* [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
@ 2023-12-10 13:00 Menglong Dong
2023-12-11 5:09 ` Yonghong Song
2023-12-11 19:15 ` Andrii Nakryiko
0 siblings, 2 replies; 7+ messages in thread
From: Menglong Dong @ 2023-12-10 13:00 UTC (permalink / raw)
To: andrii
Cc: ast, daniel, john.fastabend, martin.lau, song, yonghong.song,
kpsingh, sdf, haoluo, jolsa, bpf, linux-kernel, Menglong Dong
We can derive some new information for BPF_JNE in regs_refine_cond_op().
Take following code for example:
/* The type of "a" is u16 */
if (a > 0 && a < 100) {
/* the range of the register for a is [0, 99], not [1, 99],
* and will cause the following error:
*
* invalid zero-sized read
*
* as a can be 0.
*/
bpf_skb_store_bytes(skb, xx, xx, a, 0);
}
In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
TRUE branch, the dst_reg will be marked as known to 0. However, in the
fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
the [min, max] for a is [0, 99], not [1, 99].
For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
const and is exactly the edge of the dst reg.
Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
---
kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
1 file changed, 44 insertions(+), 1 deletion(-)
diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
index 727a59e4a647..7b074ac93190 100644
--- a/kernel/bpf/verifier.c
+++ b/kernel/bpf/verifier.c
@@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
reg->type = SCALAR_VALUE;
}
+#define CHECK_REG_MIN(value) \
+do { \
+ if ((value) == (typeof(value))imm) \
+ value++; \
+} while (0)
+
+#define CHECK_REG_MAX(value) \
+do { \
+ if ((value) == (typeof(value))imm) \
+ value--; \
+} while (0)
+
+static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
+{
+ CHECK_REG_MIN(reg->s32_min_value);
+ CHECK_REG_MAX(reg->s32_max_value);
+ CHECK_REG_MIN(reg->u32_min_value);
+ CHECK_REG_MAX(reg->u32_max_value);
+}
+
+static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
+{
+ CHECK_REG_MIN(reg->smin_value);
+ CHECK_REG_MAX(reg->smax_value);
+
+ CHECK_REG_MIN(reg->umin_value);
+ CHECK_REG_MAX(reg->umax_value);
+
+ CHECK_REG_MIN(reg->s32_min_value);
+ CHECK_REG_MAX(reg->s32_max_value);
+ CHECK_REG_MIN(reg->u32_min_value);
+ CHECK_REG_MAX(reg->u32_max_value);
+}
+
static void mark_reg_known_zero(struct bpf_verifier_env *env,
struct bpf_reg_state *regs, u32 regno)
{
@@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
}
break;
case BPF_JNE:
- /* we don't derive any new information for inequality yet */
+ /* try to recompute the bound of reg1 if reg2 is a const and
+ * is exactly the edge of reg1.
+ */
+ if (is_reg_const(reg2, is_jmp32)) {
+ val = reg_const_value(reg2, is_jmp32);
+ if (is_jmp32)
+ mark_reg32_not_equal(reg1, val);
+ else
+ mark_reg_not_equal(reg1, val);
+ }
break;
case BPF_JSET:
if (!is_reg_const(reg2, is_jmp32))
--
2.39.2
^ permalink raw reply related [flat|nested] 7+ messages in thread
* Re: [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
2023-12-10 13:00 [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs Menglong Dong
@ 2023-12-11 5:09 ` Yonghong Song
2023-12-11 9:39 ` Menglong Dong
2023-12-11 19:15 ` Andrii Nakryiko
1 sibling, 1 reply; 7+ messages in thread
From: Yonghong Song @ 2023-12-11 5:09 UTC (permalink / raw)
To: Menglong Dong, andrii
Cc: ast, daniel, john.fastabend, martin.lau, song, kpsingh, sdf,
haoluo, jolsa, bpf, linux-kernel
On 12/10/23 5:00 AM, Menglong Dong wrote:
> We can derive some new information for BPF_JNE in regs_refine_cond_op().
> Take following code for example:
>
> /* The type of "a" is u16 */
> if (a > 0 && a < 100) {
> /* the range of the register for a is [0, 99], not [1, 99],
> * and will cause the following error:
> *
> * invalid zero-sized read
> *
> * as a can be 0.
> */
> bpf_skb_store_bytes(skb, xx, xx, a, 0);
> }
Could you have a C test to demonstrate this example?
Also, you should have a set of inline asm code (progs/verifier*.c)
to test various cases as in mark_reg32_not_equal() and
mark_reg_not_equal().
>
> In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
> TRUE branch, the dst_reg will be marked as known to 0. However, in the
> fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
> the [min, max] for a is [0, 99], not [1, 99].
>
> For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
> const and is exactly the edge of the dst reg.
>
> Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
> ---
> kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 44 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 727a59e4a647..7b074ac93190 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
> reg->type = SCALAR_VALUE;
> }
>
> +#define CHECK_REG_MIN(value) \
> +do { \
> + if ((value) == (typeof(value))imm) \
> + value++; \
> +} while (0)
> +
> +#define CHECK_REG_MAX(value) \
> +do { \
> + if ((value) == (typeof(value))imm) \
> + value--; \
> +} while (0)
> +
> +static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
> +{
What if reg->s32_min_value == imm and reg->s32_max_value == imm?
Has this been handled in previous verifier logic?
> + CHECK_REG_MIN(reg->s32_min_value);
> + CHECK_REG_MAX(reg->s32_max_value);
> + CHECK_REG_MIN(reg->u32_min_value);
> + CHECK_REG_MAX(reg->u32_max_value);
> +}
> +
> +static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
> +{
> + CHECK_REG_MIN(reg->smin_value);
> + CHECK_REG_MAX(reg->smax_value);
> +
> + CHECK_REG_MIN(reg->umin_value);
> + CHECK_REG_MAX(reg->umax_value);
> +
> + CHECK_REG_MIN(reg->s32_min_value);
> + CHECK_REG_MAX(reg->s32_max_value);
> + CHECK_REG_MIN(reg->u32_min_value);
> + CHECK_REG_MAX(reg->u32_max_value);
> +}
> +
> static void mark_reg_known_zero(struct bpf_verifier_env *env,
> struct bpf_reg_state *regs, u32 regno)
> {
> @@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
> }
> break;
> case BPF_JNE:
> - /* we don't derive any new information for inequality yet */
> + /* try to recompute the bound of reg1 if reg2 is a const and
> + * is exactly the edge of reg1.
> + */
> + if (is_reg_const(reg2, is_jmp32)) {
> + val = reg_const_value(reg2, is_jmp32);
> + if (is_jmp32)
> + mark_reg32_not_equal(reg1, val);
> + else
> + mark_reg_not_equal(reg1, val);
> + }
> break;
> case BPF_JSET:
> if (!is_reg_const(reg2, is_jmp32))
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
2023-12-11 5:09 ` Yonghong Song
@ 2023-12-11 9:39 ` Menglong Dong
2023-12-11 15:03 ` Yonghong Song
0 siblings, 1 reply; 7+ messages in thread
From: Menglong Dong @ 2023-12-11 9:39 UTC (permalink / raw)
To: Yonghong Song
Cc: andrii, ast, daniel, john.fastabend, martin.lau, song, kpsingh,
sdf, haoluo, jolsa, bpf, linux-kernel
Hello,
On Mon, Dec 11, 2023 at 1:09 PM Yonghong Song <yonghong.song@linux.dev> wrote:
>
>
> On 12/10/23 5:00 AM, Menglong Dong wrote:
> > We can derive some new information for BPF_JNE in regs_refine_cond_op().
> > Take following code for example:
> >
> > /* The type of "a" is u16 */
> > if (a > 0 && a < 100) {
> > /* the range of the register for a is [0, 99], not [1, 99],
> > * and will cause the following error:
> > *
> > * invalid zero-sized read
> > *
> > * as a can be 0.
> > */
> > bpf_skb_store_bytes(skb, xx, xx, a, 0);
> > }
>
> Could you have a C test to demonstrate this example?
> Also, you should have a set of inline asm code (progs/verifier*.c)
> to test various cases as in mark_reg32_not_equal() and
> mark_reg_not_equal().
>
Yeah! I found that this part is tested in the test_progs/reg_bounds_crafted
too, and this commit failed that test case, which I should fix in the next
version.
> >
> > In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
> > TRUE branch, the dst_reg will be marked as known to 0. However, in the
> > fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
> > the [min, max] for a is [0, 99], not [1, 99].
> >
> > For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
> > const and is exactly the edge of the dst reg.
> >
> > Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
> > ---
> > kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
> > 1 file changed, 44 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 727a59e4a647..7b074ac93190 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
> > reg->type = SCALAR_VALUE;
> > }
> >
> > +#define CHECK_REG_MIN(value) \
> > +do { \
> > + if ((value) == (typeof(value))imm) \
> > + value++; \
> > +} while (0)
> > +
> > +#define CHECK_REG_MAX(value) \
> > +do { \
> > + if ((value) == (typeof(value))imm) \
> > + value--; \
> > +} while (0)
> > +
> > +static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
> > +{
>
> What if reg->s32_min_value == imm and reg->s32_max_value == imm?
> Has this been handled in previous verifier logic?
Will such a case happen? In current code path, the src reg is a const,
and the is_branch_taken() will return 0 or 1 if the
dst_reg->s32_min_value == dst_reg->s32_max_value.
Enn......maybe we can do more checking here in case that someone
calls this function in another place.
Thanks!
Menglong Dong
>
> > + CHECK_REG_MIN(reg->s32_min_value);
> > + CHECK_REG_MAX(reg->s32_max_value);
> > + CHECK_REG_MIN(reg->u32_min_value);
> > + CHECK_REG_MAX(reg->u32_max_value);
> > +}
> > +
> > +static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
> > +{
> > + CHECK_REG_MIN(reg->smin_value);
> > + CHECK_REG_MAX(reg->smax_value);
> > +
> > + CHECK_REG_MIN(reg->umin_value);
> > + CHECK_REG_MAX(reg->umax_value);
> > +
> > + CHECK_REG_MIN(reg->s32_min_value);
> > + CHECK_REG_MAX(reg->s32_max_value);
> > + CHECK_REG_MIN(reg->u32_min_value);
> > + CHECK_REG_MAX(reg->u32_max_value);
> > +}
> > +
> > static void mark_reg_known_zero(struct bpf_verifier_env *env,
> > struct bpf_reg_state *regs, u32 regno)
> > {
> > @@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
> > }
> > break;
> > case BPF_JNE:
> > - /* we don't derive any new information for inequality yet */
> > + /* try to recompute the bound of reg1 if reg2 is a const and
> > + * is exactly the edge of reg1.
> > + */
> > + if (is_reg_const(reg2, is_jmp32)) {
> > + val = reg_const_value(reg2, is_jmp32);
> > + if (is_jmp32)
> > + mark_reg32_not_equal(reg1, val);
> > + else
> > + mark_reg_not_equal(reg1, val);
> > + }
> > break;
> > case BPF_JSET:
> > if (!is_reg_const(reg2, is_jmp32))
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
2023-12-11 9:39 ` Menglong Dong
@ 2023-12-11 15:03 ` Yonghong Song
0 siblings, 0 replies; 7+ messages in thread
From: Yonghong Song @ 2023-12-11 15:03 UTC (permalink / raw)
To: Menglong Dong
Cc: andrii, ast, daniel, john.fastabend, martin.lau, song, kpsingh,
sdf, haoluo, jolsa, bpf, linux-kernel
On 12/11/23 1:39 AM, Menglong Dong wrote:
> Hello,
>
> On Mon, Dec 11, 2023 at 1:09 PM Yonghong Song <yonghong.song@linux.dev> wrote:
>>
>> On 12/10/23 5:00 AM, Menglong Dong wrote:
>>> We can derive some new information for BPF_JNE in regs_refine_cond_op().
>>> Take following code for example:
>>>
>>> /* The type of "a" is u16 */
>>> if (a > 0 && a < 100) {
>>> /* the range of the register for a is [0, 99], not [1, 99],
>>> * and will cause the following error:
>>> *
>>> * invalid zero-sized read
>>> *
>>> * as a can be 0.
>>> */
>>> bpf_skb_store_bytes(skb, xx, xx, a, 0);
>>> }
>> Could you have a C test to demonstrate this example?
>> Also, you should have a set of inline asm code (progs/verifier*.c)
>> to test various cases as in mark_reg32_not_equal() and
>> mark_reg_not_equal().
>>
> Yeah! I found that this part is tested in the test_progs/reg_bounds_crafted
> too, and this commit failed that test case, which I should fix in the next
> version.
>
>>> In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
>>> TRUE branch, the dst_reg will be marked as known to 0. However, in the
>>> fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
>>> the [min, max] for a is [0, 99], not [1, 99].
>>>
>>> For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
>>> const and is exactly the edge of the dst reg.
>>>
>>> Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
>>> ---
>>> kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
>>> 1 file changed, 44 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
>>> index 727a59e4a647..7b074ac93190 100644
>>> --- a/kernel/bpf/verifier.c
>>> +++ b/kernel/bpf/verifier.c
>>> @@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
>>> reg->type = SCALAR_VALUE;
>>> }
>>>
>>> +#define CHECK_REG_MIN(value) \
>>> +do { \
>>> + if ((value) == (typeof(value))imm) \
>>> + value++; \
>>> +} while (0)
>>> +
>>> +#define CHECK_REG_MAX(value) \
>>> +do { \
>>> + if ((value) == (typeof(value))imm) \
>>> + value--; \
>>> +} while (0)
>>> +
>>> +static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
>>> +{
>> What if reg->s32_min_value == imm and reg->s32_max_value == imm?
>> Has this been handled in previous verifier logic?
> Will such a case happen? In current code path, the src reg is a const,
> and the is_branch_taken() will return 0 or 1 if the
> dst_reg->s32_min_value == dst_reg->s32_max_value.
>
> Enn......maybe we can do more checking here in case that someone
> calls this function in another place.
I double checked the source code as well. Indeed, 'reg' should
not be a constant as it has been handled in is_branch_taken()
properly. Ignore my comments above then. Thanks!
>
> Thanks!
> Menglong Dong
>
>>> + CHECK_REG_MIN(reg->s32_min_value);
>>> + CHECK_REG_MAX(reg->s32_max_value);
>>> + CHECK_REG_MIN(reg->u32_min_value);
>>> + CHECK_REG_MAX(reg->u32_max_value);
>>> +}
>>> +
>>> +static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
>>> +{
>>> + CHECK_REG_MIN(reg->smin_value);
>>> + CHECK_REG_MAX(reg->smax_value);
>>> +
>>> + CHECK_REG_MIN(reg->umin_value);
>>> + CHECK_REG_MAX(reg->umax_value);
>>> +
>>> + CHECK_REG_MIN(reg->s32_min_value);
>>> + CHECK_REG_MAX(reg->s32_max_value);
>>> + CHECK_REG_MIN(reg->u32_min_value);
>>> + CHECK_REG_MAX(reg->u32_max_value);
>>> +}
>>> +
>>> static void mark_reg_known_zero(struct bpf_verifier_env *env,
>>> struct bpf_reg_state *regs, u32 regno)
>>> {
>>> @@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
>>> }
>>> break;
>>> case BPF_JNE:
>>> - /* we don't derive any new information for inequality yet */
>>> + /* try to recompute the bound of reg1 if reg2 is a const and
>>> + * is exactly the edge of reg1.
>>> + */
>>> + if (is_reg_const(reg2, is_jmp32)) {
>>> + val = reg_const_value(reg2, is_jmp32);
>>> + if (is_jmp32)
>>> + mark_reg32_not_equal(reg1, val);
>>> + else
>>> + mark_reg_not_equal(reg1, val);
>>> + }
>>> break;
>>> case BPF_JSET:
>>> if (!is_reg_const(reg2, is_jmp32))
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
2023-12-10 13:00 [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs Menglong Dong
2023-12-11 5:09 ` Yonghong Song
@ 2023-12-11 19:15 ` Andrii Nakryiko
2023-12-12 2:15 ` Menglong Dong
1 sibling, 1 reply; 7+ messages in thread
From: Andrii Nakryiko @ 2023-12-11 19:15 UTC (permalink / raw)
To: Menglong Dong
Cc: andrii, ast, daniel, john.fastabend, martin.lau, song,
yonghong.song, kpsingh, sdf, haoluo, jolsa, bpf, linux-kernel
On Sun, Dec 10, 2023 at 5:00 AM Menglong Dong <menglong8.dong@gmail.com> wrote:
>
> We can derive some new information for BPF_JNE in regs_refine_cond_op().
> Take following code for example:
>
> /* The type of "a" is u16 */
> if (a > 0 && a < 100) {
> /* the range of the register for a is [0, 99], not [1, 99],
> * and will cause the following error:
> *
> * invalid zero-sized read
> *
> * as a can be 0.
> */
> bpf_skb_store_bytes(skb, xx, xx, a, 0);
> }
>
> In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
> TRUE branch, the dst_reg will be marked as known to 0. However, in the
> fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
> the [min, max] for a is [0, 99], not [1, 99].
>
> For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
> const and is exactly the edge of the dst reg.
>
> Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
> ---
> kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
> 1 file changed, 44 insertions(+), 1 deletion(-)
>
> diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> index 727a59e4a647..7b074ac93190 100644
> --- a/kernel/bpf/verifier.c
> +++ b/kernel/bpf/verifier.c
> @@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
> reg->type = SCALAR_VALUE;
> }
>
> +#define CHECK_REG_MIN(value) \
> +do { \
> + if ((value) == (typeof(value))imm) \
> + value++; \
> +} while (0)
> +
> +#define CHECK_REG_MAX(value) \
> +do { \
> + if ((value) == (typeof(value))imm) \
> + value--; \
> +} while (0)
> +
> +static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
> +{
> + CHECK_REG_MIN(reg->s32_min_value);
> + CHECK_REG_MAX(reg->s32_max_value);
> + CHECK_REG_MIN(reg->u32_min_value);
> + CHECK_REG_MAX(reg->u32_max_value);
> +}
> +
> +static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
> +{
> + CHECK_REG_MIN(reg->smin_value);
> + CHECK_REG_MAX(reg->smax_value);
> +
> + CHECK_REG_MIN(reg->umin_value);
> + CHECK_REG_MAX(reg->umax_value);
> +
> + CHECK_REG_MIN(reg->s32_min_value);
> + CHECK_REG_MAX(reg->s32_max_value);
> + CHECK_REG_MIN(reg->u32_min_value);
> + CHECK_REG_MAX(reg->u32_max_value);
> +}
please don't use macros for this, this code is tricky enough without
having to jump around double-checking what exactly macros are doing.
Just code it explicitly.
Also I don't see the need for mark_reg32_not_equal() and
mark_reg_not_equal() helper functions, there is just one place where
this logic is going to be called from, so let's add code right there.
> +
> static void mark_reg_known_zero(struct bpf_verifier_env *env,
> struct bpf_reg_state *regs, u32 regno)
> {
> @@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
> }
> break;
> case BPF_JNE:
> - /* we don't derive any new information for inequality yet */
> + /* try to recompute the bound of reg1 if reg2 is a const and
> + * is exactly the edge of reg1.
> + */
> + if (is_reg_const(reg2, is_jmp32)) {
> + val = reg_const_value(reg2, is_jmp32);
> + if (is_jmp32)
> + mark_reg32_not_equal(reg1, val);
> + else
> + mark_reg_not_equal(reg1, val);
> + }
> break;
> case BPF_JSET:
> if (!is_reg_const(reg2, is_jmp32))
> --
> 2.39.2
>
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
2023-12-11 19:15 ` Andrii Nakryiko
@ 2023-12-12 2:15 ` Menglong Dong
2023-12-12 3:51 ` Andrii Nakryiko
0 siblings, 1 reply; 7+ messages in thread
From: Menglong Dong @ 2023-12-12 2:15 UTC (permalink / raw)
To: Andrii Nakryiko
Cc: andrii, ast, daniel, john.fastabend, martin.lau, song,
yonghong.song, kpsingh, sdf, haoluo, jolsa, bpf, linux-kernel
On Tue, Dec 12, 2023 at 3:16 AM Andrii Nakryiko
<andrii.nakryiko@gmail.com> wrote:
>
> On Sun, Dec 10, 2023 at 5:00 AM Menglong Dong <menglong8.dong@gmail.com> wrote:
> >
> > We can derive some new information for BPF_JNE in regs_refine_cond_op().
> > Take following code for example:
> >
> > /* The type of "a" is u16 */
> > if (a > 0 && a < 100) {
> > /* the range of the register for a is [0, 99], not [1, 99],
> > * and will cause the following error:
> > *
> > * invalid zero-sized read
> > *
> > * as a can be 0.
> > */
> > bpf_skb_store_bytes(skb, xx, xx, a, 0);
> > }
> >
> > In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
> > TRUE branch, the dst_reg will be marked as known to 0. However, in the
> > fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
> > the [min, max] for a is [0, 99], not [1, 99].
> >
> > For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
> > const and is exactly the edge of the dst reg.
> >
> > Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
> > ---
> > kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
> > 1 file changed, 44 insertions(+), 1 deletion(-)
> >
> > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > index 727a59e4a647..7b074ac93190 100644
> > --- a/kernel/bpf/verifier.c
> > +++ b/kernel/bpf/verifier.c
> > @@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
> > reg->type = SCALAR_VALUE;
> > }
> >
> > +#define CHECK_REG_MIN(value) \
> > +do { \
> > + if ((value) == (typeof(value))imm) \
> > + value++; \
> > +} while (0)
> > +
> > +#define CHECK_REG_MAX(value) \
> > +do { \
> > + if ((value) == (typeof(value))imm) \
> > + value--; \
> > +} while (0)
> > +
> > +static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
> > +{
> > + CHECK_REG_MIN(reg->s32_min_value);
> > + CHECK_REG_MAX(reg->s32_max_value);
> > + CHECK_REG_MIN(reg->u32_min_value);
> > + CHECK_REG_MAX(reg->u32_max_value);
> > +}
> > +
> > +static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
> > +{
> > + CHECK_REG_MIN(reg->smin_value);
> > + CHECK_REG_MAX(reg->smax_value);
> > +
> > + CHECK_REG_MIN(reg->umin_value);
> > + CHECK_REG_MAX(reg->umax_value);
> > +
> > + CHECK_REG_MIN(reg->s32_min_value);
> > + CHECK_REG_MAX(reg->s32_max_value);
> > + CHECK_REG_MIN(reg->u32_min_value);
> > + CHECK_REG_MAX(reg->u32_max_value);
> > +}
>
> please don't use macros for this, this code is tricky enough without
> having to jump around double-checking what exactly macros are doing.
> Just code it explicitly.
>
Okay!
> Also I don't see the need for mark_reg32_not_equal() and
> mark_reg_not_equal() helper functions, there is just one place where
> this logic is going to be called from, so let's add code right there.
>
Yeah, you are right. And I just found that you have already
implemented the test case for this logic in reg_bounds.c/range_cond().
I wonder why this logic is not implemented in the verifier yet?
Am I missing something?
Thanks!
Menglong Dong
> > +
> > static void mark_reg_known_zero(struct bpf_verifier_env *env,
> > struct bpf_reg_state *regs, u32 regno)
> > {
> > @@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
> > }
> > break;
> > case BPF_JNE:
> > - /* we don't derive any new information for inequality yet */
> > + /* try to recompute the bound of reg1 if reg2 is a const and
> > + * is exactly the edge of reg1.
> > + */
> > + if (is_reg_const(reg2, is_jmp32)) {
> > + val = reg_const_value(reg2, is_jmp32);
> > + if (is_jmp32)
> > + mark_reg32_not_equal(reg1, val);
> > + else
> > + mark_reg_not_equal(reg1, val);
> > + }
> > break;
> > case BPF_JSET:
> > if (!is_reg_const(reg2, is_jmp32))
> > --
> > 2.39.2
> >
^ permalink raw reply [flat|nested] 7+ messages in thread
* Re: [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs
2023-12-12 2:15 ` Menglong Dong
@ 2023-12-12 3:51 ` Andrii Nakryiko
0 siblings, 0 replies; 7+ messages in thread
From: Andrii Nakryiko @ 2023-12-12 3:51 UTC (permalink / raw)
To: Menglong Dong
Cc: andrii, ast, daniel, john.fastabend, martin.lau, song,
yonghong.song, kpsingh, sdf, haoluo, jolsa, bpf, linux-kernel
On Mon, Dec 11, 2023 at 6:16 PM Menglong Dong <menglong8.dong@gmail.com> wrote:
>
> On Tue, Dec 12, 2023 at 3:16 AM Andrii Nakryiko
> <andrii.nakryiko@gmail.com> wrote:
> >
> > On Sun, Dec 10, 2023 at 5:00 AM Menglong Dong <menglong8.dong@gmail.com> wrote:
> > >
> > > We can derive some new information for BPF_JNE in regs_refine_cond_op().
> > > Take following code for example:
> > >
> > > /* The type of "a" is u16 */
> > > if (a > 0 && a < 100) {
> > > /* the range of the register for a is [0, 99], not [1, 99],
> > > * and will cause the following error:
> > > *
> > > * invalid zero-sized read
> > > *
> > > * as a can be 0.
> > > */
> > > bpf_skb_store_bytes(skb, xx, xx, a, 0);
> > > }
> > >
> > > In the code above, "a > 0" will be compiled to "jmp xxx if a == 0". In the
> > > TRUE branch, the dst_reg will be marked as known to 0. However, in the
> > > fallthrough(FALSE) branch, the dst_reg will not be handled, which makes
> > > the [min, max] for a is [0, 99], not [1, 99].
> > >
> > > For BPF_JNE, we can reduce the range of the dst reg if the src reg is a
> > > const and is exactly the edge of the dst reg.
> > >
> > > Signed-off-by: Menglong Dong <menglong8.dong@gmail.com>
> > > ---
> > > kernel/bpf/verifier.c | 45 ++++++++++++++++++++++++++++++++++++++++++-
> > > 1 file changed, 44 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/kernel/bpf/verifier.c b/kernel/bpf/verifier.c
> > > index 727a59e4a647..7b074ac93190 100644
> > > --- a/kernel/bpf/verifier.c
> > > +++ b/kernel/bpf/verifier.c
> > > @@ -1764,6 +1764,40 @@ static void __mark_reg_const_zero(struct bpf_reg_state *reg)
> > > reg->type = SCALAR_VALUE;
> > > }
> > >
> > > +#define CHECK_REG_MIN(value) \
> > > +do { \
> > > + if ((value) == (typeof(value))imm) \
> > > + value++; \
> > > +} while (0)
> > > +
> > > +#define CHECK_REG_MAX(value) \
> > > +do { \
> > > + if ((value) == (typeof(value))imm) \
> > > + value--; \
> > > +} while (0)
> > > +
> > > +static void mark_reg32_not_equal(struct bpf_reg_state *reg, u64 imm)
> > > +{
> > > + CHECK_REG_MIN(reg->s32_min_value);
> > > + CHECK_REG_MAX(reg->s32_max_value);
> > > + CHECK_REG_MIN(reg->u32_min_value);
> > > + CHECK_REG_MAX(reg->u32_max_value);
> > > +}
> > > +
> > > +static void mark_reg_not_equal(struct bpf_reg_state *reg, u64 imm)
> > > +{
> > > + CHECK_REG_MIN(reg->smin_value);
> > > + CHECK_REG_MAX(reg->smax_value);
> > > +
> > > + CHECK_REG_MIN(reg->umin_value);
> > > + CHECK_REG_MAX(reg->umax_value);
> > > +
> > > + CHECK_REG_MIN(reg->s32_min_value);
> > > + CHECK_REG_MAX(reg->s32_max_value);
> > > + CHECK_REG_MIN(reg->u32_min_value);
> > > + CHECK_REG_MAX(reg->u32_max_value);
> > > +}
> >
> > please don't use macros for this, this code is tricky enough without
> > having to jump around double-checking what exactly macros are doing.
> > Just code it explicitly.
> >
>
> Okay!
>
> > Also I don't see the need for mark_reg32_not_equal() and
> > mark_reg_not_equal() helper functions, there is just one place where
> > this logic is going to be called from, so let's add code right there.
> >
>
> Yeah, you are right. And I just found that you have already
> implemented the test case for this logic in reg_bounds.c/range_cond().
> I wonder why this logic is not implemented in the verifier yet?
> Am I missing something?
No, I just didn't want to add yet more verifier changes in my original
patch set on extending reg bounds logic.
>
> Thanks!
> Menglong Dong
>
> > > +
> > > static void mark_reg_known_zero(struct bpf_verifier_env *env,
> > > struct bpf_reg_state *regs, u32 regno)
> > > {
> > > @@ -14332,7 +14366,16 @@ static void regs_refine_cond_op(struct bpf_reg_state *reg1, struct bpf_reg_state
> > > }
> > > break;
> > > case BPF_JNE:
> > > - /* we don't derive any new information for inequality yet */
> > > + /* try to recompute the bound of reg1 if reg2 is a const and
> > > + * is exactly the edge of reg1.
> > > + */
> > > + if (is_reg_const(reg2, is_jmp32)) {
> > > + val = reg_const_value(reg2, is_jmp32);
> > > + if (is_jmp32)
> > > + mark_reg32_not_equal(reg1, val);
> > > + else
> > > + mark_reg_not_equal(reg1, val);
> > > + }
> > > break;
> > > case BPF_JSET:
> > > if (!is_reg_const(reg2, is_jmp32))
> > > --
> > > 2.39.2
> > >
^ permalink raw reply [flat|nested] 7+ messages in thread
end of thread, other threads:[~2023-12-12 3:51 UTC | newest]
Thread overview: 7+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-12-10 13:00 [PATCH bpf-next] bpf: make the verifier trace the "not qeual" for regs Menglong Dong
2023-12-11 5:09 ` Yonghong Song
2023-12-11 9:39 ` Menglong Dong
2023-12-11 15:03 ` Yonghong Song
2023-12-11 19:15 ` Andrii Nakryiko
2023-12-12 2:15 ` Menglong Dong
2023-12-12 3:51 ` Andrii Nakryiko
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).