[linux-lvm] [PATCH] lvdisplay: dispaly correct status on linear type

[heming.zhao at suse.com]

thank you for your kindly review.

On 8/27/20 1:03 AM, David Teigland wrote:
> On Wed, Aug 26, 2020 at 11:50:44PM +0800, Zhao Heming wrote:
>> It commit is enhancement for 1d0dc74f9147e3c1f3681efa4166cbe2edcb6571
>> 1d0dc74f9147 only supports all raid type, this commit adds linear type.
>>
>> With this patch, for linear type LV, one of two disks missing,
>> lvdisplay will show
>> from:
>>    LV Status              available (partial)
>> to:
>>    LV Status              NOT available (partial)
> 
> It would be nice if you could come up with a more formal definition for
> what you're calling "lv_is_available", and also in the same terms define
> the existing "lv_is_partial", explaining how they differ.  Both properties
> may be useful to show in lvs reporting fields (and possibly lv_attr).
> Reporting this state only via the lvdisplay line above seems strange.
I will send the v2 patch very soon.
the v2 patch only for lvdisplay cmd. if my code acceptable, I plan update lvs
cmd in v3, add a new letter (N)ot_available in bit 9 (Volume Health).
> 
>> +			} else if (seg_type(seg, s) == AREA_PV) {
>> +				if (seg_pv(seg, s)->status & MISSING_PV)
> 
> Using only MISSING_PV is probably insufficient, because that flag is used
> only when the metadata is written while the device is missing.  i.e.
> pv->dev can be NULL when the flag is not set.
> 
I have a little confused with your comments.
These 2 lines are same as _lv_mark_if_partial_single() code:
```
            if (seg_type(lvseg, s) == AREA_PV) {
                if (is_missing_pv(seg_pv(lvseg, s))) {
```
why _lv_mark_if_partial_single can use these 2 lines?
please wait for the v2 patch, I changed some code logic, set the NOT_AVAIL_LV in
_lv_mark_if_partial_single().

> Dave
>