bpf: Reset register bounds before narrowing retval range in check_mem_access()
Subsystems (3)
PATCH v2 1/1 bpf: Reset register bounds before narrowing retval range in check_mem_access()
From: Tristan Madani <[email protected]>
When the BPF verifier processes a context load of an LSM hook return value, it calls __mark_reg_s32_range() to narrow the register to the hook's valid range. However, __mark_reg_s32_range() intersects the new range with the register's existing bounds using max_t()/min_t() rather than replacing them.
If the destination register carries stale bounds from a prior instruction (e.g. BPF_MOV64_IMM), the intersection can produce a range narrower than reality. The verifier then believes it knows the register's exact value, while at runtime the actual hook return value is loaded, creating a verifier/runtime mismatch that can be used to bypass BPF memory safety checks.
The else branch already calls mark_reg_unknown() to reset register state before any narrowing. Apply the same reset in the is_retval path so stale bounds are cleared before __mark_reg_s32_range() intersects.
---
Changes in v2:
- Use [PATCH bpf] subject prefix per Jiri Olsa review
- Rebased on bpf/master
- No code change from v1
kernel/bpf/verifier.c | 1 + 1 file changed, 1 insertion(+)
kernel/bpf/verifier.c +1 -0
@@ -6196,6 +6196,7 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, struct b*/ if (info.reg_type ==SCALAR_VALUE ) {if (info.is_retval &&get_func_retval_range (env->prog, &range)) { +mark_reg_unknown (env, regs, value_regno); err =__mark_reg_s32_range (env, regs, value_regno, range.minval, range.maxval);if (err)
Signature · 1 line
2.47.3
Eduard Zingerman
maintainerQuoted 36 lines · Tristan Madani
From: Tristan Madani <[email protected]>
When the BPF verifier processes a context load of an LSM hook return value, it calls
__mark_reg_s32_range()to narrow the register to the hook's valid range. However,__mark_reg_s32_range()intersects the new range with the register's existing bounds using max_t()/min_t() rather than replacing them.If the destination register carries stale bounds from a prior instruction (e.g.
BPF_MOV64_IMM), the intersection can produce a range narrower than reality. The verifier then believes it knows the register's exact value, while at runtime the actual hook return value is loaded, creating a verifier/runtime mismatch that can be used to bypass BPF memory safety checks.The else branch already calls
mark_reg_unknown()to reset register state before any narrowing. Apply the same reset in theis_retvalpath so stale bounds are cleared before__mark_reg_s32_range()intersects.Fixes 5d99e198be27 ("bpf, lsm: Add check for BPF LSM return value")Signed-off-by Tristan Madani <[email protected]>---
Changes in v2:
- Use [PATCH bpf] subject prefix per Jiri Olsa review
- Rebased on bpf/master
- No code change from v1kernel/bpf/verifier.c | 1 + 1 file changed, 1 insertion(+)kernel/bpf/verifier.c +0
-0
kernel/bpf/verifier.c static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, struct b · line 6199
Quoted 4 lines
kernel/bpf/verifier.c +1
-0@@ -6196,6 +6196,7 @@ static int check_mem_access(struct bpf_verifier_env *env, int insn_idx, struct b*/ if (info.reg_type ==SCALAR_VALUE ) {if (info.is_retval &&get_func_retval_range (env->prog, &range)) { +mark_reg_unknown (env, regs, value_regno);
This seem to be a real bug, could you please add a selftest demonstrating the issue and post v3?
Something like this in the tools/testing/selftests/bpf/progs/verifier_lsm.c:
/* lsm bpf prog retval load must reset stale register bounds */ SEC("lsm/file_mprotect")__failure __msg("div by zero")__naked int retval_load_resets_bounds(void)
{
asm volatile (
"r6 = 0;"
"r6 = *(u64 *)(r1 + 24);" /* return value of the LSM_HOOK(...file_mprotect...) */
"if r6 == 0 goto +1;" /* r6 value should not be predictable here */
"r6 /= 0;"
"r0 = 0;"
"exit;"
::: __clobber_all);
}Quoted 3 lines
kernel/bpf/verifier.c +0
-0err =__mark_reg_s32_range (env, regs, value_regno, range.minval, range.maxval);if (err)