CVE-2024-42063

In the Linux kernel, the following vulnerability has been resolved:<br /> <br /> bpf: Mark bpf prog stack with kmsan_unposion_memory in interpreter mode<br /> <br /> syzbot reported uninit memory usages during map_{lookup,delete}_elem.<br /> <br /> ==========<br /> BUG: KMSAN: uninit-value in __dev_map_lookup_elem kernel/bpf/devmap.c:441 [inline]<br /> BUG: KMSAN: uninit-value in dev_map_lookup_elem+0xf3/0x170 kernel/bpf/devmap.c:796<br /> __dev_map_lookup_elem kernel/bpf/devmap.c:441 [inline]<br /> dev_map_lookup_elem+0xf3/0x170 kernel/bpf/devmap.c:796<br /> ____bpf_map_lookup_elem kernel/bpf/helpers.c:42 [inline]<br /> bpf_map_lookup_elem+0x5c/0x80 kernel/bpf/helpers.c:38<br /> ___bpf_prog_run+0x13fe/0xe0f0 kernel/bpf/core.c:1997<br /> __bpf_prog_run256+0xb5/0xe0 kernel/bpf/core.c:2237<br /> ==========<br /> <br /> The reproducer should be in the interpreter mode.<br /> <br /> The C reproducer is trying to run the following bpf prog:<br /> <br /> 0: (18) r0 = 0x0<br /> 2: (18) r1 = map[id:49]<br /> 4: (b7) r8 = 16777216<br /> 5: (7b) *(u64 *)(r10 -8) = r8<br /> 6: (bf) r2 = r10<br /> 7: (07) r2 += -229<br /> ^^^^^^^^^^<br /> <br /> 8: (b7) r3 = 8<br /> 9: (b7) r4 = 0<br /> 10: (85) call dev_map_lookup_elem#1543472<br /> 11: (95) exit<br /> <br /> It is due to the "void *key" (r2) passed to the helper. bpf allows uninit<br /> stack memory access for bpf prog with the right privileges. This patch<br /> uses kmsan_unpoison_memory() to mark the stack as initialized.<br /> <br /> This should address different syzbot reports on the uninit "void *key"<br /> argument during map_{lookup,delete}_elem.