Re: [RFC] BTRFS_DEV_REPLACE_ITEM_STATE_* doesn't match with on disk

[Anand Jain <anand.jain@oracle.com>]

David, any comments on this please.

Thanks, Anand


On 11/13/2018 06:32 PM, Anand Jain wrote:
> 
> David, Gentle ping.
> 
> Thanks, Anand
> 
> On 11/12/2018 03:50 PM, Nikolay Borisov wrote:
>>
>>
>> On 12.11.18 г. 6:58 ч., Anand Jain wrote:
>>>
>>> The dev_replace_state defines are miss matched between the
>>> BTRFS_IOCTL_DEV_REPLACE_STATE_* and BTRFS_DEV_REPLACE_ITEM_STATE_* [1].
>>>
>>> [1]
>>> -----------------------------
>>> btrfs.h:#define BTRFS_IOCTL_DEV_REPLACE_STATE_FINISHED        2
>>> btrfs.h:#define BTRFS_IOCTL_DEV_REPLACE_STATE_CANCELED        3
>>> btrfs.h:#define BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED        4
>>>
>>> btrfs_tree.h:#define BTRFS_DEV_REPLACE_ITEM_STATE_SUSPENDED    2
>>> btrfs_tree.h:#define BTRFS_DEV_REPLACE_ITEM_STATE_FINISHED    3
>>> btrfs_tree.h:#define BTRFS_DEV_REPLACE_ITEM_STATE_CANCELED    4
>>> -----------------------------
>>>
>>> The BTRFS_DEV_REPLACE_ITEM_STATE_* series is unused in both btrfs.ko and
>>> btrfs-progs, the on-disk also follows BTRFS_IOCTL_DEV_REPLACE_STATE_*
>>> (we set dev_replace->replace_state using the
>>> BTRFS_IOCTL_DEV_REPLACE_STATE_* defines and write to the on-disk).
>>>
>>>   359         btrfs_set_dev_replace_replace_state(eb, ptr,
>>>   360                 dev_replace->replace_state);
>>>
>>> IMO it should be ok to delete the BTRFS_DEV_REPLACE_ITEM_STATE_*
>>> altogether? But how about the userland progs other than btrfs-progs?
>>> If not at least fix the miss match as in [2], any comments?
>>
>> Unfortunately you are right. This seems to stem from sloppy job back in
>> the days of initial dev-replace support. BTRFS_DEV_REPLACE_ITEM_STATE_*
>> were added in e922e087a35c ("Btrfs: enhance btrfs structures for device
>> replace support"), yet they were never used. And the
>> IOCTL_DEV_REPLACE_STATE* were added in e93c89c1aaaa ("Btrfs: add new
>> sources for device replace code").
>>
>> It looks like the ITEM_STATE* definitions were stillborn so to speak and
>> personally I'm in favor of removing them. They shouldn't have been
>> merged in the first place and indeed the patch doesn't even have a
>> Reviewed-by tag. So it originated from the, I'd say, spartan days of
>> btrfs development...
>>
>> David,  any code which is using BTRFS_DEV_REPLACE_ITEM_STATE_SUSPENDED
>> is inherently broken, so how about we remove those definitions, then
>> when it's compilation is broken in the future the author will actually
>> have a chance to fix it, though it's highly unlikely anyone is relying
>> on those definitions.
>>
>>
>>>
>>> [2]
>>> --------------------------------------
>>> diff --git a/include/uapi/linux/btrfs_tree.h
>>> b/include/uapi/linux/btrfs_tree.h
>>> index aff1356c2bb8..9ffa7534cadf 100644
>>> --- a/include/uapi/linux/btrfs_tree.h
>>> +++ b/include/uapi/linux/btrfs_tree.h
>>> @@ -805,9 +805,9 @@ struct btrfs_dev_stats_item {
>>>   #define 
>>> BTRFS_DEV_REPLACE_ITEM_CONT_READING_FROM_SRCDEV_MODE_AVOID     1
>>>   #define BTRFS_DEV_REPLACE_ITEM_STATE_NEVER_STARTED     0
>>>   #define BTRFS_DEV_REPLACE_ITEM_STATE_STARTED           1
>>> -#define BTRFS_DEV_REPLACE_ITEM_STATE_SUSPENDED         2
>>> -#define BTRFS_DEV_REPLACE_ITEM_STATE_FINISHED          3
>>> -#define BTRFS_DEV_REPLACE_ITEM_STATE_CANCELED          4
>>> +#define BTRFS_DEV_REPLACE_ITEM_STATE_FINISHED          2
>>> +#define BTRFS_DEV_REPLACE_ITEM_STATE_CANCELED          3
>>> +#define BTRFS_DEV_REPLACE_ITEM_STATE_SUSPENDED         4
>>>
>>>   struct btrfs_dev_replace_item {
>>>          /*
>>> --------------------------------------
>>>
>>>
>>> Thanks, Anand
>>>