On Fri, Mar 29, 2024 at 09:21:23AM -0700, Darrick J. Wong wrote:
> > +static inline void
> > xfs_quota_unreserve_blkres(struct xfs_inode *ip, int64_t blocks)
> > {
> > - return xfs_quota_reserve_blkres(ip, -blocks);
> > + /* don't return an error as unreserving quotas can't fail */
> > + xfs_quota_reserve_blkres(ip, -blocks);
>
> xfs_quota_reserve_blkres only doesn't fail if the nblks argument is
> actually negative. Can we have an ASSERT(blocks >= 0) here to guard
> against someone accidentally passing in a negative @blocks here?
Sure. Or even better just mark blocks as unsigned?
