Re: [PATCH 3/3] ntfs: mark quotas out of date on initial rw mount

[CharSyam]

Hi. Hyunchul.

  I agree that the out-of-date marking path looks questionable.

 After re-checking the code, the current ntfs driver does not appear to
 implement quota accounting: new inodes get v1.2 $STANDARD_INFORMATION, and
 the driver does not maintain the v3 owner_id/quota_charged fields or $Quota
 usage records. So fixing ntfs_mark_quotas_out_of_date() only makes an
 existing path set QUOTA_FLAG_OUT_OF_DATE, but that does not seem meaningful
 without the metadata Windows would need to rescan.

 Given that, I am wondering what removal scope you would prefer:

 1. Drop only this patch and leave the existing broken remount-rw quota
     marking path as-is for now.
 2. Remove ntfs_mark_quotas_out_of_date() and its remount-rw call site.
 3. Remove the broader unused quota-loading state as well ($Quota/$Q inode
     loading, quota_ino/quota_q_ino, NV_QuotaOutOfDate, quota.o), while keeping
     the on-disk quota layout definitions.

My current preference is option 3, because once the marking helper is gone,
the loaded $Quota inodes do not seem to have another user. But since this
changes the series from a bug fix into removing unsupported quota handling,
I would like to confirm the intended direction first.

Thanks.
DaeMyung.

2026년 5월 15일 (금) 오전 8:50, Hyunchul Lee <hyc.lee@xxxxxxxxx>님이 작성:
>
> 2026년 5월 14일 (목) 오후 10:25, CharSyam <charsyam@xxxxxxxxx>님이 작성:
> >
> >     Thanks for the review.
> >
> >     You're right that the driver does not maintain quota accounting: it
> >     creates v1.2 $STANDARD_INFORMATION for new inodes, and it does not update
> >     the v3 quota fields or $Quota usage accounting.
> >
> >     Given that I do not have modern Windows validation for the exact quota
> >     rescan behavior, I'll drop 3/3 from v3 and resend just 1/3 and 2/3.
>
> I think setting the out-of-date flag in patch 2/3 is also not meaningful.
> How about removing ntfs_mark_quotas_out_of_date() related codes for now?
>
> >
> >     Thanks,
> >     DaeMyung
>
>
>
>
> --
> Thanks,
> Hyunchul