On 2019/12/6 18:02, Guoqing Jiang wrote:
>
>
> On 12/6/19 10:40 AM, liuzhiqiang (I) wrote:
>> In raid1_sync_request func, rdev should be checked whether it is null
>> before reference.
>
> Do you have real calltrace about it?
Thanks for your comment.
Actually, it is reported by one static scanning tool.
>
>> Fixes: 06f603851f("md/raid1: avoid reading known bad blocks during resync")
>> Signed-off-by: Zhiqiang Liu <liuzhiqiang26@xxxxxxxxxx>
>> ---
>> drivers/md/raid1.c | 2 +-
>> 1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/md/raid1.c b/drivers/md/raid1.c
>> index a409ab6f30bc..0dea35052efe 100644
>> --- a/drivers/md/raid1.c
>> +++ b/drivers/md/raid1.c
>> @@ -2782,7 +2782,7 @@ static sector_t raid1_sync_request(struct mddev *mddev, sector_t sector_nr,
>> write_targets++;
>> }
>> }
>> - if (bio->bi_end_io) {
>> + if (rdev != NULL && bio->bi_end_io) {
>> atomic_inc(&rdev->nr_pending);
>> bio->bi_iter.bi_sector = sector_nr + rdev->data_offset;
>> bio_set_dev(bio, rdev->bdev);
>
> If "bio->bi_end_io" is true, I think it implys rdev exists because bio->bi_end_io is set
> when rdev != NUL. I don't object to add it to make it explicitly, but it is not a fix.
>
ok, i will remove the fixes tag in v2 patch.
> Thanks,
> Guoqing
>
> .
