AZL-88757

In the Linux kernel, the following vulnerability has been resolved:

bpf: reject negative CO-RE accessor indices in bpfcoreparse_spec()

CO-RE accessor strings are colon-separated indices that describe a path from a root BTF type to a target field, e.g. "0:1:2" walks through nested struct members. bpfcoreparsespec() parses each component with sscanf("%d"), so negative values like -1 are silently accepted. The subsequent bounds checks (accessidx >= btf_vlen(t)) only guard the upper bound and always pass for negative values because C integer promotion converts the _u16 btfvlen result to int, making the comparison (int)(-1) >= (int)(N) false for any positive N.

When -1 reaches btfmemberbitoffset() it gets cast to u32 0xffffffff, producing an out-of-bounds read far past the members array. A crafted BPF program with a negative CO-RE accessor on any struct that exists in vmlinux BTF (e.g. taskstruct) crashes the kernel deterministically during BPFPROGLOAD on any system with CONFIGDEBUGINFOBTF=y (default on major distributions). The bug is reachable with CAPBPF:

BUG: unable to handle page fault for address: ffffed11818b6626 #PF: supervisor read access in kernel mode #PF: errorcode(0x0000) - not-present page Oops: Oops: 0000 [#1] SMP KASAN NOPTI CPU: 0 UID: 0 PID: 85 Comm: poc Not tainted 7.0.0-rc6 #18 PREEMPT(full) RIP: 0010:bpfcoreparsespec (tools/lib/bpf/relocore.c:354) RAX: 00000000ffffffff Call Trace: <TASK> bpfcorecalcreloinsn (tools/lib/bpf/relocore.c:1321) bpfcoreapply (kernel/bpf/btf.c:9507) checkcorerelo (kernel/bpf/verifier.c:19475) bpfcheck (kernel/bpf/verifier.c:26031) bpfprog_load (kernel/bpf/syscall.c:3089) _sysbpf (kernel/bpf/syscall.c:6228) </TASK>

CO-RE accessor indices are inherently non-negative (struct member index, array element index, or enumerator index), so reject them immediately after parsing.