Hi This is the version 2 patch set that attempts to get discard out of the jbd2 commit kthread. When the user delete a lot data and cause discard flooding, the jbd2 commit kthread can be blocked for very long time and then all of the metadata operations are blocked due to no journal space. This version patch is based on Ted's suggestion that free the blocks after commit immediately and then use a background kworker to do the discard with the fstrim fashion. The discard maybe slower than before but it doesn't matter as the blocks has been freed back to mb allocator and nothing would be blocked. There are 7 patches, Patch 1 ~ 4, there are no functional changes in them, but just some preparation for following patches Patch 5 does the buddy cache get only after insert ext4_free_data successfully Patch 6 get rid of the s_freed_data_list which carries ext4_free_data and iterate free data rbtree to get ext4_free_data. This is also a preparation. Patch 7 introduces a async kworker to do discard which implements the core idea of this patch set. Any comments are welcome ;) Chagnes from V1: - free blocks back to mb allocator immediately after commit and do discard in fstrim fashion. Wang Jianchao (7): ext4: remove the 'group' parameter of ext4_trim_extent ext4: add new helper interface ext4_try_to_trim_range() ext4: add new helper interface ext4_insert_free_data ext4: remove the repeated comment of ext4_trim_all_free ext4: get buddy cache after insert successfully ext4: use bb_free_root to get the free data entry ext4: get discard out of jbd2 commit kthread context fs/ext4/balloc.c | 2 +- fs/ext4/ext4.h | 6 +- fs/ext4/mballoc.c | 434 ++++++++++++++++++++++++++++++++++++++++++++------------------------------------ fs/ext4/mballoc.h | 3 - 4 files changed, 241 insertions(+), 204 deletions(-)
