bpf: enforce precision of R0 on callback return
authorAndrii Nakryiko <andrii@kernel.org>
Sat, 2 Dec 2023 17:56:57 +0000 (09:56 -0800)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 25 Jan 2024 23:35:20 +0000 (15:35 -0800)
[ Upstream commit 0acd03a5bd188b0c501d285d938439618bd855c4 ]

Given verifier checks actual value, r0 has to be precise, so we need to
propagate precision properly. r0 also has to be marked as read,
otherwise subsequent state comparisons will ignore such register as
unimportant and precision won't really help here.

Fixes: 69c087ba6225 ("bpf: Add bpf_for_each_map_elem() helper")
Acked-by: Eduard Zingerman <eddyz87@gmail.com>
Acked-by: Shung-Hsi Yu <shung-hsi.yu@suse.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20231202175705.885270-4-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
kernel/bpf/verifier.c

index 824531d..2deae0d 100644 (file)
@@ -9284,6 +9284,13 @@ static int prepare_func_exit(struct bpf_verifier_env *env, int *insn_idx)
                        verbose(env, "R0 not a scalar value\n");
                        return -EACCES;
                }
+
+               /* we are going to rely on register's precise value */
+               err = mark_reg_read(env, r0, r0->parent, REG_LIVE_READ64);
+               err = err ?: mark_chain_precision(env, BPF_REG_0);
+               if (err)
+                       return err;
+
                if (!tnum_in(range, r0->var_off)) {
                        verbose_invalid_scalar(env, r0, &range, "callback return", "R0");
                        return -EINVAL;