| Message ID | 20240712064716.3385793-2-leo.lilong@huawei.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [1/2] xfs/016: fix test fail when head equal to near_end_min | expand |
On Fri, Jul 12, 2024 at 02:47:16PM +0800, Long Li wrote: > I got a failure in xfs/242 as follows, it can be easily reproduced > when I run xfs/242 as a cyclic test. > > 13. data -> unwritten -> data > 0: [0..127]: data > -1: [128..511]: unwritten > -2: [512..639]: data > +1: [128..639]: unwritten > > The root cause, as Dave pointed out in previous email [1], is that > _filter_bmap may incorrectly match the AG-OFFSET in column 5 for datadev > files. On the other hand, _filter_bmap missing a "next" to jump out when > it matches "data" in the 5th column, otherwise it might print the result > twice. The issue was introduced by commit 7d5d3f77154e ("xfs/242: fix > _filter_bmap for xfs_io bmap that does rt file properly"). The failure > disappeared when I retest xfs/242 by reverted commit 7d5d3f77154e. > > Fix it by matching the 7th column first and then the 5th column in > _filter_bmap, because the rtdev file only has 5 columns in the `bmap -vp` > output. > > [1] https://lore.kernel.org/all/Zh9UkHEesvrpSQ7J@dread.disaster.area/ > Signed-off-by: Long Li <leo.lilong@huawei.com> > --- > common/punch | 14 +++++++++----- > 1 file changed, 9 insertions(+), 5 deletions(-) > > diff --git a/common/punch b/common/punch > index 9e730404..43ccab69 100644 > --- a/common/punch > +++ b/common/punch > @@ -188,7 +188,10 @@ _filter_hole_fiemap() > _coalesce_extents > } > > -# Column 7 for datadev files and column 5 for rtdev files > +# Column 7 for datadev files and column 5 for rtdev files, To prevent the > +# 5th column in datadev files from being potentially matched incorrectly, > +# we need to match Column 7 for datadev files first, because the rtdev > +# file only has 5 columns in the `bmap -vp` output. Yeah, checking column 7 before 5 seems reasonable. Longer term, maybe these tools should emit json to cut down on the amount of string parsing that fstests has to do. Reviewed-by: Darrick J. Wong <djwong@kernel.org> --D > # 10000 Unwritten preallocated extent > # 01000 Doesn't begin on stripe unit > # 00100 Doesn't end on stripe unit > @@ -201,18 +204,19 @@ _filter_bmap() > print $1, $2, $3; > next; > } > - $5 ~ /1[01][01][01][01]/ { > + $7 ~ /1[01][01][01][01]/ { > print $1, $2, "unwritten"; > next; > } > - $5 ~ /0[01][01][01][01]/ { > + $7 ~ /0[01][01][01][01]/ { > print $1, $2, "data" > + next; > } > - $7 ~ /1[01][01][01][01]/ { > + $5 ~ /1[01][01][01][01]/ { > print $1, $2, "unwritten"; > next; > } > - $7 ~ /0[01][01][01][01]/ { > + $5 ~ /0[01][01][01][01]/ { > print $1, $2, "data" > }' | > _coalesce_extents > -- > 2.39.2 > >
diff --git a/common/punch b/common/punch index 9e730404..43ccab69 100644 --- a/common/punch +++ b/common/punch @@ -188,7 +188,10 @@ _filter_hole_fiemap() _coalesce_extents } -# Column 7 for datadev files and column 5 for rtdev files +# Column 7 for datadev files and column 5 for rtdev files, To prevent the +# 5th column in datadev files from being potentially matched incorrectly, +# we need to match Column 7 for datadev files first, because the rtdev +# file only has 5 columns in the `bmap -vp` output. # 10000 Unwritten preallocated extent # 01000 Doesn't begin on stripe unit # 00100 Doesn't end on stripe unit @@ -201,18 +204,19 @@ _filter_bmap() print $1, $2, $3; next; } - $5 ~ /1[01][01][01][01]/ { + $7 ~ /1[01][01][01][01]/ { print $1, $2, "unwritten"; next; } - $5 ~ /0[01][01][01][01]/ { + $7 ~ /0[01][01][01][01]/ { print $1, $2, "data" + next; } - $7 ~ /1[01][01][01][01]/ { + $5 ~ /1[01][01][01][01]/ { print $1, $2, "unwritten"; next; } - $7 ~ /0[01][01][01][01]/ { + $5 ~ /0[01][01][01][01]/ { print $1, $2, "data" }' | _coalesce_extents
I got a failure in xfs/242 as follows, it can be easily reproduced when I run xfs/242 as a cyclic test. 13. data -> unwritten -> data 0: [0..127]: data -1: [128..511]: unwritten -2: [512..639]: data +1: [128..639]: unwritten The root cause, as Dave pointed out in previous email [1], is that _filter_bmap may incorrectly match the AG-OFFSET in column 5 for datadev files. On the other hand, _filter_bmap missing a "next" to jump out when it matches "data" in the 5th column, otherwise it might print the result twice. The issue was introduced by commit 7d5d3f77154e ("xfs/242: fix _filter_bmap for xfs_io bmap that does rt file properly"). The failure disappeared when I retest xfs/242 by reverted commit 7d5d3f77154e. Fix it by matching the 7th column first and then the 5th column in _filter_bmap, because the rtdev file only has 5 columns in the `bmap -vp` output. [1] https://lore.kernel.org/all/Zh9UkHEesvrpSQ7J@dread.disaster.area/ Signed-off-by: Long Li <leo.lilong@huawei.com> --- common/punch | 14 +++++++++----- 1 file changed, 9 insertions(+), 5 deletions(-)