bpf: fix branch pruning logic

PATCH In review Alexei Starovoitov 2 participants 2 messages Nov 23, 2017

Subsystems (4)

Networking Network drivers BPF BPF core

lore

Share & tools

lore

Alexei Starovoitov

author

2017-11-23 00:42 UTC #0 lore raw

PATCH v2 1/1 bpf: fix branch pruning logic

when the verifier detects that register contains a runtime constant and it's compared with another constant it will prune exploration of the branch that is guaranteed not to be taken at runtime. This is all correct, but malicious program may be constructed in such a way that it always has a constant comparison and the other branch is never taken under any conditions.
In this case such path through the program will not be explored by the verifier. It won't be taken at run-time either, but since all instructions are JITed the malicious program may cause JITs to complain about using reserved fields, etc.
To fix the issue we have to track the instructions explored by the verifier and sanitize instructions that are dead at run time with NOPs. We cannot reject such dead code, since llvm generates it for valid C code, since it doesn't do as much data flow analysis as the verifier does.

Fixes 17a5267067f3 ("bpf: verifier (add verifier core)")

Signed-off-by Alexei Starovoitov <[email protected]>

Acked-by Daniel Borkmann <[email protected]>

---
v1->v2: made sanitize_dead_code() conditional. Only do it when program was successfully validated, since broken progs will be freed immediately and no need to spend time to clear insns.

for net-next we might try to remove dead code and adjust all branches instead of replacing with nops
Implementation detail: converted_op_size is unused. We can reuse that space.

---
 include/linux/bpf_verifier.h |  2 +-
 kernel/bpf/verifier.c        | 27 +++++++++++++++++++++++++++
 2 files changed, 28 insertions(+), 1 deletion(-)

include/linux/bpf_verifier.h +1 -1

index 07b96aaca256..7b418f0a62f6 100644
@@ -115,7 +115,7 @@ struct bpf_insn_aux_data {
 		struct bpf_map *map_ptr;	/* pointer for call insn into lookup_elem */
 	};
 	int ctx_field_size; /* the ctx field size for load insn, maybe 0 */
-	int converted_op_size; /* the valid value width after perceived conversion */
+	bool seen; /* this insn was processed by the verifier */
 };
 
 #define MAX_USED_MAPS 64 /* max number of maps accessed by one eBPF program */

kernel/bpf/verifier.c +27 -0

index dd54d20ace2f..0a34594dab96 100644
@@ -3825,6 +3825,7 @@ static int do_check(struct bpf_verifier_env *env)
 			return err;
 
 		regs = cur_regs(env);
+		env->insn_aux_data[insn_idx].seen = true;
 		if (class == BPF_ALU || class == BPF_ALU64) {
 			err = check_alu_op(env, insn);
 			if (err)

@@ -4020,6 +4021,7 @@ static int do_check(struct bpf_verifier_env *env)
 					return err;
 
 				insn_idx++;
+				env->insn_aux_data[insn_idx].seen = true;
 			} else {
 				verbose(env, "invalid BPF_LD mode\n");
 				return -EINVAL;

@@ -4202,6 +4204,7 @@ static int adjust_insn_aux_data(struct bpf_verifier_env *env, u32 prog_len,
 				u32 off, u32 cnt)
 {
 	struct bpf_insn_aux_data *new_data, *old_data = env->insn_aux_data;
+	int i;
 
 	if (cnt == 1)
 		return 0;

@@ -4211,6 +4214,8 @@ static int adjust_insn_aux_data(struct bpf_verifier_env *env, u32 prog_len,
 	memcpy(new_data, old_data, sizeof(struct bpf_insn_aux_data) * off);
 	memcpy(new_data + off + cnt - 1, old_data + off,
 	       sizeof(struct bpf_insn_aux_data) * (prog_len - off - cnt + 1));
+	for (i = off; i < off + cnt - 1; i++)
+		new_data[i].seen = true;
 	env->insn_aux_data = new_data;
 	vfree(old_data);
 	return 0;

@@ -4229,6 +4234,25 @@ static struct bpf_prog *bpf_patch_insn_data(struct bpf_verifier_env *env, u32 of
 	return new_prog;
 }
 
+/* The verifier does more data flow analysis than llvm and will not explore
+ * branches that are dead at run time. Malicious programs can have dead code
+ * too. Therefore replace all dead at-run-time code with nops.
+ */
+static void sanitize_dead_code(struct bpf_verifier_env *env)
+{
+	struct bpf_insn_aux_data *aux_data = env->insn_aux_data;
+	struct bpf_insn nop = BPF_MOV64_REG(BPF_REG_0, BPF_REG_0);
+	struct bpf_insn *insn = env->prog->insnsi;
+	const int insn_cnt = env->prog->len;
+	int i;
+
+	for (i = 0; i < insn_cnt; i++) {
+		if (aux_data[i].seen)
+			continue;
+		memcpy(insn + i, &nop, sizeof(nop));
+	}
+}
+
 /* convert load instructions that access fields of 'struct __sk_buff'
  * into sequence of instructions that access fields of 'struct sk_buff'
  */

@@ -4556,6 +4580,9 @@ int bpf_check(struct bpf_prog **prog, union bpf_attr *attr)
 	free_states(env);
 
 	if (ret == 0)
+		sanitize_dead_code(env);
+
+	if (ret == 0)
 		/* program is valid, convert *(u32*)(ctx + off) accesses */
 		ret = convert_ctx_accesses(env);

Signature · 1 line

2.9.5

Daniel Borkmann

maintainer

2017-11-23 10:14 UTC #1 lore raw

Quoted 19 lines · Alexei Starovoitov

when the verifier detects that register contains a runtime constant and it's compared with another constant it will prune exploration of the branch that is guaranteed not to be taken at runtime. This is all correct, but malicious program may be constructed in such a way that it always has a constant comparison and the other branch is never taken under any conditions.
In this case such path through the program will not be explored by the verifier. It won't be taken at run-time either, but since all instructions are JITed the malicious program may cause JITs to complain about using reserved fields, etc.
To fix the issue we have to track the instructions explored by the verifier and sanitize instructions that are dead at run time with NOPs. We cannot reject such dead code, since llvm generates it for valid C code, since it doesn't do as much data flow analysis as the verifier does.
Fixes 17a5267067f3 ("bpf: verifier (add verifier core)")
Signed-off-by Alexei Starovoitov <[email protected]>
Acked-by Daniel Borkmann <[email protected]>

Quoted 8 lines

---
v1->v2: made sanitize_dead_code() conditional. Only do it when program was successfully validated, since broken progs will be freed immediately and no need to spend time to clear insns.
for net-next we might try to remove dead code and adjust all branches instead of replacing with nops
Implementation detail: converted_op_size is unused. We can reuse that space.

Thanks for following up, Alexei. Applied to bpf tree and queued up for stable.