On Thu, Sep 29, 2011 at 10:31 PM, Tao Ma <tm@xxxxxx> wrote: > On 09/29/2011 01:08 PM, Yongqiang Yang wrote: >> This patch let ext4 maintain delayed extent trees. >> >> Signed-off-by: Yongqiang Yang <xiaoqiangnk@xxxxxxxxx> >> --- >> fs/ext4/ext4.h | 1 + >> fs/ext4/extents.c | 2 ++ >> fs/ext4/indirect.c | 3 +++ >> fs/ext4/inode.c | 28 ++++++++++++++++++++++++++-- >> fs/ext4/super.c | 12 +++++++++++- >> 5 files changed, 43 insertions(+), 3 deletions(-) >> >> diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h >> index d3c6b97..177ec0a 100644 >> --- a/fs/ext4/ext4.h >> +++ b/fs/ext4/ext4.h >> @@ -519,6 +519,7 @@ struct ext4_new_group_data { >> #define EXT4_GET_BLOCKS_PUNCH_OUT_EXT 0x0020 >> /* Don't normalize allocation size (used for fallocate) */ >> #define EXT4_GET_BLOCKS_NO_NORMALIZE 0x0040 >> +#define EXT4_GET_BLOCKS_DEALLOC 0x0080 >> >> /* >> * Flags used by ext4_free_blocks >> diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c >> index 9124cd2..bdbb984 100644 >> --- a/fs/ext4/extents.c >> +++ b/fs/ext4/extents.c >> @@ -3688,6 +3688,8 @@ void ext4_ext_truncate(struct inode *inode) >> >> last_block = (inode->i_size + sb->s_blocksize - 1) >> >> EXT4_BLOCK_SIZE_BITS(sb); >> + err = ext4_de_remove_space(inode, last_block, >> + EXT_MAX_BLOCKS - last_block); >> err = ext4_ext_remove_space(inode, last_block); >> >> /* In a multi-transaction truncate, we only make the final >> diff --git a/fs/ext4/indirect.c b/fs/ext4/indirect.c >> index 0962642..25cdb5b 100644 >> --- a/fs/ext4/indirect.c >> +++ b/fs/ext4/indirect.c >> @@ -22,6 +22,7 @@ >> >> #include <linux/module.h> >> #include "ext4_jbd2.h" >> +#include "ext4_extents.h" >> #include "truncate.h" >> >> #include <trace/events/ext4.h> >> @@ -1383,6 +1384,8 @@ void ext4_ind_truncate(struct inode *inode) >> down_write(&ei->i_data_sem); >> >> ext4_discard_preallocations(inode); >> + ext4_de_remove_space(inode, last_block, >> + EXT_MAX_BLOCKS - last_block); >> >> /* >> * The orphan list entry will now protect us from any crash which >> diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c >> index f86b149..0f9f108 100644 >> --- a/fs/ext4/inode.c >> +++ b/fs/ext4/inode.c >> @@ -442,7 +442,15 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode, >> up_read((&EXT4_I(inode)->i_data_sem)); >> >> if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) { >> - int ret = check_block_validity(inode, map); >> + int ret; >> + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) { >> + /* delayed alloc may be allocated by fallocate, >> + * we need to handle delayed extent here. >> + */ >> + down_write((&EXT4_I(inode)->i_data_sem)); >> + goto delayed_mapped; >> + } >> + ret = check_block_validity(inode, map); > I am not quite sure of this. So do you mean when we write_begin the > extent isn't allocated, while in the time of writepage, the extent is > fallocted, right? If this is the case, where do we update the reserve_space? > I mean in ext4_da_get_block_prep, we call ext4_da_reserve_space, and if > there is no fallocate, we will call ext4_da_update_reserve_space in > ext4_ext_handle_uninitialized_extents. So in your case, the 2nd > ext4_da_update_reserve_space wouldn't be called. I am not sure whether > there will be some problem or not. Hi Tao, What if fallocated blocks have been allocated by direct I/O? I placed a BUG() here, it was indeed triggered by xfstests 127. The comment is not clear, sorry for that. Yongqiang. > > Thanks > Tao >> if (ret != 0) >> return ret; >> } >> @@ -517,8 +525,18 @@ int ext4_map_blocks(handle_t *handle, struct inode *inode, >> (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE)) >> ext4_da_update_reserve_space(inode, retval, 1); >> } >> - if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) >> + if (flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) { >> ext4_clear_inode_state(inode, EXT4_STATE_DELALLOC_RESERVED); >> +delayed_mapped: >> + if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) { >> + int ret; >> + /* delayed allocation blocks has been allocated */ >> + ret = ext4_de_remove_space(inode, map->m_lblk, >> + map->m_len); >> + if (ret < 0) >> + retval = ret; >> + } >> + } >> >> up_write((&EXT4_I(inode)->i_data_sem)); >> if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) { >> @@ -1630,6 +1648,12 @@ static int ext4_da_get_block_prep(struct inode *inode, sector_t iblock, >> /* not enough space to reserve */ >> return ret; >> >> + down_write((&EXT4_I(inode)->i_data_sem)); >> + ret = ext4_de_add_space(inode, map.m_lblk, map.m_len); >> + up_write((&EXT4_I(inode)->i_data_sem)); >> + if (ret) >> + return ret; >> + >> map_bh(bh, inode->i_sb, invalid_block); >> set_buffer_new(bh); >> set_buffer_delay(bh); >> diff --git a/fs/ext4/super.c b/fs/ext4/super.c >> index 247fcdd..a248551 100644 >> --- a/fs/ext4/super.c >> +++ b/fs/ext4/super.c >> @@ -49,6 +49,7 @@ >> #include "xattr.h" >> #include "acl.h" >> #include "mballoc.h" >> +#include "ext4_extents.h" >> >> #define CREATE_TRACE_POINTS >> #include <trace/events/ext4.h> >> @@ -967,6 +968,7 @@ void ext4_clear_inode(struct inode *inode) >> end_writeback(inode); >> dquot_drop(inode); >> ext4_discard_preallocations(inode); >> + ext4_de_remove_space(inode, 0, EXT_MAX_BLOCKS); >> if (EXT4_I(inode)->jinode) { >> jbd2_journal_release_jbd_inode(EXT4_JOURNAL(inode), >> EXT4_I(inode)->jinode); >> @@ -4976,9 +4978,14 @@ static int __init ext4_init_fs(void) >> init_waitqueue_head(&ext4__ioend_wq[i]); >> } >> >> - err = ext4_init_pageio(); >> + err = ext4_init_de(); >> if (err) >> return err; >> + >> + err = ext4_init_pageio(); >> + if (err) >> + goto out8; >> + >> err = ext4_init_system_zone(); >> if (err) >> goto out7; >> @@ -5030,6 +5037,9 @@ out6: >> ext4_exit_system_zone(); >> out7: >> ext4_exit_pageio(); >> +out8: >> + ext4_exit_de(); >> + >> return err; >> } >> > > -- Best Wishes Yongqiang Yang -- To unsubscribe from this list: send the line "unsubscribe linux-ext4" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html