Hi all, Currently, i_mmap_lock is huge bottleneck for DAX scalability as we use in place of lock_page().
This patchset tries to recover some scalability by introducing per-mapping range-lock. The range-lock itself is implemented by Jan Kara on top of interval tree. It looks not so cheap, by should scale better than exclusive i_mmap_lock. Any comments? Jan Kara (1): lib: Implement range locks Kirill A. Shutemov (1): dax: use range_lock instead of i_mmap_lock drivers/gpu/drm/Kconfig | 1 - drivers/gpu/drm/i915/Kconfig | 1 - fs/dax.c | 30 +++++++++-------- fs/inode.c | 1 + include/linux/fs.h | 2 ++ include/linux/range_lock.h | 51 +++++++++++++++++++++++++++++ lib/Kconfig | 14 -------- lib/Kconfig.debug | 1 - lib/Makefile | 3 +- lib/range_lock.c | 78 ++++++++++++++++++++++++++++++++++++++++++++ mm/memory.c | 35 +++++++++++++------- mm/rmap.c | 1 + 12 files changed, 174 insertions(+), 44 deletions(-) create mode 100644 include/linux/range_lock.h create mode 100644 lib/range_lock.c -- 2.5.0 -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
