From 5a33911fa5ecd7395115df2e27fbd22b73357ac5 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 21 Jul 2015 23:50:24 -0400 Subject: ext4: replace ext4_io_submit->io_op with ->io_wbc ext4_io_submit_init() takes the pointer to writeback_control to test its sync_mode and determine between WRITE and WRITE_SYNC and records the result in ->io_op. This patch makes it record the pointer directly and moves the test to ext4_io_submit(). This doesn't cause any noticeable differences now but having writeback_control available throughout IO submission path will be depended upon by the planned cgroup writeback support. Signed-off-by: Tejun Heo Signed-off-by: Theodore Ts'o --- fs/ext4/page-io.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs/ext4/page-io.c') diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index 5602450f03f6..a917bfe3e70c 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -357,8 +357,10 @@ void ext4_io_submit(struct ext4_io_submit *io) struct bio *bio = io->io_bio; if (bio) { + int io_op = io->io_wbc->sync_mode == WB_SYNC_ALL ? + WRITE_SYNC : WRITE; bio_get(io->io_bio); - submit_bio(io->io_op, io->io_bio); + submit_bio(io_op, io->io_bio); bio_put(io->io_bio); } io->io_bio = NULL; @@ -367,7 +369,7 @@ void ext4_io_submit(struct ext4_io_submit *io) void ext4_io_submit_init(struct ext4_io_submit *io, struct writeback_control *wbc) { - io->io_op = (wbc->sync_mode == WB_SYNC_ALL ? WRITE_SYNC : WRITE); + io->io_wbc = wbc; io->io_bio = NULL; io->io_end = NULL; } -- cgit v1.2.3 From 001e4a8775f6e8ad52a89e0072f09aee47d5d252 Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Tue, 21 Jul 2015 23:51:26 -0400 Subject: ext4: implement cgroup writeback support For ordered and writeback data modes, all data IOs go through ext4_io_submit. This patch adds cgroup writeback support by invoking wbc_init_bio() from io_submit_init_bio() and wbc_account_io() in io_submit_add_bh(). Journal data which is written by jbd2 worker is left alone by this patch and will always be written out from the root cgroup. ext4_fill_super() is updated to set MS_CGROUPWB when data mode is either ordered or writeback. In journaled data mode, most IOs become synchronous through the journal and enabling cgroup writeback support doesn't make much sense or difference. Journaled data mode is left alone. Lightly tested with sequential data write workload. Behaves as expected. Signed-off-by: Tejun Heo Signed-off-by: Theodore Ts'o --- fs/ext4/page-io.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs/ext4/page-io.c') diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index a917bfe3e70c..58ab2e3dd114 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -383,6 +383,7 @@ static int io_submit_init_bio(struct ext4_io_submit *io, bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES)); if (!bio) return -ENOMEM; + wbc_init_bio(io->io_wbc, bio); bio->bi_iter.bi_sector = bh->b_blocknr * (bh->b_size >> 9); bio->bi_bdev = bh->b_bdev; bio->bi_end_io = ext4_end_bio; @@ -411,6 +412,7 @@ submit_and_retry: ret = bio_add_page(io->io_bio, page, bh->b_size, bh_offset(bh)); if (ret != bh->b_size) goto submit_and_retry; + wbc_account_io(io->io_wbc, page, bh->b_size); io->io_next_block++; return 0; } -- cgit v1.2.3 From 4246a0b63bd8f56a1469b12eafeb875b1041a451 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 20 Jul 2015 15:29:37 +0200 Subject: block: add a bi_error field to struct bio Currently we have two different ways to signal an I/O error on a BIO: (1) by clearing the BIO_UPTODATE flag (2) by returning a Linux errno value to the bi_end_io callback The first one has the drawback of only communicating a single possible error (-EIO), and the second one has the drawback of not beeing persistent when bios are queued up, and are not passed along from child to parent bio in the ever more popular chaining scenario. Having both mechanisms available has the additional drawback of utterly confusing driver authors and introducing bugs where various I/O submitters only deal with one of them, and the others have to add boilerplate code to deal with both kinds of error returns. So add a new bi_error field to store an errno value directly in struct bio and remove the existing mechanisms to clean all this up. Signed-off-by: Christoph Hellwig Reviewed-by: Hannes Reinecke Reviewed-by: NeilBrown Signed-off-by: Jens Axboe --- fs/ext4/page-io.c | 15 ++++++--------- 1 file changed, 6 insertions(+), 9 deletions(-) (limited to 'fs/ext4/page-io.c') diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index 5602450f03f6..aa95566f14be 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -61,7 +61,6 @@ static void buffer_io_error(struct buffer_head *bh) static void ext4_finish_bio(struct bio *bio) { int i; - int error = !test_bit(BIO_UPTODATE, &bio->bi_flags); struct bio_vec *bvec; bio_for_each_segment_all(bvec, bio, i) { @@ -88,7 +87,7 @@ static void ext4_finish_bio(struct bio *bio) } #endif - if (error) { + if (bio->bi_error) { SetPageError(page); set_bit(AS_EIO, &page->mapping->flags); } @@ -107,7 +106,7 @@ static void ext4_finish_bio(struct bio *bio) continue; } clear_buffer_async_write(bh); - if (error) + if (bio->bi_error) buffer_io_error(bh); } while ((bh = bh->b_this_page) != head); bit_spin_unlock(BH_Uptodate_Lock, &head->b_state); @@ -310,27 +309,25 @@ ext4_io_end_t *ext4_get_io_end(ext4_io_end_t *io_end) } /* BIO completion function for page writeback */ -static void ext4_end_bio(struct bio *bio, int error) +static void ext4_end_bio(struct bio *bio) { ext4_io_end_t *io_end = bio->bi_private; sector_t bi_sector = bio->bi_iter.bi_sector; BUG_ON(!io_end); bio->bi_end_io = NULL; - if (test_bit(BIO_UPTODATE, &bio->bi_flags)) - error = 0; - if (error) { + if (bio->bi_error) { struct inode *inode = io_end->inode; ext4_warning(inode->i_sb, "I/O error %d writing to inode %lu " "(offset %llu size %ld starting block %llu)", - error, inode->i_ino, + bio->bi_error, inode->i_ino, (unsigned long long) io_end->offset, (long) io_end->size, (unsigned long long) bi_sector >> (inode->i_blkbits - 9)); - mapping_set_error(inode->i_mapping, error); + mapping_set_error(inode->i_mapping, bio->bi_error); } if (io_end->flag & EXT4_IO_END_UNWRITTEN) { -- cgit v1.2.3 From b54ffb73cadcdcff9cc1ae0e11f502407e3e2e4c Mon Sep 17 00:00:00 2001 From: Kent Overstreet Date: Tue, 19 May 2015 14:31:01 +0200 Subject: block: remove bio_get_nr_vecs() We can always fill up the bio now, no need to estimate the possible size based on queue parameters. Acked-by: Steven Whitehouse Signed-off-by: Kent Overstreet [hch: rebased and wrote a changelog] Signed-off-by: Christoph Hellwig Signed-off-by: Ming Lin Signed-off-by: Jens Axboe --- fs/ext4/page-io.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs/ext4/page-io.c') diff --git a/fs/ext4/page-io.c b/fs/ext4/page-io.c index aa95566f14be..8a9d63a0c071 100644 --- a/fs/ext4/page-io.c +++ b/fs/ext4/page-io.c @@ -372,10 +372,9 @@ void ext4_io_submit_init(struct ext4_io_submit *io, static int io_submit_init_bio(struct ext4_io_submit *io, struct buffer_head *bh) { - int nvecs = bio_get_nr_vecs(bh->b_bdev); struct bio *bio; - bio = bio_alloc(GFP_NOIO, min(nvecs, BIO_MAX_PAGES)); + bio = bio_alloc(GFP_NOIO, BIO_MAX_PAGES); if (!bio) return -ENOMEM; bio->bi_iter.bi_sector = bh->b_blocknr * (bh->b_size >> 9); -- cgit v1.2.3