDF,
fortnern,
jhendersonHDF,
qkoziol,
vchoi-hdfgroup,
bmribler,
glennsong09,
mattjala and
brtnfld
as code owners
|
Are you sure we're not losing something by changing the fill value from -1 to 0, which is more likely to be there "accidentally" if the library didn't fill the dataset due to a bug. I see using an int fill value for a compound type is wrong, but the fact that it was initialized to -1 implies it cares about the fill value being something other than 0. Maybe the code was copied from a different test where it matters more? |
It looks like a copy-paste error. I thought about initializing everything to -1, but that looked messy and the test clearly doesn't care about the fill value since it was initialized to "-1 + garbage bytes". |
Compound fill values were set to the integer -1, causing valgrind to flag 'uninitialized bytes' errors. This is just a problem with the cmpd_dset test and not a core library problem.
* Do not enable szip for sanitizer runs (#4057) * Add note to H5Tset_fields about needing to set datatype precision first (#4059) * Offset of a floating-point type also needs to be accounted for * Clarify ordering of H5Tset_precision and H5Tset_fields * Fix issue where H5Tset_fields does not account for datatype offsets (#4061) H5Tset_fields did not account for any offset in a floating-point datatype, causing it to fail when a datatype's precision is correctly set such that it doesn't include the offset bits. * Ignore UserPresets and Use only C compiler for sanitizers (#4066) * Remove user presets file * Only use C compiler for sanitzers * Rename incorrectly named option (#4067) * Rename incorrectly named option * Restore the correct uses of USING_MEMCHECKER * Update release note * Fix a memory leak in the cmpd_dset test (#4071) This was due to not freeing a test buffer. It was not a core library memory leak. * Fix uninitialized bytes in cmpd_dset test (#4072) Compound fill values were set to the integer -1, causing valgrind to flag 'uninitialized bytes' errors. This is just a problem with the cmpd_dset test and not a core library problem. * Update INSTALL files (#4052) * Add NEWSLETTER and merge abi reports and add sha256sums (#4055) * Fix uninitialized bytes in selection I/O test (#4073) This was due to a complex type fill value being set to -1 instead of a proper complex value. This was a test problem and not a core library issue. * fix path for S3 build path in CI (#4076) * Correct paths for 1.14 and add lines missing from release_docs/INSTALL_CMake.txt. --------- Co-authored-by: Allen Byrne <[email protected]> Co-authored-by: jhendersonHDF <[email protected]> Co-authored-by: Dana Robinson <[email protected]>
Compound fill values were set to the integer -1, causing valgrind to flag 'uninitialized bytes' errors.
This is just a problem with the cmpd_dset test and not a core library problem.