On Mon, May 18, 2020 at 12:32:46AM -0700, Christoph Hellwig wrote:
> Btw, some time after xfsprogs 5.6, xfstests generic/590 start to fail
> with new xfs_check errors in the form of:
>
> rtblock 1048576 beyond end of expected area
> rtblock 1048576 beyond end of expected area
> rtblock 1048580 beyond end of expected area
> rtblock 1048580 beyond end of expected area
> rtblock 1048584 beyond end of expected area
> rtblock 1048584 beyond end of expected area
> rtblock 1048588 beyond end of expected area
> rtblock 1048588 beyond end of expected area
> rtblock 1048592 beyond end of expected area
> rtblock 1048592 beyond end of expected area
> rtblock 1048596 beyond end of expected area
> rtblock 1048596 beyond end of expected area
> rtblock 1048600 beyond end of expected area
>
>
That's probably the bounds check I added in commit
7161cd21b3ed3fa82aef1f13b2bcfae045208573
So either I've gotten the units wron.... aha, I see it:
static inline bool
rdbmap_boundscheck(
xfs_rfsblock_t bno)
{
return bno < mp->m_sb.sb_agblocks;
}
That should have been sb_rblocks.
--D
