Buffer Under-read in GNU C Library Affects Wide Character Handling

CVE-2026-5928

What is CVE-2026-5928?

The GNU C Library, specifically version 2.43 and earlier, contains a vulnerability in the handling of wide characters, specifically within the ungetwc function. This issue arises when wide characters encoded in overlapping single byte and multi-byte character sets are processed. As a result, the function may attempt to read bytes beyond the allocated buffer, which can lead to unintentional disclosure of adjacent heap data or, in certain conditions, cause a program crash. The underlying flaw is linked to the _IO_wdefault_pbackfail implementation, which incorrectly uses the regular character buffer instead of the appropriate wide-stream read pointer, particularly when initialization issues occur. This vulnerability is primarily triggered under specific character encodings and does not apply to standard Unicode sets.

References