Re: [PATCH v3 1/2] md: prevent adding disks with larger logical_block_size to active arrays

[Li Nan <linan666@huaweicloud.com>]

在 2025/8/25 16:10, Paul Menzel 写道:
> Dear Li,
> 
> 
> Thank you for your patch.
> 
> Am 25.08.25 um 09:59 schrieb linan666@huaweicloud.com:
>> From: Li Nan <linan122@huawei.com>
>>
>> When adding a disk to a md array, avoid updating the array's
>> logical_block_size to match the new disk. This prevents accidental
>> partition table loss that renders the array unusable.
> 
> Do you have a reproducer to test this?

Please try the following script:

```
#sd[de] lbs is 512, sdf is 4096
mdadm -CR /dev/md0 -l1 -n2 /dev/sd[de] --assume-clean

#512
cat /sys/block/md0/queue/logical_block_size

#create md0p1
printf "g\nn\n\n\n\nw\n" | fdisk /dev/md0
lsblk | grep md0

mdadm --fail /dev/md0 /dev/sdd
mdadm --add /dev/md0 /dev/sdf

#4096
cat /sys/block/md0/queue/logical_block_size

#partition loss
partprobe /dev/md0
lsblk | grep md0
```

> 
>> The later patch will introduce a way to configure the array's
>> logical_block_size.
>>
>> Signed-off-by: Li Nan <linan122@huawei.com>
>> Reviewed-by: Martin K. Petersen <martin.petersen@oracle.com>
>> ---
>>   drivers/md/md.c | 7 +++++++
>>   1 file changed, 7 insertions(+)
>>
>> diff --git a/drivers/md/md.c b/drivers/md/md.c
>> index cea8fc96abd3..206434591b97 100644
>> --- a/drivers/md/md.c
>> +++ b/drivers/md/md.c
>> @@ -6064,6 +6064,13 @@ int mddev_stack_new_rdev(struct mddev *mddev, 
>> struct md_rdev *rdev)
>>       if (mddev_is_dm(mddev))
>>           return 0;
>> +    if (queue_logical_block_size(rdev->bdev->bd_disk->queue) >
>> +        queue_logical_block_size(mddev->gendisk->queue)) {
>> +        pr_err("%s: incompatible logical_block_size, can not add\n",
>> +               mdname(mddev));
>> +        return -EINVAL;
>> +    }
>> +
>>       lim = queue_limits_start_update(mddev->gendisk->queue);
>>       queue_limits_stack_bdev(&lim, rdev->bdev, rdev->data_offset,
>>                   mddev->gendisk->disk_name);
> 
> 
> Kind regards,
> 
> Paul
> 
> 
> .

-- 
Thanks,
Nan