2018-08-13 03:12:10

by Jia-Ju Bai

Subject: [BUG] mm: truncate: a possible sleep-in-atomic-context bug in truncate_exceptional_pvec_entries()

The kernel may sleep with holding a spinlock.

The function call paths (from bottom to top) in Linux-4.16 are:

[FUNC] schedule
fs/dax.c, 259: schedule in get_unlocked_mapping_entry
fs/dax.c, 450: get_unlocked_mapping_entry in __dax_invalidate_mapping_entry
fs/dax.c, 471: __dax_invalidate_mapping_entry in dax_delete_mapping_entry
mm/truncate.c, 97: dax_delete_mapping_entry in
truncate_exceptional_pvec_entries
mm/truncate.c, 82: spin_lock_irq in truncate_exceptional_pvec_entries

I do not find a good way to fix, so I only report.
This is found by my static analysis tool (DSAC).


Thanks,
Jia-Ju Bai

2018-08-13 08:58:32

by Jan Kara

Subject: Re: [BUG] mm: truncate: a possible sleep-in-atomic-context bug in truncate_exceptional_pvec_entries()

Hi,

On Mon 13-08-18 11:10:23, Jia-Ju Bai wrote:
> The kernel may sleep with holding a spinlock.
>
> The function call paths (from bottom to top) in Linux-4.16 are:
>
> [FUNC] schedule
> fs/dax.c, 259: schedule in get_unlocked_mapping_entry
> fs/dax.c, 450: get_unlocked_mapping_entry in __dax_invalidate_mapping_entry
> fs/dax.c, 471: __dax_invalidate_mapping_entry in dax_delete_mapping_entry
> mm/truncate.c, 97: dax_delete_mapping_entry in
> truncate_exceptional_pvec_entries
> mm/truncate.c, 82: spin_lock_irq in truncate_exceptional_pvec_entries
>
> I do not find a good way to fix, so I only report.
> This is found by my static analysis tool (DSAC).

Thanks for report but this is a false positive. Note that the lock is
acquired only if we are not operating on DAX mapping but we can get to
dax_delete_mapping_entry() only if we are operating on DAX mapping.

Honza

--
Jan Kara <[email protected]>
SUSE Labs, CR

2018-08-13 09:32:07

by Jia-Ju Bai

Subject: Re: [BUG] mm: truncate: a possible sleep-in-atomic-context bug in truncate_exceptional_pvec_entries()

On 2018/8/13 16:56, Jan Kara wrote:
> Hi,
>
> On Mon 13-08-18 11:10:23, Jia-Ju Bai wrote:
>> The kernel may sleep with holding a spinlock.
>>
>> The function call paths (from bottom to top) in Linux-4.16 are:
>>
>> [FUNC] schedule
>> fs/dax.c, 259: schedule in get_unlocked_mapping_entry
>> fs/dax.c, 450: get_unlocked_mapping_entry in __dax_invalidate_mapping_entry
>> fs/dax.c, 471: __dax_invalidate_mapping_entry in dax_delete_mapping_entry
>> mm/truncate.c, 97: dax_delete_mapping_entry in
>> truncate_exceptional_pvec_entries
>> mm/truncate.c, 82: spin_lock_irq in truncate_exceptional_pvec_entries
>>
>> I do not find a good way to fix, so I only report.
>> This is found by my static analysis tool (DSAC).
> Thanks for report but this is a false positive. Note that the lock is
> acquired only if we are not operating on DAX mapping but we can get to
> dax_delete_mapping_entry() only if we are operating on DAX mapping.

Thanks for your reply :)
My tool does not well check the path condition here...
Sorry for this false report.


Best wishes,
Jia-Ju Bai