Merge git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable (ccc5ff94) · Commits · 戴 / test

fs/btrfs/async-thread.c

+47 −13

Original line number	Diff line number	Diff line
		@@ -25,6 +25,7 @@
		#define WORK_QUEUED_BIT 0
		#define WORK_DONE_BIT 1
		#define WORK_ORDER_DONE_BIT 2
		#define WORK_HIGH_PRIO_BIT 3

		/*
		* container for the kthread task pointer and the list of pending work
		@@ -36,6 +37,7 @@ struct btrfs_worker_thread {

		/* list of struct btrfs_work that are waiting for service */
		struct list_head pending;
		struct list_head prio_pending;

		/* list of worker threads from struct btrfs_workers */
		struct list_head worker_list;
		@@ -103,10 +105,16 @@ static noinline int run_ordered_completions(struct btrfs_workers *workers,

		spin_lock_irqsave(&workers->lock, flags);

		while (!list_empty(&workers->order_list)) {
		while (1) {
		if (!list_empty(&workers->prio_order_list)) {
		work = list_entry(workers->prio_order_list.next,
		struct btrfs_work, order_list);
		} else if (!list_empty(&workers->order_list)) {
		work = list_entry(workers->order_list.next,
		struct btrfs_work, order_list);

		} else {
		break;
		}
		if (!test_bit(WORK_DONE_BIT, &work->flags))
		break;

		@@ -143,8 +151,14 @@ static int worker_loop(void *arg)
		do {
		spin_lock_irq(&worker->lock);
		again_locked:
		while (!list_empty(&worker->pending)) {
		while (1) {
		if (!list_empty(&worker->prio_pending))
		cur = worker->prio_pending.next;
		else if (!list_empty(&worker->pending))
		cur = worker->pending.next;
		else
		break;

		work = list_entry(cur, struct btrfs_work, list);
		list_del(&work->list);
		clear_bit(WORK_QUEUED_BIT, &work->flags);
		@@ -163,7 +177,6 @@ again_locked:

		spin_lock_irq(&worker->lock);
		check_idle_worker(worker);

		}
		if (freezing(current)) {
		worker->working = 0;
		@@ -178,7 +191,8 @@ again_locked:
		* jump_in?
		*/
		smp_mb();
		if (!list_empty(&worker->pending))
		if (!list_empty(&worker->pending) \|\|
		!list_empty(&worker->prio_pending))
		continue;

		/*
		@@ -191,7 +205,8 @@ again_locked:
		*/
		schedule_timeout(1);
		smp_mb();
		if (!list_empty(&worker->pending))
		if (!list_empty(&worker->pending) \|\|
		!list_empty(&worker->prio_pending))
		continue;

		if (kthread_should_stop())
		@@ -200,7 +215,8 @@ again_locked:
		/* still no more work?, sleep for real */
		spin_lock_irq(&worker->lock);
		set_current_state(TASK_INTERRUPTIBLE);
		if (!list_empty(&worker->pending))
		if (!list_empty(&worker->pending) \|\|
		!list_empty(&worker->prio_pending))
		goto again_locked;

		/*
		@@ -248,6 +264,7 @@ void btrfs_init_workers(struct btrfs_workers workers, char name, int max)
		INIT_LIST_HEAD(&workers->worker_list);
		INIT_LIST_HEAD(&workers->idle_list);
		INIT_LIST_HEAD(&workers->order_list);
		INIT_LIST_HEAD(&workers->prio_order_list);
		spin_lock_init(&workers->lock);
		workers->max_workers = max;
		workers->idle_thresh = 32;
		@@ -273,6 +290,7 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers)
		}

		INIT_LIST_HEAD(&worker->pending);
		INIT_LIST_HEAD(&worker->prio_pending);
		INIT_LIST_HEAD(&worker->worker_list);
		spin_lock_init(&worker->lock);
		atomic_set(&worker->num_pending, 0);
		@@ -396,6 +414,9 @@ int btrfs_requeue_work(struct btrfs_work *work)
		goto out;

		spin_lock_irqsave(&worker->lock, flags);
		if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
		list_add_tail(&work->list, &worker->prio_pending);
		else
		list_add_tail(&work->list, &worker->pending);
		atomic_inc(&worker->num_pending);

		@@ -422,6 +443,11 @@ out:
		return 0;
		}

		void btrfs_set_work_high_prio(struct btrfs_work *work)
		{
		set_bit(WORK_HIGH_PRIO_BIT, &work->flags);
		}

		/*
		* places a struct btrfs_work into the pending queue of one of the kthreads
		*/
		@@ -438,7 +464,12 @@ int btrfs_queue_worker(struct btrfs_workers workers, struct btrfs_work work)
		worker = find_worker(workers);
		if (workers->ordered) {
		spin_lock_irqsave(&workers->lock, flags);
		if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags)) {
		list_add_tail(&work->order_list,
		&workers->prio_order_list);
		} else {
		list_add_tail(&work->order_list, &workers->order_list);
		}
		spin_unlock_irqrestore(&workers->lock, flags);
		} else {
		INIT_LIST_HEAD(&work->order_list);
		@@ -446,6 +477,9 @@ int btrfs_queue_worker(struct btrfs_workers workers, struct btrfs_work work)

		spin_lock_irqsave(&worker->lock, flags);

		if (test_bit(WORK_HIGH_PRIO_BIT, &work->flags))
		list_add_tail(&work->list, &worker->prio_pending);
		else
		list_add_tail(&work->list, &worker->pending);
		atomic_inc(&worker->num_pending);
		check_busy_worker(worker);

fs/btrfs/async-thread.h

+2 −0

Original line number	Diff line number	Diff line
		@@ -85,6 +85,7 @@ struct btrfs_workers {
		* of work items waiting for completion
		*/
		struct list_head order_list;
		struct list_head prio_order_list;

		/* lock for finding the next worker thread to queue on */
		spinlock_t lock;
		@@ -98,4 +99,5 @@ int btrfs_start_workers(struct btrfs_workers *workers, int num_workers);
		int btrfs_stop_workers(struct btrfs_workers *workers);
		void btrfs_init_workers(struct btrfs_workers workers, char name, int max);
		int btrfs_requeue_work(struct btrfs_work *work);
		void btrfs_set_work_high_prio(struct btrfs_work *work);
		#endif

fs/btrfs/ctree.c

+12 −5

Original line number	Diff line number	Diff line
		@@ -1325,12 +1325,12 @@ static noinline int reada_for_balance(struct btrfs_root *root,
		int ret = 0;
		int blocksize;

		parent = path->nodes[level - 1];
		parent = path->nodes[level + 1];
		if (!parent)
		return 0;

		nritems = btrfs_header_nritems(parent);
		slot = path->slots[level];
		slot = path->slots[level + 1];
		blocksize = btrfs_level_size(root, level);

		if (slot > 0) {
		@@ -1341,7 +1341,7 @@ static noinline int reada_for_balance(struct btrfs_root *root,
		block1 = 0;
		free_extent_buffer(eb);
		}
		if (slot < nritems) {
		if (slot + 1 < nritems) {
		block2 = btrfs_node_blockptr(parent, slot + 1);
		gen = btrfs_node_ptr_generation(parent, slot + 1);
		eb = btrfs_find_tree_block(root, block2, blocksize);
		@@ -1351,7 +1351,11 @@ static noinline int reada_for_balance(struct btrfs_root *root,
		}
		if (block1 \|\| block2) {
		ret = -EAGAIN;

		/* release the whole path */
		btrfs_release_path(root, path);

		/* read the blocks */
		if (block1)
		readahead_tree_block(root, block1, blocksize, 0);
		if (block2)
		@@ -1361,7 +1365,7 @@ static noinline int reada_for_balance(struct btrfs_root *root,
		eb = read_tree_block(root, block1, blocksize, 0);
		free_extent_buffer(eb);
		}
		if (block1) {
		if (block2) {
		eb = read_tree_block(root, block2, blocksize, 0);
		free_extent_buffer(eb);
		}
		@@ -1481,12 +1485,15 @@ read_block_for_search(struct btrfs_trans_handle *trans,
		* of the btree by dropping locks before
		* we read.
		*/
		btrfs_release_path(NULL, p);
		btrfs_unlock_up_safe(p, level + 1);
		btrfs_set_path_blocking(p);

		if (tmp)
		free_extent_buffer(tmp);
		if (p->reada)
		reada_for_search(root, p, level, slot, key->objectid);

		btrfs_release_path(NULL, p);
		tmp = read_tree_block(root, blocknr, blocksize, gen);
		if (tmp)
		free_extent_buffer(tmp);

fs/btrfs/disk-io.c

+7 −2

Original line number	Diff line number	Diff line
		@@ -579,6 +579,10 @@ int btrfs_wq_submit_bio(struct btrfs_fs_info fs_info, struct inode inode,
		async->bio_flags = bio_flags;

		atomic_inc(&fs_info->nr_async_submits);

		if (rw & (1 << BIO_RW_SYNCIO))
		btrfs_set_work_high_prio(&async->work);

		btrfs_queue_worker(&fs_info->workers, &async->work);
		#if 0
		int limit = btrfs_async_submit_limit(fs_info);
		@@ -656,6 +660,7 @@ static int btree_submit_bio_hook(struct inode inode, int rw, struct bio bio,
		return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
		mirror_num, 0);
		}

		/*
		* kthread helpers are used to submit writes so that checksumming
		* can happen in parallel across all CPUs
		@@ -2095,10 +2100,10 @@ static int write_dev_supers(struct btrfs_device *device,
		device->barriers = 0;
		get_bh(bh);
		lock_buffer(bh);
		ret = submit_bh(WRITE, bh);
		ret = submit_bh(WRITE_SYNC, bh);
		}
		} else {
		ret = submit_bh(WRITE, bh);
		ret = submit_bh(WRITE_SYNC, bh);
		}

		if (!ret && wait) {

fs/btrfs/extent_io.c

+63 −23

Original line number	Diff line number	Diff line
		@@ -50,7 +50,10 @@ struct extent_page_data {
		/* tells writepage not to lock the state bits for this range
		* it still does the unlocking
		*/
		int extent_locked;
		unsigned int extent_locked:1;

		/* tells the submit_bio code to use a WRITE_SYNC */
		unsigned int sync_io:1;
		};

		int __init extent_io_init(void)
		@@ -2101,6 +2104,16 @@ int extent_read_full_page(struct extent_io_tree tree, struct page page,
		return ret;
		}

		static noinline void update_nr_written(struct page *page,
		struct writeback_control *wbc,
		unsigned long nr_written)
		{
		wbc->nr_to_write -= nr_written;
		if (wbc->range_cyclic \|\| (wbc->nr_to_write > 0 &&
		wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
		page->mapping->writeback_index = page->index + nr_written;
		}

		/*
		* the writepage semantics are similar to regular writepage. extent
		* records are inserted to lock ranges in the tree, and as dirty areas
		@@ -2136,8 +2149,14 @@ static int __extent_writepage(struct page page, struct writeback_control wbc,
		u64 delalloc_end;
		int page_started;
		int compressed;
		int write_flags;
		unsigned long nr_written = 0;

		if (wbc->sync_mode == WB_SYNC_ALL)
		write_flags = WRITE_SYNC_PLUG;
		else
		write_flags = WRITE;

		WARN_ON(!PageLocked(page));
		pg_offset = i_size & (PAGE_CACHE_SIZE - 1);
		if (page->index > end_index \|\|
		@@ -2164,6 +2183,12 @@ static int __extent_writepage(struct page page, struct writeback_control wbc,
		delalloc_end = 0;
		page_started = 0;
		if (!epd->extent_locked) {
		/*
		* make sure the wbc mapping index is at least updated
		* to this page.
		*/
		update_nr_written(page, wbc, 0);

		while (delalloc_end < page_end) {
		nr_delalloc = find_lock_delalloc_range(inode, tree,
		page,
		@@ -2185,7 +2210,13 @@ static int __extent_writepage(struct page page, struct writeback_control wbc,
		*/
		if (page_started) {
		ret = 0;
		goto update_nr_written;
		/*
		* we've unlocked the page, so we can't update
		* the mapping's writeback index, just update
		* nr_to_write.
		*/
		wbc->nr_to_write -= nr_written;
		goto done_unlocked;
		}
		}
		lock_extent(tree, start, page_end, GFP_NOFS);
		@@ -2198,13 +2229,18 @@ static int __extent_writepage(struct page page, struct writeback_control wbc,
		if (ret == -EAGAIN) {
		unlock_extent(tree, start, page_end, GFP_NOFS);
		redirty_page_for_writepage(wbc, page);
		update_nr_written(page, wbc, nr_written);
		unlock_page(page);
		ret = 0;
		goto update_nr_written;
		goto done_unlocked;
		}
		}

		nr_written++;
		/*
		* we don't want to touch the inode after unlocking the page,
		* so we update the mapping writeback index now
		*/
		update_nr_written(page, wbc, nr_written + 1);

		end = page_end;
		if (test_range_bit(tree, start, page_end, EXTENT_DELALLOC, 0))
		@@ -2314,9 +2350,9 @@ static int __extent_writepage(struct page page, struct writeback_control wbc,
		(unsigned long long)end);
		}

		ret = submit_extent_page(WRITE, tree, page, sector,
		iosize, pg_offset, bdev,
		&epd->bio, max_nr,
		ret = submit_extent_page(write_flags, tree, page,
		sector, iosize, pg_offset,
		bdev, &epd->bio, max_nr,
		end_bio_extent_writepage,
		0, 0, 0);
		if (ret)
		@@ -2336,11 +2372,8 @@ done:
		unlock_extent(tree, unlock_start, page_end, GFP_NOFS);
		unlock_page(page);

		update_nr_written:
		wbc->nr_to_write -= nr_written;
		if (wbc->range_cyclic \|\| (wbc->nr_to_write > 0 &&
		wbc->range_start == 0 && wbc->range_end == LLONG_MAX))
		page->mapping->writeback_index = page->index + nr_written;
		done_unlocked:

		return 0;
		}

		@@ -2460,15 +2493,23 @@ retry:
		return ret;
		}

		static noinline void flush_write_bio(void *data)
		static void flush_epd_write_bio(struct extent_page_data *epd)
		{
		struct extent_page_data *epd = data;
		if (epd->bio) {
		if (epd->sync_io)
		submit_one_bio(WRITE_SYNC, epd->bio, 0, 0);
		else
		submit_one_bio(WRITE, epd->bio, 0, 0);
		epd->bio = NULL;
		}
		}

		static noinline void flush_write_bio(void *data)
		{
		struct extent_page_data *epd = data;
		flush_epd_write_bio(epd);
		}

		int extent_write_full_page(struct extent_io_tree tree, struct page page,
		get_extent_t *get_extent,
		struct writeback_control *wbc)
		@@ -2480,23 +2521,22 @@ int extent_write_full_page(struct extent_io_tree tree, struct page page,
		.tree = tree,
		.get_extent = get_extent,
		.extent_locked = 0,
		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
		};
		struct writeback_control wbc_writepages = {
		.bdi = wbc->bdi,
		.sync_mode = WB_SYNC_NONE,
		.sync_mode = wbc->sync_mode,
		.older_than_this = NULL,
		.nr_to_write = 64,
		.range_start = page_offset(page) + PAGE_CACHE_SIZE,
		.range_end = (loff_t)-1,
		};


		ret = __extent_writepage(page, wbc, &epd);

		extent_write_cache_pages(tree, mapping, &wbc_writepages,
		__extent_writepage, &epd, flush_write_bio);
		if (epd.bio)
		submit_one_bio(WRITE, epd.bio, 0, 0);
		flush_epd_write_bio(&epd);
		return ret;
		}

		@@ -2515,6 +2555,7 @@ int extent_write_locked_range(struct extent_io_tree tree, struct inode inode,
		.tree = tree,
		.get_extent = get_extent,
		.extent_locked = 1,
		.sync_io = mode == WB_SYNC_ALL,
		};
		struct writeback_control wbc_writepages = {
		.bdi = inode->i_mapping->backing_dev_info,
		@@ -2540,8 +2581,7 @@ int extent_write_locked_range(struct extent_io_tree tree, struct inode inode,
		start += PAGE_CACHE_SIZE;
		}

		if (epd.bio)
		submit_one_bio(WRITE, epd.bio, 0, 0);
		flush_epd_write_bio(&epd);
		return ret;
		}

		@@ -2556,13 +2596,13 @@ int extent_writepages(struct extent_io_tree *tree,
		.tree = tree,
		.get_extent = get_extent,
		.extent_locked = 0,
		.sync_io = wbc->sync_mode == WB_SYNC_ALL,
		};

		ret = extent_write_cache_pages(tree, mapping, wbc,
		__extent_writepage, &epd,
		flush_write_bio);
		if (epd.bio)
		submit_one_bio(WRITE, epd.bio, 0, 0);
		flush_epd_write_bio(&epd);
		return ret;
		}

Admin message