This patch series adds the support for blocksize < pagesize for dioread_nolock feature. Since in case of blocksize < pagesize, we can have multiple small buffers of page as unwritten extents, we need to maintain a vector of these unwritten extents which needs the conversion after the IO is complete. Thus, we maintain a list of tuple <offset, size> pair (io_end_vec) for this & traverse this list to do the unwritten to written conversion. Appreciate any reviews/comments on this patches. Tests completed =============== All (which also passes in default config) "quick" group xfstests are passing. Tested xfstests with below configurations. dioread_nolock with blocksize < pagesize dioread_nolock with blocksize == pagesize without dioread_nolock with blocksize < pagesize without dioread_nolock with blocksize == pagesize ltp/fsx test with multiple iterations of 1 million ops did not show any error. About patches ============= Patch 1 - 3: These are some cleanup and refactoring patches. Patch 4: This patch adds the required support. Patch 5: This patch removes the checks which was not allowing to mount with dioread_nolock when blocksize != pagesize was true. _Patches can be cleanly applied on today's linus tree master branch_ Ritesh Harjani (5): ext4: keep uniform naming convention for io & io_end variables ext4: Add API to bring in support for unwritten io_end_vec conversion ext4: Refactor mpage_map_and_submit_buffers function ext4: Add support for blocksize < pagesize in dioread_nolock ext4: Enable blocksize < pagesize for dioread_nolock fs/ext4/ext4.h | 13 ++++- fs/ext4/extents.c | 49 ++++++++++++----- fs/ext4/inode.c | 136 ++++++++++++++++++++++++++++++---------------- fs/ext4/page-io.c | 110 ++++++++++++++++++++++++------------- fs/ext4/super.c | 10 ---- 5 files changed, 208 insertions(+), 110 deletions(-) -- 2.21.0
