In current implementation the atomic write may not start before the direct write and the test fails.
The patch adds a delay after starting the atomic write process so it has a chance to actually start wrtiting data and set the internal state of the file correctly. Signed-off-by: Jan Prusakowski <jprusakow...@google.com> --- tests/f2fs/004 | 3 +++ 1 file changed, 3 insertions(+) diff --git a/tests/f2fs/004 b/tests/f2fs/004 index e08cee11..ac824c43 100755 --- a/tests/f2fs/004 +++ b/tests/f2fs/004 @@ -40,6 +40,9 @@ touch $dbfile $F2FS_IO_PROG write 1 0 32 zero atomic_commit $dbfile 3000 >> $seqres.full & atomic_write_pid=$! +# wait a bit to allow the atomic write to start +sleep 2 + # simulate concurrent direct read/write IO $XFS_IO_PROG -d -c "pread 0 128k" $dbfile $XFS_IO_PROG -d -c "pwrite 0 128k" $dbfile -- 2.49.0.1204.g71687c7c1d-goog _______________________________________________ Linux-f2fs-devel mailing list Linux-f2fs-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel