On Tue, Nov 29, 2022 at 10:30:21AM +1100, Dave Chinner wrote:
> On Mon, Nov 28, 2022 at 02:34:05PM -0800, Darrick J. Wong wrote:
> > From: Darrick J. Wong <djwong@xxxxxxxxxx>
> >
> > Add a new error injection knob so that we can arbitrarily slow down
> > writeback to test for race conditions and aberrant reclaim behavior if
> > the writeback mechanisms are slow to issue writeback. This will enable
> > functional testing for the ifork sequence counters introduced in commit
> > 745b3f76d1c8 ("xfs: maintain a sequence count for inode fork
> > manipulations").
> >
> > Signed-off-by: Darrick J. Wong <djwong@xxxxxxxxxx>
> > ---
> > fs/xfs/libxfs/xfs_errortag.h | 4 +++-
> > fs/xfs/xfs_aops.c | 12 ++++++++++--
> > fs/xfs/xfs_error.c | 11 +++++++++++
> > fs/xfs/xfs_error.h | 22 ++++++++++++++++++++++
> > 4 files changed, 46 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/xfs/libxfs/xfs_errortag.h b/fs/xfs/libxfs/xfs_errortag.h
> > index 580ccbd5aadc..f5f629174eca 100644
> > --- a/fs/xfs/libxfs/xfs_errortag.h
> > +++ b/fs/xfs/libxfs/xfs_errortag.h
> > @@ -61,7 +61,8 @@
> > #define XFS_ERRTAG_LARP 39
> > #define XFS_ERRTAG_DA_LEAF_SPLIT 40
> > #define XFS_ERRTAG_ATTR_LEAF_TO_NODE 41
> > -#define XFS_ERRTAG_MAX 42
> > +#define XFS_ERRTAG_WB_DELAY_MS 42
> > +#define XFS_ERRTAG_MAX 43
> >
> > /*
> > * Random factors for above tags, 1 means always, 2 means 1/2 time, etc.
> > @@ -107,5 +108,6 @@
> > #define XFS_RANDOM_LARP 1
> > #define XFS_RANDOM_DA_LEAF_SPLIT 1
> > #define XFS_RANDOM_ATTR_LEAF_TO_NODE 1
> > +#define XFS_RANDOM_WB_DELAY_MS 3000
> >
> > #endif /* __XFS_ERRORTAG_H_ */
> > diff --git a/fs/xfs/xfs_aops.c b/fs/xfs/xfs_aops.c
> > index a22d90af40c8..4a13260527b9 100644
> > --- a/fs/xfs/xfs_aops.c
> > +++ b/fs/xfs/xfs_aops.c
> > @@ -17,6 +17,8 @@
> > #include "xfs_bmap.h"
> > #include "xfs_bmap_util.h"
> > #include "xfs_reflink.h"
> > +#include "xfs_errortag.h"
> > +#include "xfs_error.h"
> >
> > struct xfs_writepage_ctx {
> > struct iomap_writepage_ctx ctx;
> > @@ -217,11 +219,15 @@ xfs_imap_valid(
> > * checked (and found nothing at this offset) could have added
> > * overlapping blocks.
> > */
> > - if (XFS_WPC(wpc)->data_seq != READ_ONCE(ip->i_df.if_seq))
> > + if (XFS_WPC(wpc)->data_seq != READ_ONCE(ip->i_df.if_seq)) {
> > + XFS_ERRORTAG_REPORT(ip->i_mount, XFS_ERRTAG_WB_DELAY_MS);
> > return false;
> > + }
> > if (xfs_inode_has_cow_data(ip) &&
> > - XFS_WPC(wpc)->cow_seq != READ_ONCE(ip->i_cowfp->if_seq))
> > + XFS_WPC(wpc)->cow_seq != READ_ONCE(ip->i_cowfp->if_seq)) {
> > + XFS_ERRORTAG_REPORT(ip->i_mount, XFS_ERRTAG_WB_DELAY_MS);
> > return false;
>
> These should be tracepoints, right?
>
> Otherwise I don't see a problem with the delay code.
Yeah. I suppose this'll be the first fstest that has to go wrangle with
setting up its own tracepoint filtering. :)
--D
> -Dave.
> --
> Dave Chinner
> david@xxxxxxxxxxxxx
