On Mon, Oct 09 2017, Zhilong Liu wrote:
> On 10/05/2017 07:44 PM, Tomasz Majchrzak wrote:
>> On Tue, Sep 05, 2017 at 05:41:36PM +0800, Zhilong Liu wrote:
>>> Grow: Changing component size must be larger than current
>>> chunk size against stripe raids, otherwise Grow_reshape()
>>> would set s->size to '0'.
>>>
>>> Signed-off-by: Zhilong Liu <zlliu@xxxxxxxx>
>>> ---
>>> Grow.c | 6 ++++++
>>> 1 file changed, 6 insertions(+)
>>>
>>> diff --git a/Grow.c b/Grow.c
>>> index 534ba80..f0a21ff 100644
>>> --- a/Grow.c
>>> +++ b/Grow.c
>>> @@ -1816,6 +1816,12 @@ int Grow_reshape(char *devname, int fd,
>>> return 1;
>>> }
>>>
>>> + if (array.level > 1 &&
>>> + (array.chunk_size / 1024) > (int)s->size) {
>>> + pr_err("component size must be larger than chunk size.\n");
>>> + return 1;
>>> + }
>>> +
>>> st = super_by_fd(fd, &subarray);
>>> if (!st) {
>>> pr_err("Unable to determine metadata format for %s\n", devname);
>>> --
>>> 2.6.6
>>
>> Hi Zhilong Liu,
>>
>> Above patch has broken RAID level conversion:
>>
>> Before the patch:
>>
>> mdadm --create /dev/md/raid --level=10 --raid-devices=4 /dev/nvme0n1
>> /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1 --run --assume-clean
>> MDADM_EXPERIMENTAL=1 mdadm --grow /dev/md/raid --level=0
>> mdadm: level of /dev/md/raid changed to raid0
>>
>> After the patch:
>>
>> mdadm --create /dev/md/raid --level=10 --raid-devices=4 /dev/nvme0n1
>> /dev/nvme1n1 /dev/nvme2n1 /dev/nvme3n1 --run --assume-clean
>> MDADM_EXPERIMENTAL=1 mdadm --grow /dev/md/raid --level=0
>> mdadm: component size must be larger than chunk size.
>>
>> Could you take a look at it, please?
>
> Hi, Tomek;
> I just back from vacation, and this is very important info for me,
> thanks very much.
> Against this issue, the codes should be proper like:
>
> diff --git a/Grow.c b/Grow.c
> index 1149753..180fd78 100644
> --- a/Grow.c
> +++ b/Grow.c
> @@ -1814,7 +1814,8 @@ int Grow_reshape(char *devname, int fd,
> }
>
> if (array.level > 1 &&
> - (array.chunk_size / 1024) > (int)s->size) {
> + (array.chunk_size / 1024) > (int)s->size &&
> + s->level == UnSet) {
> pr_err("component size must be larger than chunk size.\n");
> return 1;
> }
I think the correct fix for this bug is to add a test for "s->size > 0".
if "s->size == 0", no size change has been requested.
NeilBrown
>
> Is this changing good for you? Any ideas is very welcome.
>
> Thanks,
> -Zhilong
>
>> Thanks,
>>
>> Tomek
>>
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
Attachment:
signature.asc
Description: PGP signature
