The reported bug in Android 17's bionic library, specifically affecting the wcscspn() and wcspbrk() wide string functions, reveals a critical flaw in the platform's core components and its development transparency. Discovered in September 2026 builds, the issue causes these functions to misinterpret wide characters (those with code points U+0100 or greater) as matches, even when they are not part of the specified "reject set"—provided the reject set itself consists solely of ASCII characters. This means that any application scanning wide strings for common ASCII delimiters, such as newlines or wildcard characters, will prematurely stop and report an incorrect match if the string contains non-ASCII characters like emojis, Cyrillic, or CJK characters. The problem was not present in the June 2026 android17-release sources, indicating a regression introduced within a three-month window.

Android 17

The core of the problem lies within the bionic C library, a fundamental component of the Android operating system. The bug specifically targets Android 17 builds, which were released in September 2026. This particular version of Android introduced a change in how wcscspn() and wcspbrk() are implemented, moving away from a plain nested loop approach found in external/llvm-libc to a more optimized, but flawed, mechanism. The new implementation attempts to use a 256-byte boolean table for faster lookups when the reject set contains only ASCII characters. However, it incorrectly terminates the scan when it encounters any wide character (U+0100 or above) in the haystack string, regardless of whether that character is actually in the reject set. This optimization, while intended to improve performance, inadvertently introduced a severe logical error that impacts global character support.

wcscspn()

The wcscspn() function, along with its complement wcspbrk(), is designed to scan wide character strings. wcscspn() calculates the length of the initial segment of a wide string that consists entirely of characters not in a specified set of wide characters. Conversely, wcspbrk() locates the first occurrence in a wide string of any character from a specified set. The bug's mechanism, as detailed by francisdb through disassembly of libc.so, shows that when the reject set contains only characters less than U+0100, the function's optimized path incorrectly stops scanning upon encountering any character greater than or equal to U+0100. This behavior is fundamentally incorrect for wcscspn() because such a character, by definition, cannot be in the ASCII-only reject set, and thus the scan should continue. The correct behavior, as observed in wcsspn(), is to stop only if the character is in the set.

Visual Pinball

The real-world impact of this bug is exemplified by applications like Visual Pinball on Android. The libwinevbs component, which is a Unix build of Wine's vbscript engine used by Visual Pinball, relies on wcspbrk(ptr, L"\n\r") to parse and skip comments in VBScript source code. After the September 2026 Android 17 update, if a VBScript file contained a comment with an emoji, the lexer would incorrectly identify the emoji's position as the end of the comment line instead of the actual newline character. This led to "Invalid character" errors and prevented tables from loading. Another affected function, FileSystem.MoveFolder, similarly misinterpreted paths containing non-ASCII characters as wildcard patterns due to its use of wcspbrk(src, L"*?><"). These incidents highlight how a low-level library bug can cascade into significant application failures, forcing developers to implement downstream workarounds, such as replacing the problematic wcspbrk() calls with custom plain loops, as seen in vpinball/libwinevbs#41. The necessity for such workarounds, coupled with the difficulty of diagnosing the issue due to delayed public source code, underscores a broader challenge in the Android open-source ecosystem.