From 87e99511ea54510ffb60b98001d108794d5037f8 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 11 Aug 2010 17:05:45 +0200 Subject: kill BH_Ordered flag Instead of abusing a buffer_head flag just add a variant of sync_dirty_buffer which allows passing the exact type of write flag required. Signed-off-by: Christoph Hellwig Signed-off-by: Al Viro --- fs/jbd2/commit.c | 39 +++++++++++++++------------------------ 1 file changed, 15 insertions(+), 24 deletions(-) (limited to 'fs/jbd2/commit.c') diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index f52e5e8049f1..7c068c189d80 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -101,7 +101,6 @@ static int journal_submit_commit_record(journal_t *journal, struct commit_header *tmp; struct buffer_head *bh; int ret; - int barrier_done = 0; struct timespec now = current_kernel_time(); if (is_journal_aborted(journal)) @@ -136,30 +135,22 @@ static int journal_submit_commit_record(journal_t *journal, if (journal->j_flags & JBD2_BARRIER && !JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) { - set_buffer_ordered(bh); - barrier_done = 1; - } - ret = submit_bh(WRITE_SYNC_PLUG, bh); - if (barrier_done) - clear_buffer_ordered(bh); - - /* is it possible for another commit to fail at roughly - * the same time as this one? If so, we don't want to - * trust the barrier flag in the super, but instead want - * to remember if we sent a barrier request - */ - if (ret == -EOPNOTSUPP && barrier_done) { - printk(KERN_WARNING - "JBD2: Disabling barriers on %s, " - "not supported by device\n", journal->j_devname); - write_lock(&journal->j_state_lock); - journal->j_flags &= ~JBD2_BARRIER; - write_unlock(&journal->j_state_lock); + ret = submit_bh(WRITE_SYNC_PLUG | WRITE_BARRIER, bh); + if (ret == -EOPNOTSUPP) { + printk(KERN_WARNING + "JBD2: Disabling barriers on %s, " + "not supported by device\n", journal->j_devname); + write_lock(&journal->j_state_lock); + journal->j_flags &= ~JBD2_BARRIER; + write_unlock(&journal->j_state_lock); - /* And try again, without the barrier */ - lock_buffer(bh); - set_buffer_uptodate(bh); - clear_buffer_dirty(bh); + /* And try again, without the barrier */ + lock_buffer(bh); + set_buffer_uptodate(bh); + clear_buffer_dirty(bh); + ret = submit_bh(WRITE_SYNC_PLUG, bh); + } + } else { ret = submit_bh(WRITE_SYNC_PLUG, bh); } *cbh = bh; -- cgit v1.2.3 From f73bee49855fe968e87af18e44f2d17e1e675e74 Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Wed, 18 Aug 2010 15:56:56 +0200 Subject: jbd2: Modify ASYNC_COMMIT code to not rely on queue draining on barrier Currently JBD2 relies blkdev_issue_flush() draining the queue when ASYNC_COMMIT feature is set. This property is going away so make JBD2 wait for buffers it needs on its own before submitting the cache flush. Signed-off-by: Jan Kara Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- fs/jbd2/commit.c | 29 ++++++++++++++++------------- 1 file changed, 16 insertions(+), 13 deletions(-) (limited to 'fs/jbd2/commit.c') diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index 7c068c189d80..d6aeb1f6cfe0 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -701,6 +701,16 @@ start_journal_io: } } + err = journal_finish_inode_data_buffers(journal, commit_transaction); + if (err) { + printk(KERN_WARNING + "JBD2: Detected IO errors while flushing file data " + "on %s\n", journal->j_devname); + if (journal->j_flags & JBD2_ABORT_ON_SYNCDATA_ERR) + jbd2_journal_abort(journal, err); + err = 0; + } + /* * If the journal is not located on the file system device, * then we must flush the file system device before we issue @@ -719,19 +729,6 @@ start_journal_io: &cbh, crc32_sum); if (err) __jbd2_journal_abort_hard(journal); - if (journal->j_flags & JBD2_BARRIER) - blkdev_issue_flush(journal->j_dev, GFP_KERNEL, NULL, - BLKDEV_IFL_WAIT); - } - - err = journal_finish_inode_data_buffers(journal, commit_transaction); - if (err) { - printk(KERN_WARNING - "JBD2: Detected IO errors while flushing file data " - "on %s\n", journal->j_devname); - if (journal->j_flags & JBD2_ABORT_ON_SYNCDATA_ERR) - jbd2_journal_abort(journal, err); - err = 0; } /* Lo and behold: we have just managed to send a transaction to @@ -845,6 +842,12 @@ wait_for_iobuf: } if (!err && !is_journal_aborted(journal)) err = journal_wait_on_commit_record(journal, cbh); + if (JBD2_HAS_INCOMPAT_FEATURE(journal, + JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT) && + journal->j_flags & JBD2_BARRIER) { + blkdev_issue_flush(journal->j_dev, GFP_KERNEL, NULL, + BLKDEV_IFL_WAIT); + } if (err) jbd2_journal_abort(journal, err); -- cgit v1.2.3 From 9c35575bbe6b1dd4914a5323c8df8b3159edcc75 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 18 Aug 2010 05:29:17 -0400 Subject: jbd2: replace barriers with explicit flush / FUA usage Switch to the WRITE_FLUSH_FUA flag for journal commits and remove the EOPNOTSUPP detection for barriers. Signed-off-by: Christoph Hellwig Acked-by: Jan Kara Signed-off-by: Tejun Heo Signed-off-by: Jens Axboe --- fs/jbd2/commit.c | 43 ++++--------------------------------------- 1 file changed, 4 insertions(+), 39 deletions(-) (limited to 'fs/jbd2/commit.c') diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index d6aeb1f6cfe0..f204e27f44d1 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -134,25 +134,11 @@ static int journal_submit_commit_record(journal_t *journal, if (journal->j_flags & JBD2_BARRIER && !JBD2_HAS_INCOMPAT_FEATURE(journal, - JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) { - ret = submit_bh(WRITE_SYNC_PLUG | WRITE_BARRIER, bh); - if (ret == -EOPNOTSUPP) { - printk(KERN_WARNING - "JBD2: Disabling barriers on %s, " - "not supported by device\n", journal->j_devname); - write_lock(&journal->j_state_lock); - journal->j_flags &= ~JBD2_BARRIER; - write_unlock(&journal->j_state_lock); - - /* And try again, without the barrier */ - lock_buffer(bh); - set_buffer_uptodate(bh); - clear_buffer_dirty(bh); - ret = submit_bh(WRITE_SYNC_PLUG, bh); - } - } else { + JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) + ret = submit_bh(WRITE_SYNC_PLUG | WRITE_FLUSH_FUA, bh); + else ret = submit_bh(WRITE_SYNC_PLUG, bh); - } + *cbh = bh; return ret; } @@ -166,29 +152,8 @@ static int journal_wait_on_commit_record(journal_t *journal, { int ret = 0; -retry: clear_buffer_dirty(bh); wait_on_buffer(bh); - if (buffer_eopnotsupp(bh) && (journal->j_flags & JBD2_BARRIER)) { - printk(KERN_WARNING - "JBD2: %s: disabling barries on %s - not supported " - "by device\n", __func__, journal->j_devname); - write_lock(&journal->j_state_lock); - journal->j_flags &= ~JBD2_BARRIER; - write_unlock(&journal->j_state_lock); - - lock_buffer(bh); - clear_buffer_dirty(bh); - set_buffer_uptodate(bh); - bh->b_end_io = journal_end_buffer_io_sync; - - ret = submit_bh(WRITE_SYNC_PLUG, bh); - if (ret) { - unlock_buffer(bh); - return ret; - } - goto retry; - } if (unlikely(!buffer_uptodate(bh))) ret = -EIO; -- cgit v1.2.3 From dd3932eddf428571762596e17b65f5dc92ca361b Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Thu, 16 Sep 2010 20:51:46 +0200 Subject: block: remove BLKDEV_IFL_WAIT All the blkdev_issue_* helpers can only sanely be used for synchronous caller. To issue cache flushes or barriers asynchronously the caller needs to set up a bio by itself with a completion callback to move the asynchronous state machine ahead. So drop the BLKDEV_IFL_WAIT flag that is always specified when calling blkdev_issue_* and also remove the now unused flags argument to blkdev_issue_flush and blkdev_issue_zeroout. For blkdev_issue_discard we need to keep it for the secure discard flag, which gains a more descriptive name and loses the bitops vs flag confusion. Signed-off-by: Christoph Hellwig Signed-off-by: Jens Axboe --- fs/jbd2/commit.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'fs/jbd2/commit.c') diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index f204e27f44d1..cb43c605cfaa 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -684,8 +684,7 @@ start_journal_io: if (commit_transaction->t_flushed_data_blocks && (journal->j_fs_dev != journal->j_dev) && (journal->j_flags & JBD2_BARRIER)) - blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL, NULL, - BLKDEV_IFL_WAIT); + blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL, NULL); /* Done it all: now write the commit record asynchronously. */ if (JBD2_HAS_INCOMPAT_FEATURE(journal, @@ -810,8 +809,7 @@ wait_for_iobuf: if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT) && journal->j_flags & JBD2_BARRIER) { - blkdev_issue_flush(journal->j_dev, GFP_KERNEL, NULL, - BLKDEV_IFL_WAIT); + blkdev_issue_flush(journal->j_dev, GFP_KERNEL, NULL); } if (err) -- cgit v1.2.3 From 749ef9f8423054e326f3a246327ed2db4b6d395f Mon Sep 17 00:00:00 2001 From: Corrado Zoccolo Date: Mon, 20 Sep 2010 15:24:50 +0200 Subject: cfq: improve fsync performance for small files Fsync performance for small files achieved by cfq on high-end disks is lower than what deadline can achieve, due to idling introduced between the sync write happening in process context and the journal commit. Moreover, when competing with a sequential reader, a process writing small files and fsync-ing them is starved. This patch fixes the two problems by: - marking journal commits as WRITE_SYNC, so that they get the REQ_NOIDLE flag set, - force all queues that have REQ_NOIDLE requests to be put in the noidle tree. Having the queue associated to the fsync-ing process and the one associated to journal commits in the noidle tree allows: - switching between them without idling, - fairness vs. competing idling queues, since they will be serviced only after the noidle tree expires its slice. Acked-by: Vivek Goyal Reviewed-by: Jeff Moyer Tested-by: Jeff Moyer Signed-off-by: Corrado Zoccolo Signed-off-by: Jens Axboe --- fs/jbd2/commit.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs/jbd2/commit.c') diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index 7c068c189d80..80910f51d4b4 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -360,7 +360,7 @@ void jbd2_journal_commit_transaction(journal_t *journal) int tag_bytes = journal_tag_bytes(journal); struct buffer_head *cbh = NULL; /* For transactional checksums */ __u32 crc32_sum = ~0; - int write_op = WRITE; + int write_op = WRITE_SYNC; /* * First job: lock down the current transaction and wait for -- cgit v1.2.3 From 39e3ac2599a5f9aba499b5f8af809108e70a6163 Mon Sep 17 00:00:00 2001 From: Brian King Date: Wed, 27 Oct 2010 21:25:12 -0400 Subject: jbd2: Fix I/O hang in jbd2_journal_release_jbd_inode This fixes a hang seen in jbd2_journal_release_jbd_inode on a lot of Power 6 systems running with ext4. When we get in the hung state, all I/O to the disk in question gets blocked where we stay indefinitely. Looking at the task list, I can see we are stuck in jbd2_journal_release_jbd_inode waiting on a wake up. I added some debug code to detect this scenario and dump additional data if we were stuck in jbd2_journal_release_jbd_inode for longer than 30 minutes. When it hit, I was able to see that i_flags was 0, suggesting we missed the wake up. This patch changes i_flags to be an unsigned long, uses bit operators to access it, and adds barriers around the accesses. Prior to applying this patch, we were regularly hitting this hang on numerous systems in our test environment. After applying the patch, the hangs no longer occur. Signed-off-by: Brian King Signed-off-by: "Theodore Ts'o" --- fs/jbd2/commit.c | 12 ++++++++---- 1 file changed, 8 insertions(+), 4 deletions(-) (limited to 'fs/jbd2/commit.c') diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c index 7c068c189d80..6494c81e3b0a 100644 --- a/fs/jbd2/commit.c +++ b/fs/jbd2/commit.c @@ -26,7 +26,9 @@ #include #include #include +#include #include +#include /* * Default IO end handler for temporary BJ_IO buffer_heads. @@ -236,7 +238,7 @@ static int journal_submit_data_buffers(journal_t *journal, spin_lock(&journal->j_list_lock); list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) { mapping = jinode->i_vfs_inode->i_mapping; - jinode->i_flags |= JI_COMMIT_RUNNING; + set_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); spin_unlock(&journal->j_list_lock); /* * submit the inode data buffers. We use writepage @@ -251,7 +253,8 @@ static int journal_submit_data_buffers(journal_t *journal, spin_lock(&journal->j_list_lock); J_ASSERT(jinode->i_transaction == commit_transaction); commit_transaction->t_flushed_data_blocks = 1; - jinode->i_flags &= ~JI_COMMIT_RUNNING; + clear_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); + smp_mb__after_clear_bit(); wake_up_bit(&jinode->i_flags, __JI_COMMIT_RUNNING); } spin_unlock(&journal->j_list_lock); @@ -272,7 +275,7 @@ static int journal_finish_inode_data_buffers(journal_t *journal, /* For locking, see the comment in journal_submit_data_buffers() */ spin_lock(&journal->j_list_lock); list_for_each_entry(jinode, &commit_transaction->t_inode_list, i_list) { - jinode->i_flags |= JI_COMMIT_RUNNING; + set_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); spin_unlock(&journal->j_list_lock); err = filemap_fdatawait(jinode->i_vfs_inode->i_mapping); if (err) { @@ -288,7 +291,8 @@ static int journal_finish_inode_data_buffers(journal_t *journal, ret = err; } spin_lock(&journal->j_list_lock); - jinode->i_flags &= ~JI_COMMIT_RUNNING; + clear_bit(__JI_COMMIT_RUNNING, &jinode->i_flags); + smp_mb__after_clear_bit(); wake_up_bit(&jinode->i_flags, __JI_COMMIT_RUNNING); } -- cgit v1.2.3