struct inode *inode;
struct bio *bio;
struct list_head list;
- extent_submit_bio_hook_t *submit_bio_hook;
+ extent_submit_bio_hook_t *submit_bio_start;
+ extent_submit_bio_hook_t *submit_bio_done;
int rw;
int mirror_num;
unsigned long bio_flags;
btrfs_async_submit_limit(info);
}
-static void run_one_async_submit(struct btrfs_work *work)
+static void run_one_async_start(struct btrfs_work *work)
+{
+ struct btrfs_fs_info *fs_info;
+ struct async_submit_bio *async;
+
+ async = container_of(work, struct async_submit_bio, work);
+ fs_info = BTRFS_I(async->inode)->root->fs_info;
+ async->submit_bio_start(async->inode, async->rw, async->bio,
+ async->mirror_num, async->bio_flags);
+}
+
+static void run_one_async_done(struct btrfs_work *work)
{
struct btrfs_fs_info *fs_info;
struct async_submit_bio *async;
waitqueue_active(&fs_info->async_submit_wait))
wake_up(&fs_info->async_submit_wait);
- async->submit_bio_hook(async->inode, async->rw, async->bio,
+ async->submit_bio_done(async->inode, async->rw, async->bio,
async->mirror_num, async->bio_flags);
+}
+
+static void run_one_async_free(struct btrfs_work *work)
+{
+ struct async_submit_bio *async;
+
+ async = container_of(work, struct async_submit_bio, work);
kfree(async);
}
int btrfs_wq_submit_bio(struct btrfs_fs_info *fs_info, struct inode *inode,
int rw, struct bio *bio, int mirror_num,
unsigned long bio_flags,
- extent_submit_bio_hook_t *submit_bio_hook)
+ extent_submit_bio_hook_t *submit_bio_start,
+ extent_submit_bio_hook_t *submit_bio_done)
{
struct async_submit_bio *async;
int limit = btrfs_async_submit_limit(fs_info);
async->rw = rw;
async->bio = bio;
async->mirror_num = mirror_num;
- async->submit_bio_hook = submit_bio_hook;
- async->work.func = run_one_async_submit;
+ async->submit_bio_start = submit_bio_start;
+ async->submit_bio_done = submit_bio_done;
+
+ async->work.func = run_one_async_start;
+ async->work.ordered_func = run_one_async_done;
+ async->work.ordered_free = run_one_async_free;
+
async->work.flags = 0;
async->bio_flags = bio_flags;
(atomic_read(&fs_info->nr_async_bios) < limit),
HZ/10);
}
+
+ while(atomic_read(&fs_info->async_submit_draining) &&
+ atomic_read(&fs_info->nr_async_submits)) {
+ wait_event(fs_info->async_submit_wait,
+ (atomic_read(&fs_info->nr_async_submits) == 0));
+ }
+
return 0;
}
return 0;
}
-static int __btree_submit_bio_hook(struct inode *inode, int rw, struct bio *bio,
- int mirror_num, unsigned long bio_flags)
+static int __btree_submit_bio_start(struct inode *inode, int rw,
+ struct bio *bio, int mirror_num,
+ unsigned long bio_flags)
{
- struct btrfs_root *root = BTRFS_I(inode)->root;
- int ret;
-
/*
* when we're called for a write, we're already in the async
* submission context. Just jump into btrfs_map_bio
*/
- if (rw & (1 << BIO_RW)) {
- btree_csum_one_bio(bio);
- return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
- mirror_num, 1);
- }
+ btree_csum_one_bio(bio);
+ return 0;
+}
+static int __btree_submit_bio_done(struct inode *inode, int rw, struct bio *bio,
+ int mirror_num, unsigned long bio_flags)
+{
/*
- * called for a read, do the setup so that checksum validation
- * can happen in the async kernel threads
+ * when we're called for a write, we're already in the async
+ * submission context. Just jump into btrfs_map_bio
*/
- ret = btrfs_bio_wq_end_io(root->fs_info, bio, 1);
- BUG_ON(ret);
-
return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio, mirror_num, 1);
}
* can happen in parallel across all CPUs
*/
if (!(rw & (1 << BIO_RW))) {
- return __btree_submit_bio_hook(inode, rw, bio, mirror_num, 0);
+ int ret;
+ /*
+ * called for a read, do the setup so that checksum validation
+ * can happen in the async kernel threads
+ */
+ ret = btrfs_bio_wq_end_io(BTRFS_I(inode)->root->fs_info,
+ bio, 1);
+ BUG_ON(ret);
+
+ return btrfs_map_bio(BTRFS_I(inode)->root, rw, bio,
+ mirror_num, 1);
}
return btrfs_wq_submit_bio(BTRFS_I(inode)->root->fs_info,
inode, rw, bio, mirror_num, 0,
- __btree_submit_bio_hook);
+ __btree_submit_bio_start,
+ __btree_submit_bio_done);
}
static int btree_writepage(struct page *page, struct writeback_control *wbc)
{
int ret;
u32 blocksize;
+ u64 generation;
__setup_root(tree_root->nodesize, tree_root->leafsize,
tree_root->sectorsize, tree_root->stripesize,
&root->root_item, &root->root_key);
BUG_ON(ret);
+ generation = btrfs_root_generation(&root->root_item);
blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
- blocksize, 0);
+ blocksize, generation);
BUG_ON(!root->node);
return 0;
}
struct btrfs_path *path;
struct extent_buffer *l;
u64 highest_inode;
+ u64 generation;
u32 blocksize;
int ret = 0;
kfree(root);
return ERR_PTR(ret);
}
+ generation = btrfs_root_generation(&root->root_item);
blocksize = btrfs_level_size(root, btrfs_root_level(&root->root_item));
root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
- blocksize, 0);
+ blocksize, generation);
BUG_ON(!root->node);
insert:
if (location->objectid != BTRFS_TREE_LOG_OBJECTID) {
return;
inode = mapping->host;
+
+ /*
+ * don't do the expensive searching for a small number of
+ * devices
+ */
+ if (BTRFS_I(inode)->root->fs_info->fs_devices->open_devices <= 2) {
+ __unplug_io_fn(bdi, page);
+ return;
+ }
+
offset = page_offset(page);
em_tree = &BTRFS_I(inode)->extent_tree;
u32 leafsize;
u32 blocksize;
u32 stripesize;
+ u64 generation;
struct buffer_head *bh;
struct btrfs_root *extent_root = kzalloc(sizeof(struct btrfs_root),
GFP_NOFS);
INIT_LIST_HEAD(&fs_info->space_info);
btrfs_mapping_init(&fs_info->mapping_tree);
atomic_set(&fs_info->nr_async_submits, 0);
+ atomic_set(&fs_info->async_delalloc_pages, 0);
atomic_set(&fs_info->async_submit_draining, 0);
atomic_set(&fs_info->nr_async_bios, 0);
atomic_set(&fs_info->throttles, 0);
mutex_init(&fs_info->trans_mutex);
mutex_init(&fs_info->tree_log_mutex);
mutex_init(&fs_info->drop_mutex);
- mutex_init(&fs_info->alloc_mutex);
+ mutex_init(&fs_info->extent_ins_mutex);
+ mutex_init(&fs_info->pinned_mutex);
mutex_init(&fs_info->chunk_mutex);
mutex_init(&fs_info->transaction_kthread_mutex);
mutex_init(&fs_info->cleaner_mutex);
btrfs_init_workers(&fs_info->workers, "worker",
fs_info->thread_pool_size);
+ btrfs_init_workers(&fs_info->delalloc_workers, "delalloc",
+ fs_info->thread_pool_size);
+
btrfs_init_workers(&fs_info->submit_workers, "submit",
min_t(u64, fs_devices->num_devices,
fs_info->thread_pool_size));
*/
fs_info->submit_workers.idle_thresh = 64;
- /* fs_info->workers is responsible for checksumming file data
- * blocks and metadata. Using a larger idle thresh allows each
- * worker thread to operate on things in roughly the order they
- * were sent by the writeback daemons, improving overall locality
- * of the IO going down the pipe.
- */
- fs_info->workers.idle_thresh = 128;
+ fs_info->workers.idle_thresh = 16;
+ fs_info->workers.ordered = 1;
+
+ fs_info->delalloc_workers.idle_thresh = 2;
+ fs_info->delalloc_workers.ordered = 1;
btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1);
btrfs_init_workers(&fs_info->endio_workers, "endio",
btrfs_start_workers(&fs_info->workers, 1);
btrfs_start_workers(&fs_info->submit_workers, 1);
+ btrfs_start_workers(&fs_info->delalloc_workers, 1);
btrfs_start_workers(&fs_info->fixup_workers, 1);
btrfs_start_workers(&fs_info->endio_workers, fs_info->thread_pool_size);
btrfs_start_workers(&fs_info->endio_write_workers,
blocksize = btrfs_level_size(tree_root,
btrfs_super_chunk_root_level(disk_super));
+ generation = btrfs_super_chunk_root_generation(disk_super);
__setup_root(nodesize, leafsize, sectorsize, stripesize,
chunk_root, fs_info, BTRFS_CHUNK_TREE_OBJECTID);
chunk_root->node = read_tree_block(chunk_root,
btrfs_super_chunk_root(disk_super),
- blocksize, 0);
+ blocksize, generation);
BUG_ON(!chunk_root->node);
read_extent_buffer(chunk_root->node, fs_info->chunk_tree_uuid,
blocksize = btrfs_level_size(tree_root,
btrfs_super_root_level(disk_super));
-
+ generation = btrfs_super_generation(disk_super);
tree_root->node = read_tree_block(tree_root,
btrfs_super_root(disk_super),
- blocksize, 0);
+ blocksize, generation);
if (!tree_root->node)
goto fail_sb_buffer;
log_tree_root, fs_info, BTRFS_TREE_LOG_OBJECTID);
log_tree_root->node = read_tree_block(tree_root, bytenr,
- blocksize, 0);
+ blocksize,
+ generation + 1);
ret = btrfs_recover_log_trees(log_tree_root);
BUG_ON(ret);
}
+ fs_info->last_trans_committed = btrfs_super_generation(disk_super);
ret = btrfs_cleanup_reloc_trees(tree_root);
BUG_ON(ret);
- fs_info->last_trans_committed = btrfs_super_generation(disk_super);
return tree_root;
fail_cleaner:
fail_sys_array:
fail_sb_buffer:
btrfs_stop_workers(&fs_info->fixup_workers);
+ btrfs_stop_workers(&fs_info->delalloc_workers);
btrfs_stop_workers(&fs_info->workers);
btrfs_stop_workers(&fs_info->endio_workers);
btrfs_stop_workers(&fs_info->endio_write_workers);
truncate_inode_pages(fs_info->btree_inode->i_mapping, 0);
btrfs_stop_workers(&fs_info->fixup_workers);
+ btrfs_stop_workers(&fs_info->delalloc_workers);
btrfs_stop_workers(&fs_info->workers);
btrfs_stop_workers(&fs_info->endio_workers);
btrfs_stop_workers(&fs_info->endio_write_workers);
struct extent_io_tree *tree;
u64 num_dirty;
u64 start = 0;
- unsigned long thresh = 96 * 1024 * 1024;
+ unsigned long thresh = 32 * 1024 * 1024;
tree = &BTRFS_I(root->fs_info->btree_inode)->io_tree;
if (current_is_pdflush() || current->flags & PF_MEMALLOC)