[bpf-next,v3,06/11] bpf: Fix compare error in function retval_range_within

Commit Message

Xu Kuohai April 11, 2024, 12:27 p.m. UTC

From: Xu Kuohai <[e-mail]>

After checking lsm hook return range in verifier, the test case
"test_progs -t test_lsm" failed, and the failure log says:

libbpf: prog 'test_int_hook': BPF program load failed: Invalid argument
libbpf: prog 'test_int_hook': -- BEGIN PROG LOAD LOG --
0: R1=ctx() R10=fp0
; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
0: (79) r0 = *(u64 *)(r1 +24)         ; R0_w=scalar(smin=smin32=-4095,smax=smax32=0) R1=ctx()

[...]

24: (b4) w0 = -1                      ; R0_w=0xffffffff
; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
25: (95) exit
At program exit the register R0 has smin=4294967295 smax=4294967295 should have been in [-4095, 0]

It can be seen that instruction "w0 = -1" zero extended -1 to 64-bit
register r0, setting both smin and smax values of r0 to 4294967295.
This resulted in a false reject when r0 was checked with range [-4095, 0].

Given bpf_retval_range is a 32-bit range, this patch fixes it by
changing the compare between r0 and return range from 64-bit
operation to 32-bit operation.

Fixes: 8fa4ecd49b81 ("bpf: enforce exact retval range on subprog/callback exit")
Signed-off-by: Xu Kuohai <[e-mail]>
---
 kernel/bpf/verifier.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

Comments

Shung-Hsi Yu April 12, 2024, 8:53 a.m. UTC | #1

On Thu, Apr 11, 2024 at 08:27:47PM +0800, Xu Kuohai wrote:
> [...]
> 24: (b4) w0 = -1                      ; R0_w=0xffffffff
> ; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
> 25: (95) exit
> At program exit the register R0 has smin=4294967295 smax=4294967295 should have been in [-4095, 0]
> 
> It can be seen that instruction "w0 = -1" zero extended -1 to 64-bit
> register r0, setting both smin and smax values of r0 to 4294967295.
> This resulted in a false reject when r0 was checked with range [-4095, 0].
> 
> Given bpf_retval_range is a 32-bit range, this patch fixes it by
> changing the compare between r0 and return range from 64-bit
> operation to 32-bit operation.
> 
> Fixes: 8fa4ecd49b81 ("bpf: enforce exact retval range on subprog/callback exit")
> Signed-off-by: Xu Kuohai <[e-mail]>
> ---
>  kernel/bpf/verifier.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 05c7c5f2bec0..5393d576c76f 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -9879,7 +9879,7 @@ static bool in_rbtree_lock_required_cb(struct bpf_verifier_env *env)
>  
>  static bool retval_range_within(struct bpf_retval_range range, const struct bpf_reg_state *reg)
>  {
> -	return range.minval <= reg->smin_value && reg->smax_value <= range.maxval;
> +	return range.minval <= reg->s32_min_value && reg->s32_max_value <= range.maxval;

Logic-wise LGTM

While the status-quo is that the return value is always truncated to
32-bit, looking back there was an attempt to use 64-bit return value for
bpf_prog_run[1] (not merged due to issue on 32-bit architectures). Also
from the reading of BPF standardization ABI it would be inferred that
return value is in 64-bit range:

  BPF has 10 general purpose registers and a read-only frame pointer register,
  all of which are 64-bits wide.
  
  The BPF calling convention is defined as:
  
  * R0: return value from function calls, and exit value for BPF programs
  ...

So add relevant people into the thread for opinions.

1: https://lore.kernel.org/bpf/20221115193911.u6prvskdzr5jevni@apollo/

Andrii Nakryiko April 25, 2024, 11:41 p.m. UTC | #2

On Thu, Apr 11, 2024 at 5:24 AM Xu Kuohai <[e-mail]> wrote:
>
> From: Xu Kuohai <[e-mail]>
>
> After checking lsm hook return range in verifier, the test case
> "test_progs -t test_lsm" failed, and the failure log says:
>
> libbpf: prog 'test_int_hook': BPF program load failed: Invalid argument
> libbpf: prog 'test_int_hook': -- BEGIN PROG LOAD LOG --
> 0: R1=ctx() R10=fp0
> ; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
> 0: (79) r0 = *(u64 *)(r1 +24)         ; R0_w=scalar(smin=smin32=-4095,smax=smax32=0) R1=ctx()
>
> [...]
>
> 24: (b4) w0 = -1                      ; R0_w=0xffffffff
> ; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
> 25: (95) exit
> At program exit the register R0 has smin=4294967295 smax=4294967295 should have been in [-4095, 0]
>
> It can be seen that instruction "w0 = -1" zero extended -1 to 64-bit
> register r0, setting both smin and smax values of r0 to 4294967295.
> This resulted in a false reject when r0 was checked with range [-4095, 0].
>
> Given bpf_retval_range is a 32-bit range, this patch fixes it by
> changing the compare between r0 and return range from 64-bit
> operation to 32-bit operation.
>
> Fixes: 8fa4ecd49b81 ("bpf: enforce exact retval range on subprog/callback exit")
> Signed-off-by: Xu Kuohai <[e-mail]>
> ---
>  kernel/bpf/verifier.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 05c7c5f2bec0..5393d576c76f 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -9879,7 +9879,7 @@ static bool in_rbtree_lock_required_cb(struct bpf_verifier_env *env)
>
>  static bool retval_range_within(struct bpf_retval_range range, const struct bpf_reg_state *reg)
>  {
> -       return range.minval <= reg->smin_value && reg->smax_value <= range.maxval;
> +       return range.minval <= reg->s32_min_value && reg->s32_max_value <= range.maxval;

are all BPF programs treated as if they return int instead of long? If
not, we probably should have a bool flag in bpf_retval_range whether
comparison should be 32-bit or 64-bit?

>  }
>
>  static int prepare_func_exit(struct bpf_verifier_env *env, int *insn_idx)
> --
> 2.30.2
>

Xu Kuohai April 26, 2024, 8:08 a.m. UTC | #3

On 4/26/2024 7:41 AM, Andrii Nakryiko wrote:
> On Thu, Apr 11, 2024 at 5:24 AM Xu Kuohai <[e-mail]> wrote:
>>
>> From: Xu Kuohai <[e-mail]>
>>
>> After checking lsm hook return range in verifier, the test case
>> "test_progs -t test_lsm" failed, and the failure log says:
>>
>> libbpf: prog 'test_int_hook': BPF program load failed: Invalid argument
>> libbpf: prog 'test_int_hook': -- BEGIN PROG LOAD LOG --
>> 0: R1=ctx() R10=fp0
>> ; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
>> 0: (79) r0 = *(u64 *)(r1 +24)         ; R0_w=scalar(smin=smin32=-4095,smax=smax32=0) R1=ctx()
>>
>> [...]
>>
>> 24: (b4) w0 = -1                      ; R0_w=0xffffffff
>> ; int BPF_PROG(test_int_hook, struct vm_area_struct *vma, @ lsm.c:89
>> 25: (95) exit
>> At program exit the register R0 has smin=4294967295 smax=4294967295 should have been in [-4095, 0]
>>
>> It can be seen that instruction "w0 = -1" zero extended -1 to 64-bit
>> register r0, setting both smin and smax values of r0 to 4294967295.
>> This resulted in a false reject when r0 was checked with range [-4095, 0].
>>
>> Given bpf_retval_range is a 32-bit range, this patch fixes it by
>> changing the compare between r0 and return range from 64-bit
>> operation to 32-bit operation.
>>
>> Fixes: 8fa4ecd49b81 ("bpf: enforce exact retval range on subprog/callback exit")
>> Signed-off-by: Xu Kuohai <[e-mail]>
>> ---
>>   kernel/bpf/verifier.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
>> index 05c7c5f2bec0..5393d576c76f 100644
>> --- a/kernel/bpf/verifier.c
>> +++ b/kernel/bpf/verifier.c
>> @@ -9879,7 +9879,7 @@ static bool in_rbtree_lock_required_cb(struct bpf_verifier_env *env)
>>
>>   static bool retval_range_within(struct bpf_retval_range range, const struct bpf_reg_state *reg)
>>   {
>> -       return range.minval <= reg->smin_value && reg->smax_value <= range.maxval;
>> +       return range.minval <= reg->s32_min_value && reg->s32_max_value <= range.maxval;
> 
> are all BPF programs treated as if they return int instead of long? If
> not, we probably should have a bool flag in bpf_retval_range whether
> comparison should be 32-bit or 64-bit?
>

It seems that when a fmod_return prog is attached to a kernel function
that returns long value, the bpf prog should also return long value.
To confirm it, I'll try to find an example or construct a case for this.

>>   }
>>
>>   static int prepare_func_exit(struct bpf_verifier_env *env, int *insn_idx)
>> --
>> 2.30.2
>>
>

Patch

diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 05c7c5f2bec0..5393d576c76f 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -9879,7 +9879,7 @@  static bool in_rbtree_lock_required_cb(struct bpf_verifier_env *env)
 
 static bool retval_range_within(struct bpf_retval_range range, const struct bpf_reg_state *reg)
 {
-	return range.minval <= reg->smin_value && reg->smax_value <= range.maxval;
+	return range.minval <= reg->s32_min_value && reg->s32_max_value <= range.maxval;
 }
 
 static int prepare_func_exit(struct bpf_verifier_env *env, int *insn_idx)