bpf: allow adjusted map element values to spill
authorGianluca Borello <g.borello@gmail.com>
Mon, 9 Jan 2017 18:19:48 +0000 (10:19 -0800)
committerDavid S. Miller <davem@davemloft.net>
Mon, 9 Jan 2017 21:56:27 +0000 (16:56 -0500)
commit 484611357c19 ("bpf: allow access into map value arrays")
introduces the ability to do pointer math inside a map element value via
the PTR_TO_MAP_VALUE_ADJ register type.

The current support doesn't handle the case where a PTR_TO_MAP_VALUE_ADJ
is spilled into the stack, limiting several use cases, especially when
generating bpf code from a compiler.

Handle this case by explicitly enabling the register type
PTR_TO_MAP_VALUE_ADJ to be spilled. Also, make sure that min_value and
max_value are reset just for BPF_LDX operations that don't result in a
restore of a spilled register from stack.

Signed-off-by: Gianluca Borello <g.borello@gmail.com>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
kernel/bpf/verifier.c
tools/testing/selftests/bpf/test_verifier.c

index b7014606745b7772c1a844ab5afe74ad88143e5b..59ed07b9b4eaace5e46472f940d0b38f5bb73912 100644 (file)
@@ -481,6 +481,13 @@ static void reset_reg_range_values(struct bpf_reg_state *regs, u32 regno)
        regs[regno].max_value = BPF_REGISTER_MAX_RANGE;
 }
 
+static void mark_reg_unknown_value_and_range(struct bpf_reg_state *regs,
+                                            u32 regno)
+{
+       mark_reg_unknown_value(regs, regno);
+       reset_reg_range_values(regs, regno);
+}
+
 enum reg_arg_type {
        SRC_OP,         /* register is used as source operand */
        DST_OP,         /* register is used as destination operand */
@@ -532,6 +539,7 @@ static bool is_spillable_regtype(enum bpf_reg_type type)
        switch (type) {
        case PTR_TO_MAP_VALUE:
        case PTR_TO_MAP_VALUE_OR_NULL:
+       case PTR_TO_MAP_VALUE_ADJ:
        case PTR_TO_STACK:
        case PTR_TO_CTX:
        case PTR_TO_PACKET:
@@ -616,7 +624,8 @@ static int check_stack_read(struct bpf_verifier_state *state, int off, int size,
                }
                if (value_regno >= 0)
                        /* have read misc data from the stack */
-                       mark_reg_unknown_value(state->regs, value_regno);
+                       mark_reg_unknown_value_and_range(state->regs,
+                                                        value_regno);
                return 0;
        }
 }
@@ -825,7 +834,8 @@ static int check_mem_access(struct bpf_verifier_env *env, u32 regno, int off,
                else
                        err = check_map_access(env, regno, off, size);
                if (!err && t == BPF_READ && value_regno >= 0)
-                       mark_reg_unknown_value(state->regs, value_regno);
+                       mark_reg_unknown_value_and_range(state->regs,
+                                                        value_regno);
 
        } else if (reg->type == PTR_TO_CTX) {
                enum bpf_reg_type reg_type = UNKNOWN_VALUE;
@@ -837,7 +847,8 @@ static int check_mem_access(struct bpf_verifier_env *env, u32 regno, int off,
                }
                err = check_ctx_access(env, off, size, t, &reg_type);
                if (!err && t == BPF_READ && value_regno >= 0) {
-                       mark_reg_unknown_value(state->regs, value_regno);
+                       mark_reg_unknown_value_and_range(state->regs,
+                                                        value_regno);
                        /* note that reg.[id|off|range] == 0 */
                        state->regs[value_regno].type = reg_type;
                }
@@ -870,7 +881,8 @@ static int check_mem_access(struct bpf_verifier_env *env, u32 regno, int off,
                }
                err = check_packet_access(env, regno, off, size);
                if (!err && t == BPF_READ && value_regno >= 0)
-                       mark_reg_unknown_value(state->regs, value_regno);
+                       mark_reg_unknown_value_and_range(state->regs,
+                                                        value_regno);
        } else {
                verbose("R%d invalid mem access '%s'\n",
                        regno, reg_type_str[reg->type]);
@@ -2744,7 +2756,6 @@ static int do_check(struct bpf_verifier_env *env)
                        if (err)
                                return err;
 
-                       reset_reg_range_values(regs, insn->dst_reg);
                        if (BPF_SIZE(insn->code) != BPF_W &&
                            BPF_SIZE(insn->code) != BPF_DW) {
                                insn_idx++;
index b7732e557bf9e005b1cfbb4815762c86900757ce..e7b075819c08a40eb93ccf4c9c7c82b810321cd3 100644 (file)
@@ -3396,6 +3396,52 @@ static struct bpf_test tests[] = {
                .result = REJECT,
                .prog_type = BPF_PROG_TYPE_TRACEPOINT,
        },
+       {
+               "map element value is preserved across register spilling",
+               .insns = {
+                       BPF_MOV64_REG(BPF_REG_2, BPF_REG_10),
+                       BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, -8),
+                       BPF_ST_MEM(BPF_DW, BPF_REG_2, 0, 0),
+                       BPF_LD_MAP_FD(BPF_REG_1, 0),
+                       BPF_EMIT_CALL(BPF_FUNC_map_lookup_elem),
+                       BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 6),
+                       BPF_ST_MEM(BPF_DW, BPF_REG_0, 0, 42),
+                       BPF_MOV64_REG(BPF_REG_1, BPF_REG_10),
+                       BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, -184),
+                       BPF_STX_MEM(BPF_DW, BPF_REG_1, BPF_REG_0, 0),
+                       BPF_LDX_MEM(BPF_DW, BPF_REG_3, BPF_REG_1, 0),
+                       BPF_ST_MEM(BPF_DW, BPF_REG_3, 0, 42),
+                       BPF_EXIT_INSN(),
+               },
+               .fixup_map2 = { 3 },
+               .errstr_unpriv = "R0 leaks addr",
+               .result = ACCEPT,
+               .result_unpriv = REJECT,
+       },
+       {
+               "map element value (adjusted) is preserved across register spilling",
+               .insns = {
+                       BPF_MOV64_REG(BPF_REG_2, BPF_REG_10),
+                       BPF_ALU64_IMM(BPF_ADD, BPF_REG_2, -8),
+                       BPF_ST_MEM(BPF_DW, BPF_REG_2, 0, 0),
+                       BPF_LD_MAP_FD(BPF_REG_1, 0),
+                       BPF_EMIT_CALL(BPF_FUNC_map_lookup_elem),
+                       BPF_JMP_IMM(BPF_JEQ, BPF_REG_0, 0, 7),
+                       BPF_ALU64_IMM(BPF_ADD, BPF_REG_0,
+                               offsetof(struct test_val, foo)),
+                       BPF_ST_MEM(BPF_DW, BPF_REG_0, 0, 42),
+                       BPF_MOV64_REG(BPF_REG_1, BPF_REG_10),
+                       BPF_ALU64_IMM(BPF_ADD, BPF_REG_1, -184),
+                       BPF_STX_MEM(BPF_DW, BPF_REG_1, BPF_REG_0, 0),
+                       BPF_LDX_MEM(BPF_DW, BPF_REG_3, BPF_REG_1, 0),
+                       BPF_ST_MEM(BPF_DW, BPF_REG_3, 0, 42),
+                       BPF_EXIT_INSN(),
+               },
+               .fixup_map2 = { 3 },
+               .errstr_unpriv = "R0 pointer arithmetic prohibited",
+               .result = ACCEPT,
+               .result_unpriv = REJECT,
+       },
 };
 
 static int probe_filter_length(const struct bpf_insn *fp)