I've numbered the new test as 999 for the moment so as not to collide with tests being added while I've been working on this. I can change that and resend if this should go in. I'm working on a set of kernel patches to change how writeback errors are handled and reported in the kernel. Instead of reporting a writeback error to only the first fsync caller on the file, I aim to make the kernel report them once on every file description: http://www.spinics.net/lists/kernel/msg2504453.html This patch adds a test for the new behavior, on local filesystems that can handle journalling to a separate device. Basically, open many fds to the same file, turn on dm_error, write to each of the fds, and then fsync them all to ensure that they all get an error back. Then, flip the device back to working, reopen the files and ensure that no error is reported. With the kernel patch series in place, ext4 and xfs now pass. btrfs still clears the error after the first fsync, so it seems like it still needs a bit of work. Jeff Layton (3): generic: add a writeback error handling test ext4: allow ext4 to use $SCRATCH_LOGDEV btrfs: allow it to use $SCRATCH_LOGDEV common/dmerror | 13 +++-- common/rc | 5 ++ src/Makefile | 2 +- src/fsync-err.c | 138 ++++++++++++++++++++++++++++++++++++++++++++++++++ tests/generic/999 | 75 +++++++++++++++++++++++++++ tests/generic/999.out | 3 ++ tests/generic/group | 1 + tools/dmerror | 47 +++++++++++++++++ 8 files changed, 278 insertions(+), 6 deletions(-) create mode 100644 src/fsync-err.c create mode 100755 tests/generic/999 create mode 100644 tests/generic/999.out create mode 100755 tools/dmerror -- 2.9.3
