Re: [PATCH] f2fs: submit cached bio to avoid endless PageWriteback

From: Jaegeuk Kim
Date: Tue Sep 11 2018 - 18:39:22 EST


On 09/11, Chao Yu wrote:
> When migrating encrypted block from background GC thread, we only add
> them into f2fs inner bio cache, but forget to submit the cached bio, it
> may cause potential deadlock when we are waiting page writebacked, fix
> it.
>
> Signed-off-by: Chao Yu <yuchao0@xxxxxxxxxx>
> ---
> fs/f2fs/gc.c | 42 +++++++++++++++++++++++++++++-------------
> 1 file changed, 29 insertions(+), 13 deletions(-)
>
> diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c
> index c4ea4009cf05..a2ea0d445345 100644
> --- a/fs/f2fs/gc.c
> +++ b/fs/f2fs/gc.c
> @@ -687,7 +687,7 @@ static int ra_data_block(struct inode *inode, pgoff_t index)
> * Move data block via META_MAPPING while keeping locked data page.
> * This can be used to move blocks, aka LBAs, directly on disk.
> */
> -static void move_data_block(struct inode *inode, block_t bidx,
> +static int move_data_block(struct inode *inode, block_t bidx,
> int gc_type, unsigned int segno, int off)
> {
> struct f2fs_io_info fio = {
> @@ -706,25 +706,29 @@ static void move_data_block(struct inode *inode, block_t bidx,
> struct node_info ni;
> struct page *page, *mpage;
> block_t newaddr;
> - int err;
> + int err = 0;
> bool lfs_mode = test_opt(fio.sbi, LFS);
>
> /* do not read out */
> page = f2fs_grab_cache_page(inode->i_mapping, bidx, false);
> if (!page)
> - return;
> + return -ENOMEM;
>
> - if (!check_valid_map(F2FS_I_SB(inode), segno, off))
> + if (!check_valid_map(F2FS_I_SB(inode), segno, off)) {
> + err = -ENOENT;
> goto out;
> + }
>
> if (f2fs_is_atomic_file(inode)) {
> F2FS_I(inode)->i_gc_failures[GC_FAILURE_ATOMIC]++;
> F2FS_I_SB(inode)->skipped_atomic_files[gc_type]++;
> + err = -EAGAIN;
> goto out;
> }
>
> if (f2fs_is_pinned_file(inode)) {
> f2fs_pin_file_control(inode, true);
> + err = -EAGAIN;
> goto out;
> }
>
> @@ -735,6 +739,7 @@ static void move_data_block(struct inode *inode, block_t bidx,
>
> if (unlikely(dn.data_blkaddr == NULL_ADDR)) {
> ClearPageUptodate(page);
> + err = -ENOENT;
> goto put_out;
> }
>
> @@ -817,6 +822,7 @@ static void move_data_block(struct inode *inode, block_t bidx,
> fio.new_blkaddr = newaddr;
> f2fs_submit_page_write(&fio);
> if (fio.retry) {
> + err = -EAGAIN;
> if (PageWriteback(fio.encrypted_page))
> end_page_writeback(fio.encrypted_page);
> goto put_page_out;
> @@ -840,6 +846,8 @@ static void move_data_block(struct inode *inode, block_t bidx,
> f2fs_put_dnode(&dn);
> out:
> f2fs_put_page(page, 1);
> +
> + return err;
> }
>
> static void move_data_page(struct inode *inode, block_t bidx, int gc_type,
> @@ -919,7 +927,7 @@ static void move_data_page(struct inode *inode, block_t bidx, int gc_type,
> * If the parent node is not valid or the data block address is different,
> * the victim data block is ignored.
> */
> -static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
> +static int gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
> struct gc_inode_list *gc_list, unsigned int segno, int gc_type)
> {
> struct super_block *sb = sbi->sb;
> @@ -927,6 +935,7 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
> block_t start_addr;
> int off;
> int phase = 0;
> + int submitted = 0;
>
> start_addr = START_BLOCK(sbi, segno);
>
> @@ -943,7 +952,7 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
>
> /* stop BG_GC if there is not enough free sections. */
> if (gc_type == BG_GC && has_not_enough_free_secs(sbi, 0, 0))
> - return;
> + return submitted;
>
> if (check_valid_map(sbi, segno, off) == 0)
> continue;
> @@ -1015,6 +1024,7 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
> if (inode) {
> struct f2fs_inode_info *fi = F2FS_I(inode);
> bool locked = false;
> + int err;
>
> if (S_ISREG(inode->i_mode)) {
> if (!down_write_trylock(&fi->i_gc_rwsem[READ]))
> @@ -1033,12 +1043,15 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
>
> start_bidx = f2fs_start_bidx_of_node(nofs, inode)
> + ofs_in_node;
> - if (f2fs_post_read_required(inode))
> - move_data_block(inode, start_bidx, gc_type,
> - segno, off);
> - else
> + if (f2fs_post_read_required(inode)) {
> + err = move_data_block(inode, start_bidx,
> + gc_type, segno, off);
> + if (!err)
> + submitted++;
> + } else {
> move_data_page(inode, start_bidx, gc_type,
> segno, off);
> + }
>
> if (locked) {
> up_write(&fi->i_gc_rwsem[WRITE]);
> @@ -1051,6 +1064,8 @@ static void gc_data_segment(struct f2fs_sb_info *sbi, struct f2fs_summary *sum,
>
> if (++phase < 5)
> goto next_step;
> +
> + return submitted;
> }
>
> static int __get_victim(struct f2fs_sb_info *sbi, unsigned int *victim,
> @@ -1078,6 +1093,7 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
> unsigned char type = IS_DATASEG(get_seg_entry(sbi, segno)->type) ?
> SUM_TYPE_DATA : SUM_TYPE_NODE;
> unsigned int units = 0;
> + int submitted = 0;
>
> /* start segno may point to middle of section, so adjust it */
> if (__is_large_section(sbi))
> @@ -1136,8 +1152,8 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
> if (type == SUM_TYPE_NODE)
> gc_node_segment(sbi, sum->entries, segno, gc_type);
> else
> - gc_data_segment(sbi, sum->entries, gc_list, segno,
> - gc_type);
> + submitted += gc_data_segment(sbi, sum->entries, gc_list,
> + segno, gc_type);
>
> stat_inc_seg_count(sbi, type, gc_type);
>
> @@ -1152,7 +1168,7 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi,
> f2fs_put_page(sum_page, 0);
> }
>
> - if (gc_type == FG_GC)
> + if (gc_type == FG_GC || submitted)

Do we need to check submitted for both of node and data? Then, we can remove
FG_GC case?

> f2fs_submit_merged_write(sbi,
> (type == SUM_TYPE_NODE) ? NODE : DATA);
>
> --
> 2.18.0.rc1