Message ID | 20160715004629.43143-5-jaegeuk@kernel.org (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
On Thu, Jul 14, 2016 at 05:46:29PM -0700, Jaegeuk Kim wrote: > The main reason is that blk_start_plug can be used to wake up from low-power > mode before submitting further bios. Where does that myth come from? -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Jul 14, 2016 at 07:21:32PM -0700, Christoph Hellwig wrote: > On Thu, Jul 14, 2016 at 05:46:29PM -0700, Jaegeuk Kim wrote: > > The main reason is that blk_start_plug can be used to wake up from low-power > > mode before submitting further bios. > > Where does that myth come from? From kernel guys working on android. And, I'm in also doubt they may have done something regarding to power mode change, by the fact that I've seen annoying performance jitters comparing to ext4's ones all the time in some android products. Frankly speaking, I cannot find a concrete reason to remove plugging entirely, since I can't see big performance differences in all my test results in severs as well. Thanks, -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Thu, Jul 14, 2016 at 08:05:02PM -0700, Jaegeuk Kim wrote:
> >From kernel guys working on android.
Well, until it's mainline it simply doesn't matter, so NAK to this
patch. Tying power behavior to plugs also sounds pretty broken, so we'd
probably come up with something better if they bothered to actually
submit their patches upsteam or at least explaining what they want
to archive.
--
To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
On Mon, Jul 18, 2016 at 08:59:52PM -0700, Christoph Hellwig wrote: > On Thu, Jul 14, 2016 at 08:05:02PM -0700, Jaegeuk Kim wrote: > > >From kernel guys working on android. > > Well, until it's mainline it simply doesn't matter, so NAK to this > patch. Tying power behavior to plugs also sounds pretty broken, so we'd > probably come up with something better if they bothered to actually > submit their patches upsteam or at least explaining what they want > to archive. Actually, the initial patch which drops plugs is not mainlined as well. So, at this time, I just want to revert it by this patch, since I recognized that there might be whatever possible use-cases of plugs in another way someday. So for now, I lost the reason to eliminate the plugs. Moreover, since every filesystems use plugs, f2fs doesn't need to be an exceptional case in terms of that (except hm-smr). So, let me sync f2fs with other filesystems at this moment. I totally agree that it'd be best to see their patches upstream, but I've witnessed that it becomes a quite tough challenge to them. Thanks, -- To unsubscribe from this list: send the line "unsubscribe linux-fsdevel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/f2fs/checkpoint.c b/fs/f2fs/checkpoint.c index 8ea8953..be1c54b 100644 --- a/fs/f2fs/checkpoint.c +++ b/fs/f2fs/checkpoint.c @@ -265,6 +265,7 @@ static int f2fs_write_meta_pages(struct address_space *mapping, struct writeback_control *wbc) { struct f2fs_sb_info *sbi = F2FS_M_SB(mapping); + struct blk_plug plug; long diff, written; /* collect a number of dirty meta pages and write together */ @@ -277,7 +278,9 @@ static int f2fs_write_meta_pages(struct address_space *mapping, /* if mounting is failed, skip writing node pages */ mutex_lock(&sbi->cp_mutex); diff = nr_pages_to_write(sbi, META, wbc); + blk_start_plug(&plug); written = sync_meta_pages(sbi, META, wbc->nr_to_write); + blk_finish_plug(&plug); mutex_unlock(&sbi->cp_mutex); wbc->nr_to_write = max((long)0, wbc->nr_to_write - written - diff); return 0; @@ -899,8 +902,11 @@ static int block_operations(struct f2fs_sb_info *sbi) .nr_to_write = LONG_MAX, .for_reclaim = 0, }; + struct blk_plug plug; int err = 0; + blk_start_plug(&plug); + retry_flush_dents: f2fs_lock_all(sbi); /* write all the dirty dentry pages */ @@ -937,6 +943,7 @@ retry_flush_nodes: goto retry_flush_nodes; } out: + blk_finish_plug(&plug); return err; } diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c index 6500995..b7a530e 100644 --- a/fs/f2fs/data.c +++ b/fs/f2fs/data.c @@ -1438,6 +1438,7 @@ static int f2fs_write_data_pages(struct address_space *mapping, { struct inode *inode = mapping->host; struct f2fs_sb_info *sbi = F2FS_I_SB(inode); + struct blk_plug plug; int ret; /* deal with chardevs and other special file */ @@ -1463,7 +1464,9 @@ static int f2fs_write_data_pages(struct address_space *mapping, trace_f2fs_writepages(mapping->host, wbc, DATA); + blk_start_plug(&plug); ret = f2fs_write_cache_pages(mapping, wbc); + blk_finish_plug(&plug); /* * if some pages were truncated, we cannot guarantee its mapping->host * to detect pending bios. diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index c529884..001f062 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -2180,6 +2180,7 @@ static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from) { struct file *file = iocb->ki_filp; struct inode *inode = file_inode(file); + struct blk_plug plug; ssize_t ret; if (f2fs_encrypted_inode(inode) && @@ -2191,8 +2192,11 @@ static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from) ret = generic_write_checks(iocb, from); if (ret > 0) { ret = f2fs_preallocate_blocks(iocb, from); - if (!ret) + if (!ret) { + blk_start_plug(&plug); ret = __generic_file_write_iter(iocb, from); + blk_finish_plug(&plug); + } } inode_unlock(inode); diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c index c612137..90d5fda 100644 --- a/fs/f2fs/gc.c +++ b/fs/f2fs/gc.c @@ -788,6 +788,7 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi, { struct page *sum_page; struct f2fs_summary_block *sum; + struct blk_plug plug; unsigned int segno = start_segno; unsigned int end_segno = start_segno + sbi->segs_per_sec; int seg_freed = 0; @@ -805,6 +806,8 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi, unlock_page(sum_page); } + blk_start_plug(&plug); + for (segno = start_segno; segno < end_segno; segno++) { if (get_valid_blocks(sbi, segno, 1) == 0) @@ -842,6 +845,8 @@ static int do_garbage_collect(struct f2fs_sb_info *sbi, f2fs_submit_merged_bio(sbi, (type == SUM_TYPE_NODE) ? NODE : DATA, WRITE); + blk_finish_plug(&plug); + if (gc_type == FG_GC) { while (start_segno < end_segno) if (get_valid_blocks(sbi, start_segno++, 1) == 0) diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index d78f61d..79a93c6 100644 --- a/fs/f2fs/node.c +++ b/fs/f2fs/node.c @@ -1618,6 +1618,7 @@ static int f2fs_write_node_pages(struct address_space *mapping, struct writeback_control *wbc) { struct f2fs_sb_info *sbi = F2FS_M_SB(mapping); + struct blk_plug plug; long diff; /* balancing f2fs's metadata in background */ @@ -1631,7 +1632,9 @@ static int f2fs_write_node_pages(struct address_space *mapping, diff = nr_pages_to_write(sbi, NODE, wbc); wbc->sync_mode = WB_SYNC_NONE; + blk_start_plug(&plug); sync_node_pages(sbi, wbc); + blk_finish_plug(&plug); wbc->nr_to_write = max((long)0, wbc->nr_to_write - diff); return 0; diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index e87aa05..d45e6bb 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -381,8 +381,13 @@ void f2fs_balance_fs_bg(struct f2fs_sb_info *sbi) excess_prefree_segs(sbi) || excess_dirty_nats(sbi) || (is_idle(sbi) && f2fs_time_over(sbi, CP_TIME))) { - if (test_opt(sbi, DATA_FLUSH)) + if (test_opt(sbi, DATA_FLUSH)) { + struct blk_plug plug; + + blk_start_plug(&plug); sync_dirty_inodes(sbi, FILE_INODE); + blk_finish_plug(&plug); + } f2fs_sync_fs(sbi->sb, true); stat_inc_bg_cp_count(sbi->stat_info); }
This patch reverts 19a5f5e2ef37 (f2fs: drop any block plugging), and adds blk_plug in write paths additionally. The main reason is that blk_start_plug can be used to wake up from low-power mode before submitting further bios. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> --- fs/f2fs/checkpoint.c | 7 +++++++ fs/f2fs/data.c | 3 +++ fs/f2fs/file.c | 6 +++++- fs/f2fs/gc.c | 5 +++++ fs/f2fs/node.c | 3 +++ fs/f2fs/segment.c | 7 ++++++- 6 files changed, 29 insertions(+), 2 deletions(-)