Clover database contains FileInfo objects for every source file. One of FileInfo fields is an absolute path to given file. When database is written, it uses current platform file separator for file name ('\' for windows, '/' for Linux/MacOS). However, when database is read it interprets file name using again current platform file separator.
As a consequence, when database is generated on Windows and report generated on Linux, it cannot resolve file name properly - it takes whole path as a single path segment, resulting in paths like:
As a consequence report generation does create html file containing source code with highlighted coverage.