On Wed, Feb 12, 2020 at 07:52:30PM -0600, Eric Sandeen wrote:
> On 2/12/20 7:41 PM, Darrick J. Wong wrote:
> > On Wed, Feb 12, 2020 at 05:51:18PM -0600, Eric Sandeen wrote:
> >> On 12/31/19 7:11 PM, Darrick J. Wong wrote:
> >>> From: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
> >>>
> >>> Refactor the open-coded test for whether or not we're over quota.
> >>
> >> Ooh, nice. This was horrible.
>
> ...
>
> >>> +}
> >>> +
> >>> /*
> >>> * Check the limits and timers of a dquot and start or reset timers
> >>> * if necessary.
> >>> @@ -117,6 +128,8 @@ xfs_qm_adjust_dqtimers(
> >>> struct xfs_mount *mp,
> >>> struct xfs_disk_dquot *d)
> >>> {
> >>> + bool over;
> >>> +
> >>> ASSERT(d->d_id);
> >>>
> >>> #ifdef DEBUG
> >>> @@ -131,71 +144,47 @@ xfs_qm_adjust_dqtimers(
> >>> be64_to_cpu(d->d_rtb_hardlimit));
> >>> #endif
> >>>
> >>> + over = xfs_quota_exceeded(&d->d_bcount, &d->d_blk_softlimit,
> >>> + &d->d_blk_hardlimit);
> >>> if (!d->d_btimer) {
> >>> - if ((d->d_blk_softlimit && (be64_to_cpu(d->d_bcount) > be64_to_cpu(d->d_blk_softlimit))) ||
> >>> - (d->d_blk_hardlimit && (be64_to_cpu(d->d_bcount) > be64_to_cpu(d->d_blk_hardlimit)))) {
> >>> + if (over) {
> >>
> >> I wonder why we check the hard limit. Isn't exceeding the soft limit
> >> enough to start the timer? Unrelated to the refactoring tho.
> >
> > Suppose there's only a hard limit set?
>
> then you get EDQUOT straightaway and timers don't matter?
Hm. Maybe the idea here was that you always start the timer even if you
just hard-failed the operation? So that we don't have to deal with
weird cases where timers don't always get started?
> I guess if you set a soft limit after you go over a hard-limit-only and ...
> meh, ain't broke don't fix?
"Behavior changes should be not be in refactoring patches"? :)
> -Eric
>
