In the Linux kernel, the following vulnerability has been resolved: bpf: sync_linked_regs() must preserve subreg_def Range propagation must not affect subreg_def marks, otherwise the following example is rewritten by verifier incorrectly when BPF_F_TEST_RND_HI32 flag is set: 0: call bpf_ktime_get_ns call bpf_ktime_get_ns 1: r0 &= 0x7fffffff after verifier r0 &= 0x7fffffff 2: w1 = w0 rewrites w1 = w0 3: if w0 < 10 goto +0 --------------> r11 = 0x2f5674a6 (r) 4: r1 >>= 32 r11 <<= 32 (r) 5: r0 = r1 r1 |= r11 (r) 6: exit; if w0 < 0xa goto pc+0 r1 >>= 32 r0 = r1 exit (or zero extension of w1 at (2) is missing for architectures that require zero extension for upper register half). The following happens w/o this patch: - r0 is marked as not a subreg at (0); - w1 is marked as subreg at (2); - w1 subreg_def is overridden at (3) by copy_register_state(); - w1 is read at (5) but mark_insn_zext() does not mark (2) for zero extension, because w1 subreg_def is not set; - because of BPF_F_TEST_RND_HI32 flag verifier inserts random value for hi32 bits of (2) (marked (r)); - this random value is read at (5).
5.13.0-19.195.15.0-100.1105.15.0-101.1115.15.0-102.1125.15.0-105.1155.15.0-106.1165.15.0-107.1175.15.0-112.1225.15.0-113.1235.15.0-116.126+59 more5.15.0-135.1465.3.0-18.195.3.0-24.265.4.0-9.126.5.0-9.96.6.0-14.146.8.0-11.116.8.0-20.206.8.0-22.226.8.0-28.286.8.0-31.316.8.0-35.356.8.0-36.366.8.0-38.38+18 more6.8.0-60.636.11.0-8.86.12.0-12.124.2.0-16.194.2.0-17.214.2.0-19.234.3.0-1.104.3.0-2.114.3.0-5.164.3.0-6.174.3.0-7.184.4.0-2.164.13.0-16.193.11.0-12.195.19.0-1007.7~22.04.15.19.0-1009.9~22.04.15.19.0-1010.10~22.04.15.19.0-1011.11~22.04.15.19.0-1012.12~22.04.15.19.0-1013.13~22.04.15.19.0-1014.14~22.04.15.19.0-1015.15~22.04.15.13.0-1005.65.15.0-1002.45.15.0-1003.55.15.0-1004.65.15.0-1005.75.15.0-1008.105.15.0-1009.115.15.0-1011.145.15.0-1013.175.15.0-1014.18+51 more5.15.0-1080.874.4.0-1002.2