{"id":"CVE-2026-98134","title":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: check_cond_jmp_op(): properly infer if register is null\n\nNicholas Carlini reported a bug when verifier can incorrectly infer\nthat a pointer is non-null","summary":"In the Linux kernel, the following vulnerability has been resolved:\n\nbpf: check_cond_jmp_op(): properly infer if register is null\n\nNicholas Carlini reported a bug when verifier can incorrectly infer\nthat a pointer is non-null. The bug oc…","severity":"none","vendor":"Linux","product":"Linux","affected":["Linux >= befae75856ab406a3f3fab2aa2118cf3b2dfe3e6 < b55c9f019e169ed0d01394f96e172286a8b21b99","Linux >= befae75856ab406a3f3fab2aa2118cf3b2dfe3e6 < d3ef6c097ba078e1f8c7239d76a0ce8b61e75095","Linux 6.2"],"published":"2026-09-25","updated":"2026-09-25","sourceUpdated":"2026-09-25T11:17:44.820","source":"NVD","sourceUrl":"https://nvd.nist.gov/vuln/detail/CVE-2026-98134","references":[{"url":"https://git.kernel.org/stable/c/b55c9f019e169ed0d01394f96e172286a8b21b99","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"},{"url":"https://git.kernel.org/stable/c/d3ef6c097ba078e1f8c7239d76a0ce8b61e75095","label":"416baaa9-dc9f-4396-8d5f-8c081fb06d67"}],"tags":["nvd","cve.org"],"ingestedAt":"2026-09-25T11:06:38.812Z","slug":"CVE-2026-98134","body":"## Overview\n\nIn the Linux kernel, the following vulnerability has been resolved:\n\nbpf: check_cond_jmp_op(): properly infer if register is null\n\nNicholas Carlini reported a bug when verifier can incorrectly infer\nthat a pointer is non-null. The bug occurs when two pointers are\ncompared and one of them has a type w/o PTR_MAYBE_NULL flag,\nbut which allows a value to be NULL at runtime.\nHere is an example:\n\n  // `a` is PTR_TO_MEM | MEM_RDONLY | PTR_UNTRUSTED\n  // `a` is 0 at runtime.\n  // `b` is PTR_TO_MAP_VALUE | PTR_MAYBE_NULL\n  void *a = bpf_rdonly_cast(0, 0);\n  int  *b = bpf_map_lookup_elem(...);\n\n  if (a == b)\n    *b = 42;  // verifier does not catch null pointer dereference\n\nThis happens because of a special case in check_cond_jmp_op(),\nwhich attempts to strip PTR_MAYBE_NULL flags from pointer types,\nwhen processing comparisons like `rA == rB`, if either rA or rB can't\nbe null.\n\nThe non-null property is derived based on the absence of\nPTR_MAYBE_NULL flag on rA's or rB's type. But that is not sufficient\nfor types like PTR_TO_MEM, as in the example.\n\nThis patch replaces type_may_be_null() call with reg_not_null(),\nwhich contains an allowlist of types for which absence of\nPTR_MAYBE_NULL actually means that the value can't be NULL at runtime.\n\nAt the moment, the list in the reg_not_null() omits two types for\nwhich PTR_MAYBE_NULL is applicable: PTR_TO_XDP_SOCK and PTR_TO_BUF.\nIn order to remain backward compatible, and assuming that only\ncomparison between pointers of the same type makes sense,\nthis commit extends reg_not_null(). W/o such an extension e.g.\nverifier_jeq_infer_not_null/null_ptr_to_map_value fails.\n\nreg_not_null() can be extended further, but I deem that out of scope\nfor the fix at hand. Explicit base_type(...) != PTR_TO_BTF_ID\nchecks in the check_cond_jmp_op() can be removed with migration to\nreg_not_null(), but that is a behavioural change, as the special case\nwould start matching for PTR_TO_BTF_ID that is also is_trusted_reg().\nI omit the behavioural change from this commit.\n\n## Remediation\n\nRefer to the linked advisories for vendor-supplied fixes and affected version ranges.","depth":"sunlit","depthScore":3,"depthScoreParts":{"impact":2.8,"likelihood":0,"exploitation":0,"ransomware":0},"changes":[]}