Btrfs: Add ordered async work queues
Btrfs uses kernel threads to create async work queues for cpu intensive
operations such as checksumming and decompression. These work well,
but they make it difficult to keep IO order intact.
A single writepages call from pdflush or fsync will turn into a number
of bios, and each bio is checksummed in parallel. Once the checksum is
computed, the bio is sent down to the disk, and since we don't control
the order in which the parallel operations happen, they might go down to
the disk in almost any order.
The code deals with this somewhat by having deep work queues for a single
kernel thread, making it very likely that a single thread will process all
the bios for a single inode.
This patch introduces an explicitly ordered work queue. As work structs
are placed into the queue they are put onto the tail of a list. They have
three callbacks:
->func (cpu intensive processing here)
->ordered_func (order sensitive processing here)
->ordered_free (free the work struct, all processing is done)
The work struct has three callbacks. The func callback does the cpu intensive
work, and when it completes the work struct is marked as done.
Every time a work struct completes, the list is checked to see if the head
is marked as done. If so the ordered_func callback is used to do the
order sensitive processing and the ordered_free callback is used to do
any cleanup. Then we loop back and check the head of the list again.
This patch also changes the checksumming code to use the ordered workqueues.
One a 4 drive array, it increases streaming writes from 280MB/s to 350MB/s.
Signed-off-by: Chris Mason <chris.mason@oracle.com>
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 94b4e50..e0a28f7 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -80,7 +80,8 @@
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;
@@ -452,7 +453,18 @@
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;
@@ -470,15 +482,23 @@
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);
@@ -491,8 +511,13 @@
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;
@@ -533,29 +558,25 @@
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);
}
@@ -567,11 +588,22 @@
* 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)
@@ -1534,7 +1566,8 @@
* 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 = 8;
+ fs_info->workers.ordered = 1;
btrfs_init_workers(&fs_info->fixup_workers, "fixup", 1);
btrfs_init_workers(&fs_info->endio_workers, "endio",