Executive brief
HDF5 is a high-performance library used for managing and storing large amounts of complex data. A vulnerability exists where a specially crafted, malicious HDF5 file can cause the library to crash or read unintended memory locations when the file is opened. This could lead to service disruptions or application instability if a user is tricked into processing a corrupted file.
Technical details
An out-of-bounds read vulnerability exists in HDF5 due to a lack of consistency checks between an array datatype's total size, number of elements, and individual element size in H5Odtype.c. Specifically, the library fails to verify that (element size * number of elements) equals the stored full size of the datatype. An attacker can exploit this by providing a maliciously altered HDF5 file with mismatched metadata, triggering out-of-bounds reads in various components such as H5Aint.c, H5Tconv_compound.c, and H5Dselect.c. This can result in application crashes or stack corruption. The issue is addressed in version 2.1.0.
Affected products
- HDFGroup hdf5 < 2.1.0
Timeline
- 2026-06-26: advisory: GitHub advisory published by maintainer
- 2026-07-20: disclosed: NVD publication date