Re: [PATCH bpf-next v3 4/9] bpf/verifier: improve register value range tracking with ARSH

[Yonghong Song <yhs@fb.com>]

On 4/22/18 5:16 PM, Alexei Starovoitov wrote:
> On Fri, Apr 20, 2018 at 03:18:37PM -0700, Yonghong Song wrote:
>> When helpers like bpf_get_stack returns an int value
>> and later on used for arithmetic computation, the LSH and ARSH
>> operations are often required to get proper sign extension into
>> 64-bit. For example, without this patch:
>>      54: R0=inv(id=0,umax_value=800)
>>      54: (bf) r8 = r0
>>      55: R0=inv(id=0,umax_value=800) R8_w=inv(id=0,umax_value=800)
>>      55: (67) r8 <<= 32
>>      56: R8_w=inv(id=0,umax_value=3435973836800,var_off=(0x0; 0x3ff00000000))
>>      56: (c7) r8 s>>= 32
>>      57: R8=inv(id=0)
>> With this patch:
>>      54: R0=inv(id=0,umax_value=800)
>>      54: (bf) r8 = r0
>>      55: R0=inv(id=0,umax_value=800) R8_w=inv(id=0,umax_value=800)
>>      55: (67) r8 <<= 32
>>      56: R8_w=inv(id=0,umax_value=3435973836800,var_off=(0x0; 0x3ff00000000))
>>      56: (c7) r8 s>>= 32
>>      57: R8=inv(id=0, umax_value=800,var_off=(0x0; 0x3ff))
>> With better range of "R8", later on when "R8" is added to other register,
>> e.g., a map pointer or scalar-value register, the better register
>> range can be derived and verifier failure may be avoided.
>>
>> In our later example,
>>      ......
>>      usize = bpf_get_stack(ctx, raw_data, max_len, BPF_F_USER_STACK);
>>      if (usize < 0)
>>          return 0;
>>      ksize = bpf_get_stack(ctx, raw_data + usize, max_len - usize, 0);
>>      ......
>> Without improving ARSH value range tracking, the register representing
>> "max_len - usize" will have smin_value equal to S64_MIN and will be
>> rejected by verifier.
>>
>> Signed-off-by: Yonghong Song <yhs@fb.com>
>> ---
>>   kernel/bpf/verifier.c | 26 ++++++++++++++++++++++++++
>>   1 file changed, 26 insertions(+)
>>
>> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
>> index 3c8bb92..01c215d 100644
>> --- a/kernel/bpf/verifier.c
>> +++ b/kernel/bpf/verifier.c
>> @@ -2975,6 +2975,32 @@ static int adjust_scalar_min_max_vals(struct bpf_verifier_env *env,
>>   		/* We may learn something more from the var_off */
>>   		__update_reg_bounds(dst_reg);
>>   		break;
>> +	case BPF_ARSH:
>> +		if (umax_val >= insn_bitness) {
>> +			/* Shifts greater than 31 or 63 are undefined.
>> +			 * This includes shifts by a negative number.
>> +			 */
>> +			mark_reg_unknown(env, regs, insn->dst_reg);
>> +			break;
>> +		}
>> +		if (dst_reg->smin_value < 0)
>> +			dst_reg->smin_value >>= umin_val;
>> +		else
>> +			dst_reg->smin_value >>= umax_val;
>> +		if (dst_reg->smax_value < 0)
>> +			dst_reg->smax_value >>= umax_val;
>> +		else
>> +			dst_reg->smax_value >>= umin_val;
>> +		if (src_known)
>> +			dst_reg->var_off = tnum_rshift(dst_reg->var_off,
>> +						       umin_val);
>> +		else
>> +			dst_reg->var_off = tnum_rshift(tnum_unknown, umin_val);
>> +		dst_reg->umin_value >>= umax_val;
>> +		dst_reg->umax_value >>= umin_val;
>> +		/* We may learn something more from the var_off */
>> +		__update_reg_bounds(dst_reg);
> 
> I'm struggling to understand how these bounds are computed.
> Could you add examples in the comments?

Okay, let me try to add some comments for better understanding.

> In particular if dst_reg is unknown (tnum.mask == -1)
> the above tnum_rshift() will clear upper bits and will make it
> 64-bit positive, but that doesn't seem correct.
> What am I missing?

Considering this is arith shift, we probably should just have
dst_reg->var_off = tnum_unknown to be conservative.

I could miss something here as well. Let me try to write more
detailed explanation, hopefully to cover all corner cases.