riscv: stacktrace: Fix missing the first frame

Message ID 20221205132936.493245-1-liushixin2@huawei.com
State New
Headers
Series riscv: stacktrace: Fix missing the first frame |

Commit Message

Liu Shixin Dec. 5, 2022, 1:29 p.m. UTC
  When running kfence_test, I found some testcases failed like this:

 # test_out_of_bounds_read: EXPECTATION FAILED at mm/kfence/kfence_test.c:346
 Expected report_matches(&expect) to be true, but is false
 not ok 1 - test_out_of_bounds_read

The corresponding call-trace is:

 BUG: KFENCE: out-of-bounds read in kunit_try_run_case+0x38/0x84

 Out-of-bounds read at 0x(____ptrval____) (32B right of kfence-#10):
  kunit_try_run_case+0x38/0x84
  kunit_generic_run_threadfn_adapter+0x12/0x1e
  kthread+0xc8/0xde
  ret_from_exception+0x0/0xc

The kfence_test using the first frame of call trace to check whether the
testcase is succeed or not. Patch a7c5c7e8ff78 skip first frame for all
case, which results the kfence_test failed. Indeed, we only need to skip
the first frame for case (task==NULL || task==current).

With this patch, the call-trace will be:

 BUG: KFENCE: out-of-bounds read in test_out_of_bounds_read+0x88/0x19e

 Out-of-bounds read at 0x(____ptrval____) (1B left of kfence-#7):
  test_out_of_bounds_read+0x88/0x19e
  kunit_try_run_case+0x38/0x84
  kunit_generic_run_threadfn_adapter+0x12/0x1e
  kthread+0xc8/0xde
  ret_from_exception+0x0/0xc

Fixes: a7c5c7e8ff78 ("riscv: eliminate unreliable __builtin_frame_address(1)")
Signed-off-by: Liu Shixin <liushixin2@huawei.com>
---
 arch/riscv/kernel/stacktrace.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
  

Comments

Conor Dooley Dec. 6, 2022, 4:15 p.m. UTC | #1
On Mon, Dec 05, 2022 at 09:29:36PM +0800, Liu Shixin wrote:
> When running kfence_test, I found some testcases failed like this:
> 
>  # test_out_of_bounds_read: EXPECTATION FAILED at mm/kfence/kfence_test.c:346
>  Expected report_matches(&expect) to be true, but is false
>  not ok 1 - test_out_of_bounds_read
> 
> The corresponding call-trace is:
> 
>  BUG: KFENCE: out-of-bounds read in kunit_try_run_case+0x38/0x84
> 
>  Out-of-bounds read at 0x(____ptrval____) (32B right of kfence-#10):
>   kunit_try_run_case+0x38/0x84
>   kunit_generic_run_threadfn_adapter+0x12/0x1e
>   kthread+0xc8/0xde
>   ret_from_exception+0x0/0xc
> 
> The kfence_test using the first frame of call trace to check whether the
> testcase is succeed or not. Patch a7c5c7e8ff78 skip first frame for all
> case, which results the kfence_test failed. Indeed, we only need to skip
> the first frame for case (task==NULL || task==current).
> 
> With this patch, the call-trace will be:
> 
>  BUG: KFENCE: out-of-bounds read in test_out_of_bounds_read+0x88/0x19e
> 
>  Out-of-bounds read at 0x(____ptrval____) (1B left of kfence-#7):
>   test_out_of_bounds_read+0x88/0x19e
>   kunit_try_run_case+0x38/0x84
>   kunit_generic_run_threadfn_adapter+0x12/0x1e
>   kthread+0xc8/0xde
>   ret_from_exception+0x0/0xc
> 
> Fixes: a7c5c7e8ff78 ("riscv: eliminate unreliable __builtin_frame_address(1)")

This fixes tag is not right, did checkpatch not warn about it?

The correct fixes tag would be:
Fixes: 6a00ef449370 ("riscv: eliminate unreliable __builtin_frame_address(1)")

Maybe consider automating the creation of fixes tags, like so:
git log -1 --format='Fixes: %h (\"%s\")'

Thanks,
Conor.

> Signed-off-by: Liu Shixin <liushixin2@huawei.com>
> ---
>  arch/riscv/kernel/stacktrace.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c
> index 08d11a53f39e..5fe2ae4cf135 100644
> --- a/arch/riscv/kernel/stacktrace.c
> +++ b/arch/riscv/kernel/stacktrace.c
> @@ -30,6 +30,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
>  		fp = (unsigned long)__builtin_frame_address(0);
>  		sp = current_stack_pointer;
>  		pc = (unsigned long)walk_stackframe;
> +		level = -1;
>  	} else {
>  		/* task blocked in __switch_to */
>  		fp = task->thread.s[0];
> @@ -41,7 +42,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
>  		unsigned long low, high;
>  		struct stackframe *frame;
>  
> -		if (unlikely(!__kernel_text_address(pc) || (level++ >= 1 && !fn(arg, pc))))
> +		if (unlikely(!__kernel_text_address(pc) || (level++ >= 0 && !fn(arg, pc))))
>  			break;
>  
>  		/* Validate frame pointer */
> -- 
> 2.25.1
>
  
Liu Shixin Dec. 7, 2022, 2:02 a.m. UTC | #2
On 2022/12/7 0:15, Conor Dooley wrote:
> On Mon, Dec 05, 2022 at 09:29:36PM +0800, Liu Shixin wrote:
>> When running kfence_test, I found some testcases failed like this:
>>
>>  # test_out_of_bounds_read: EXPECTATION FAILED at mm/kfence/kfence_test.c:346
>>  Expected report_matches(&expect) to be true, but is false
>>  not ok 1 - test_out_of_bounds_read
>>
>> The corresponding call-trace is:
>>
>>  BUG: KFENCE: out-of-bounds read in kunit_try_run_case+0x38/0x84
>>
>>  Out-of-bounds read at 0x(____ptrval____) (32B right of kfence-#10):
>>   kunit_try_run_case+0x38/0x84
>>   kunit_generic_run_threadfn_adapter+0x12/0x1e
>>   kthread+0xc8/0xde
>>   ret_from_exception+0x0/0xc
>>
>> The kfence_test using the first frame of call trace to check whether the
>> testcase is succeed or not. Patch a7c5c7e8ff78 skip first frame for all
>> case, which results the kfence_test failed. Indeed, we only need to skip
>> the first frame for case (task==NULL || task==current).
>>
>> With this patch, the call-trace will be:
>>
>>  BUG: KFENCE: out-of-bounds read in test_out_of_bounds_read+0x88/0x19e
>>
>>  Out-of-bounds read at 0x(____ptrval____) (1B left of kfence-#7):
>>   test_out_of_bounds_read+0x88/0x19e
>>   kunit_try_run_case+0x38/0x84
>>   kunit_generic_run_threadfn_adapter+0x12/0x1e
>>   kthread+0xc8/0xde
>>   ret_from_exception+0x0/0xc
>>
>> Fixes: a7c5c7e8ff78 ("riscv: eliminate unreliable __builtin_frame_address(1)")
> This fixes tag is not right, did checkpatch not warn about it?
Yes, there are no warn. Maybe it's because I do have this commit in my local repository.
I confused the original commit with my local commit.

Thanks for your reminder. I'll fix it as soon as possible.
> The correct fixes tag would be:
> Fixes: 6a00ef449370 ("riscv: eliminate unreliable __builtin_frame_address(1)")
>
> Maybe consider automating the creation of fixes tags, like so:
> git log -1 --format='Fixes: %h (\"%s\")'
Thanks, it seems to be convenient.

>
> Thanks,
> Conor.
>
>> Signed-off-by: Liu Shixin <liushixin2@huawei.com>
>> ---
>>  arch/riscv/kernel/stacktrace.c | 3 ++-
>>  1 file changed, 2 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c
>> index 08d11a53f39e..5fe2ae4cf135 100644
>> --- a/arch/riscv/kernel/stacktrace.c
>> +++ b/arch/riscv/kernel/stacktrace.c
>> @@ -30,6 +30,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
>>  		fp = (unsigned long)__builtin_frame_address(0);
>>  		sp = current_stack_pointer;
>>  		pc = (unsigned long)walk_stackframe;
>> +		level = -1;
>>  	} else {
>>  		/* task blocked in __switch_to */
>>  		fp = task->thread.s[0];
>> @@ -41,7 +42,7 @@ void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
>>  		unsigned long low, high;
>>  		struct stackframe *frame;
>>  
>> -		if (unlikely(!__kernel_text_address(pc) || (level++ >= 1 && !fn(arg, pc))))
>> +		if (unlikely(!__kernel_text_address(pc) || (level++ >= 0 && !fn(arg, pc))))
>>  			break;
>>  
>>  		/* Validate frame pointer */
>> -- 
>> 2.25.1
>>
  

Patch

diff --git a/arch/riscv/kernel/stacktrace.c b/arch/riscv/kernel/stacktrace.c
index 08d11a53f39e..5fe2ae4cf135 100644
--- a/arch/riscv/kernel/stacktrace.c
+++ b/arch/riscv/kernel/stacktrace.c
@@ -30,6 +30,7 @@  void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
 		fp = (unsigned long)__builtin_frame_address(0);
 		sp = current_stack_pointer;
 		pc = (unsigned long)walk_stackframe;
+		level = -1;
 	} else {
 		/* task blocked in __switch_to */
 		fp = task->thread.s[0];
@@ -41,7 +42,7 @@  void notrace walk_stackframe(struct task_struct *task, struct pt_regs *regs,
 		unsigned long low, high;
 		struct stackframe *frame;
 
-		if (unlikely(!__kernel_text_address(pc) || (level++ >= 1 && !fn(arg, pc))))
+		if (unlikely(!__kernel_text_address(pc) || (level++ >= 0 && !fn(arg, pc))))
 			break;
 
 		/* Validate frame pointer */