struct btrfs_root *root;
};
+struct btrfs_dio_data {
+ u64 outstanding_extents;
+ u64 reserve;
+ u64 unsubmitted_oe_range_start;
+ u64 unsubmitted_oe_range_end;
+};
+
static const struct inode_operations btrfs_dir_inode_operations;
static const struct inode_operations btrfs_symlink_inode_operations;
static const struct inode_operations btrfs_dir_ro_inode_operations;
static const struct address_space_operations btrfs_aops;
static const struct address_space_operations btrfs_symlink_aops;
static const struct file_operations btrfs_dir_file_operations;
-static struct extent_io_ops btrfs_extent_io_ops;
+static const struct extent_io_ops btrfs_extent_io_ops;
static struct kmem_cache *btrfs_inode_cachep;
-static struct kmem_cache *btrfs_delalloc_work_cachep;
struct kmem_cache *btrfs_trans_handle_cachep;
struct kmem_cache *btrfs_transaction_cachep;
struct kmem_cache *btrfs_path_cachep;
struct kmem_cache *btrfs_free_space_cachep;
#define S_SHIFT 12
-static unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = {
+static const unsigned char btrfs_type_by_mode[S_IFMT >> S_SHIFT] = {
[S_IFREG >> S_SHIFT] = BTRFS_FT_REG_FILE,
[S_IFDIR >> S_SHIFT] = BTRFS_FT_DIR,
[S_IFCHR >> S_SHIFT] = BTRFS_FT_CHRDEV,
unsigned long nr_pages_ret = 0;
unsigned long total_compressed = 0;
unsigned long total_in = 0;
- unsigned long max_compressed = 128 * 1024;
- unsigned long max_uncompressed = 128 * 1024;
+ unsigned long max_compressed = SZ_128K;
+ unsigned long max_uncompressed = SZ_128K;
int i;
int will_compress;
int compress_type = root->fs_info->compress_type;
int redirty = 0;
/* if this is a small write inside eof, kick off a defrag */
- if ((end - start + 1) < 16 * 1024 &&
+ if ((end - start + 1) < SZ_16K &&
(start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
btrfs_add_inode_defrag(NULL, inode);
again:
will_compress = 0;
nr_pages = (end >> PAGE_CACHE_SHIFT) - (start >> PAGE_CACHE_SHIFT) + 1;
- nr_pages = min(nr_pages, (128 * 1024UL) / PAGE_CACHE_SIZE);
+ nr_pages = min_t(unsigned long, nr_pages, SZ_128K / PAGE_CACHE_SIZE);
/*
* we don't want to send crud past the end of i_size through
disk_num_bytes = num_bytes;
/* if this is a small write inside eof, kick off defrag */
- if (num_bytes < 64 * 1024 &&
+ if (num_bytes < SZ_64K &&
(start > 0 || end + 1 < BTRFS_I(inode)->disk_i_size))
btrfs_add_inode_defrag(NULL, inode);
* atomic_sub_return implies a barrier for waitqueue_active
*/
if (atomic_sub_return(nr_pages, &root->fs_info->async_delalloc_pages) <
- 5 * 1024 * 1024 &&
+ 5 * SZ_1M &&
waitqueue_active(&root->fs_info->async_submit_wait))
wake_up(&root->fs_info->async_submit_wait);
struct btrfs_root *root = BTRFS_I(inode)->root;
unsigned long nr_pages;
u64 cur_end;
- int limit = 10 * 1024 * 1024;
+ int limit = 10 * SZ_1M;
clear_extent_bit(&BTRFS_I(inode)->io_tree, start, end, EXTENT_LOCKED,
1, 0, NULL, GFP_NOFS);
!btrfs_test_opt(root, FORCE_COMPRESS))
cur_end = end;
else
- cur_end = min(end, start + 512 * 1024 - 1);
+ cur_end = min(end, start + SZ_512K - 1);
async_cow->end = cur_end;
INIT_LIST_HEAD(&async_cow->extents);
page_start = page_offset(page);
page_end = page_offset(page) + PAGE_CACHE_SIZE - 1;
- lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end, 0,
+ lock_extent_bits(&BTRFS_I(inode)->io_tree, page_start, page_end,
&cached_state);
/* already ordered? We're done */
lock_start = backref->file_pos;
lock_end = backref->file_pos + backref->num_bytes - 1;
lock_extent_bits(&BTRFS_I(inode)->io_tree, lock_start, lock_end,
- 0, &cached);
+ &cached);
ordered = btrfs_lookup_first_ordered_extent(inode, lock_end);
if (ordered) {
lock_extent_bits(io_tree, ordered_extent->file_offset,
ordered_extent->file_offset + ordered_extent->len - 1,
- 0, &cached_state);
+ &cached_state);
ret = test_range_bit(io_tree, ordered_extent->file_offset,
ordered_extent->file_offset + ordered_extent->len - 1,
start, (size_t)(end - start + 1));
}
-struct delayed_iput {
- struct list_head list;
- struct inode *inode;
-};
-
-/* JDM: If this is fs-wide, why can't we add a pointer to
- * btrfs_inode instead and avoid the allocation? */
void btrfs_add_delayed_iput(struct inode *inode)
{
struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info;
- struct delayed_iput *delayed;
+ struct btrfs_inode *binode = BTRFS_I(inode);
if (atomic_add_unless(&inode->i_count, -1, 1))
return;
- delayed = kmalloc(sizeof(*delayed), GFP_NOFS | __GFP_NOFAIL);
- delayed->inode = inode;
-
spin_lock(&fs_info->delayed_iput_lock);
- list_add_tail(&delayed->list, &fs_info->delayed_iputs);
+ if (binode->delayed_iput_count == 0) {
+ ASSERT(list_empty(&binode->delayed_iput));
+ list_add_tail(&binode->delayed_iput, &fs_info->delayed_iputs);
+ } else {
+ binode->delayed_iput_count++;
+ }
spin_unlock(&fs_info->delayed_iput_lock);
}
void btrfs_run_delayed_iputs(struct btrfs_root *root)
{
- LIST_HEAD(list);
struct btrfs_fs_info *fs_info = root->fs_info;
- struct delayed_iput *delayed;
- int empty;
-
- spin_lock(&fs_info->delayed_iput_lock);
- empty = list_empty(&fs_info->delayed_iputs);
- spin_unlock(&fs_info->delayed_iput_lock);
- if (empty)
- return;
down_read(&fs_info->delayed_iput_sem);
-
spin_lock(&fs_info->delayed_iput_lock);
- list_splice_init(&fs_info->delayed_iputs, &list);
- spin_unlock(&fs_info->delayed_iput_lock);
-
- while (!list_empty(&list)) {
- delayed = list_entry(list.next, struct delayed_iput, list);
- list_del(&delayed->list);
- iput(delayed->inode);
- kfree(delayed);
+ while (!list_empty(&fs_info->delayed_iputs)) {
+ struct btrfs_inode *inode;
+
+ inode = list_first_entry(&fs_info->delayed_iputs,
+ struct btrfs_inode, delayed_iput);
+ if (inode->delayed_iput_count) {
+ inode->delayed_iput_count--;
+ list_move_tail(&inode->delayed_iput,
+ &fs_info->delayed_iputs);
+ } else {
+ list_del_init(&inode->delayed_iput);
+ }
+ spin_unlock(&fs_info->delayed_iput_lock);
+ iput(&inode->vfs_inode);
+ spin_lock(&fs_info->delayed_iput_lock);
}
-
+ spin_unlock(&fs_info->delayed_iput_lock);
up_read(&root->fs_info->delayed_iput_sem);
}
ret = -ENOMEM;
goto out;
}
- path->reada = -1;
+ path->reada = READA_BACK;
key.objectid = BTRFS_ORPHAN_OBJECTID;
key.type = BTRFS_ORPHAN_ITEM_KEY;
path = btrfs_alloc_path();
if (!path)
return -ENOMEM;
- path->reada = -1;
+ path->reada = READA_BACK;
/*
* We want to drop from the next block forward in case this new size is
* up a huge file in a single leaf. Most of the time that
* bytes_deleted is > 0, it will be huge by the time we get here
*/
- if (be_nice && bytes_deleted > 32 * 1024 * 1024) {
+ if (be_nice && bytes_deleted > SZ_32M) {
if (btrfs_should_end_transaction(trans, root)) {
err = -EAGAIN;
goto error;
btrfs_free_path(path);
- if (be_nice && bytes_deleted > 32 * 1024 * 1024) {
+ if (be_nice && bytes_deleted > SZ_32M) {
unsigned long updates = trans->delayed_ref_updates;
if (updates) {
trans->delayed_ref_updates = 0;
}
wait_on_page_writeback(page);
- lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state);
+ lock_extent_bits(io_tree, page_start, page_end, &cached_state);
set_page_extent_mapped(page);
ordered = btrfs_lookup_ordered_extent(inode, page_start);
while (1) {
struct btrfs_ordered_extent *ordered;
- lock_extent_bits(io_tree, hole_start, block_end - 1, 0,
+ lock_extent_bits(io_tree, hole_start, block_end - 1,
&cached_state);
ordered = btrfs_lookup_ordered_range(inode, hole_start,
block_end - hole_start);
end = state->end;
spin_unlock(&io_tree->lock);
- lock_extent_bits(io_tree, start, end, 0, &cached_state);
+ lock_extent_bits(io_tree, start, end, &cached_state);
/*
* If still has DELALLOC flag, the extent didn't reach disk,
no_delete:
btrfs_remove_delayed_node(inode);
clear_inode(inode);
- return;
}
/*
if (!path)
return -ENOMEM;
- path->reada = 1;
+ path->reada = READA_FORWARD;
if (key_type == BTRFS_DIR_INDEX_KEY) {
INIT_LIST_HEAD(&ins_list);
static int btrfs_link(struct dentry *old_dentry, struct inode *dir,
struct dentry *dentry)
{
- struct btrfs_trans_handle *trans;
+ struct btrfs_trans_handle *trans = NULL;
struct btrfs_root *root = BTRFS_I(dir)->root;
struct inode *inode = d_inode(old_dentry);
u64 index;
trans = btrfs_start_transaction(root, 5);
if (IS_ERR(trans)) {
err = PTR_ERR(trans);
+ trans = NULL;
goto fail;
}
btrfs_log_new_name(trans, inode, NULL, parent);
}
- btrfs_end_transaction(trans, root);
btrfs_balance_delayed_items(root);
fail:
+ if (trans)
+ btrfs_end_transaction(trans, root);
if (drop_inode) {
inode_dec_link_count(inode);
iput(inode);
}
static noinline int uncompress_inline(struct btrfs_path *path,
- struct inode *inode, struct page *page,
+ struct page *page,
size_t pg_offset, u64 extent_offset,
struct btrfs_file_extent_item *item)
{
* Chances are we'll be called again, so go ahead and do
* readahead
*/
- path->reada = 1;
+ path->reada = READA_FORWARD;
}
ret = btrfs_lookup_file_extent(trans, root, path,
if (create == 0 && !PageUptodate(page)) {
if (btrfs_file_extent_compression(leaf, item) !=
BTRFS_COMPRESS_NONE) {
- ret = uncompress_inline(path, inode, page,
- pg_offset,
+ ret = uncompress_inline(path, page, pg_offset,
extent_offset, item);
if (ret) {
err = ret;
while (1) {
lock_extent_bits(&BTRFS_I(inode)->io_tree, lockstart, lockend,
- 0, cached_state);
+ cached_state);
/*
* We're concerned with the entire range that we're going to be
* doing DIO to, so we need to make sure theres no ordered
btrfs_start_ordered_extent(inode, ordered, 1);
btrfs_put_ordered_extent(ordered);
} else {
- /* Screw you mmap */
- ret = btrfs_fdatawrite_range(inode, lockstart, lockend);
- if (ret)
- break;
- ret = filemap_fdatawait_range(inode->i_mapping,
- lockstart,
- lockend);
- if (ret)
- break;
-
/*
- * If we found a page that couldn't be invalidated just
- * fall back to buffered.
+ * We could trigger writeback for this range (and wait
+ * for it to complete) and then invalidate the pages for
+ * this range (through invalidate_inode_pages2_range()),
+ * but that can lead us to a deadlock with a concurrent
+ * call to readpages() (a buffered read or a defrag call
+ * triggered a readahead) on a page lock due to an
+ * ordered dio extent we created before but did not have
+ * yet a corresponding bio submitted (whence it can not
+ * complete), which makes readpages() wait for that
+ * ordered extent to complete while holding a lock on
+ * that page.
*/
- ret = invalidate_inode_pages2_range(inode->i_mapping,
- lockstart >> PAGE_CACHE_SHIFT,
- lockend >> PAGE_CACHE_SHIFT);
- if (ret)
- break;
+ ret = -ENOTBLK;
+ break;
}
cond_resched();
return em;
}
-struct btrfs_dio_data {
- u64 outstanding_extents;
- u64 reserve;
-};
-
static void adjust_dio_outstanding_extents(struct inode *inode,
struct btrfs_dio_data *dio_data,
const u64 len)
btrfs_free_reserved_data_space(inode, start, len);
WARN_ON(dio_data->reserve < len);
dio_data->reserve -= len;
+ dio_data->unsubmitted_oe_range_end = start + len;
current->journal_info = dio_data;
}
bio_put(bio);
}
-static void btrfs_endio_direct_write(struct bio *bio)
+static void btrfs_endio_direct_write_update_ordered(struct inode *inode,
+ const u64 offset,
+ const u64 bytes,
+ const int uptodate)
{
- struct btrfs_dio_private *dip = bio->bi_private;
- struct inode *inode = dip->inode;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct btrfs_ordered_extent *ordered = NULL;
- u64 ordered_offset = dip->logical_offset;
- u64 ordered_bytes = dip->bytes;
- struct bio *dio_bio;
+ u64 ordered_offset = offset;
+ u64 ordered_bytes = bytes;
int ret;
again:
ret = btrfs_dec_test_first_ordered_pending(inode, &ordered,
&ordered_offset,
ordered_bytes,
- !bio->bi_error);
+ uptodate);
if (!ret)
goto out_test;
* our bio might span multiple ordered extents. If we haven't
* completed the accounting for the whole dio, go back and try again
*/
- if (ordered_offset < dip->logical_offset + dip->bytes) {
- ordered_bytes = dip->logical_offset + dip->bytes -
- ordered_offset;
+ if (ordered_offset < offset + bytes) {
+ ordered_bytes = offset + bytes - ordered_offset;
ordered = NULL;
goto again;
}
- dio_bio = dip->dio_bio;
+}
+
+static void btrfs_endio_direct_write(struct bio *bio)
+{
+ struct btrfs_dio_private *dip = bio->bi_private;
+ struct bio *dio_bio = dip->dio_bio;
+
+ btrfs_endio_direct_write_update_ordered(dip->inode,
+ dip->logical_offset,
+ dip->bytes,
+ !bio->bi_error);
kfree(dip);
dip->subio_endio = btrfs_subio_endio_read;
}
+ /*
+ * Reset the range for unsubmitted ordered extents (to a 0 length range)
+ * even if we fail to submit a bio, because in such case we do the
+ * corresponding error handling below and it must not be done a second
+ * time by btrfs_direct_IO().
+ */
+ if (write) {
+ struct btrfs_dio_data *dio_data = current->journal_info;
+
+ dio_data->unsubmitted_oe_range_end = dip->logical_offset +
+ dip->bytes;
+ dio_data->unsubmitted_oe_range_start =
+ dio_data->unsubmitted_oe_range_end;
+ }
+
ret = btrfs_submit_direct_hook(rw, dip, skip_sum);
if (!ret)
return;
dip = NULL;
io_bio = NULL;
} else {
- if (write) {
- struct btrfs_ordered_extent *ordered;
-
- ordered = btrfs_lookup_ordered_extent(inode,
- file_offset);
- set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
- /*
- * Decrements our ref on the ordered extent and removes
- * the ordered extent from the inode's ordered tree,
- * doing all the proper resource cleanup such as for the
- * reserved space and waking up any waiters for this
- * ordered extent (through btrfs_remove_ordered_extent).
- */
- btrfs_finish_ordered_io(ordered);
- } else {
+ if (write)
+ btrfs_endio_direct_write_update_ordered(inode,
+ file_offset,
+ dio_bio->bi_iter.bi_size,
+ 0);
+ else
unlock_extent(&BTRFS_I(inode)->io_tree, file_offset,
file_offset + dio_bio->bi_iter.bi_size - 1);
- }
+
dio_bio->bi_error = -EIO;
/*
* Releases and cleans up our dio_bio, no need to bio_put()
* originally calculated. Abuse current->journal_info for this.
*/
dio_data.reserve = round_up(count, root->sectorsize);
+ dio_data.unsubmitted_oe_range_start = (u64)offset;
+ dio_data.unsubmitted_oe_range_end = (u64)offset;
current->journal_info = &dio_data;
} else if (test_bit(BTRFS_INODE_READDIO_NEED_LOCK,
&BTRFS_I(inode)->runtime_flags)) {
if (dio_data.reserve)
btrfs_delalloc_release_space(inode, offset,
dio_data.reserve);
+ /*
+ * On error we might have left some ordered extents
+ * without submitting corresponding bios for them, so
+ * cleanup them up to avoid other tasks getting them
+ * and waiting for them to complete forever.
+ */
+ if (dio_data.unsubmitted_oe_range_start <
+ dio_data.unsubmitted_oe_range_end)
+ btrfs_endio_direct_write_update_ordered(inode,
+ dio_data.unsubmitted_oe_range_start,
+ dio_data.unsubmitted_oe_range_end -
+ dio_data.unsubmitted_oe_range_start,
+ 0);
} else if (ret >= 0 && (size_t)ret < count)
btrfs_delalloc_release_space(inode, offset,
count - (size_t)ret);
static int btrfs_writepage(struct page *page, struct writeback_control *wbc)
{
struct extent_io_tree *tree;
-
+ struct inode *inode = page->mapping->host;
+ int ret;
if (current->flags & PF_MEMALLOC) {
redirty_page_for_writepage(wbc, page);
unlock_page(page);
return 0;
}
+
+ /*
+ * If we are under memory pressure we will call this directly from the
+ * VM, we need to make sure we have the inode referenced for the ordered
+ * extent. If not just return like we didn't do anything.
+ */
+ if (!igrab(inode)) {
+ redirty_page_for_writepage(wbc, page);
+ return AOP_WRITEPAGE_ACTIVATE;
+ }
tree = &BTRFS_I(page->mapping->host)->io_tree;
- return extent_write_full_page(tree, page, btrfs_get_extent, wbc);
+ ret = extent_write_full_page(tree, page, btrfs_get_extent, wbc);
+ btrfs_add_delayed_iput(inode);
+ return ret;
}
static int btrfs_writepages(struct address_space *mapping,
}
if (!inode_evicting)
- lock_extent_bits(tree, page_start, page_end, 0, &cached_state);
+ lock_extent_bits(tree, page_start, page_end, &cached_state);
ordered = btrfs_lookup_ordered_extent(inode, page_start);
if (ordered) {
/*
btrfs_put_ordered_extent(ordered);
if (!inode_evicting) {
cached_state = NULL;
- lock_extent_bits(tree, page_start, page_end, 0,
+ lock_extent_bits(tree, page_start, page_end,
&cached_state);
}
}
}
wait_on_page_writeback(page);
- lock_extent_bits(io_tree, page_start, page_end, 0, &cached_state);
+ lock_extent_bits(io_tree, page_start, page_end, &cached_state);
set_page_extent_mapped(page);
/*
ei->dir_index = 0;
ei->last_unlink_trans = 0;
ei->last_log_commit = 0;
+ ei->delayed_iput_count = 0;
spin_lock_init(&ei->lock);
ei->outstanding_extents = 0;
mutex_init(&ei->delalloc_mutex);
btrfs_ordered_inode_tree_init(&ei->ordered_tree);
INIT_LIST_HEAD(&ei->delalloc_inodes);
+ INIT_LIST_HEAD(&ei->delayed_iput);
RB_CLEAR_NODE(&ei->rb_node);
return inode;
kmem_cache_destroy(btrfs_path_cachep);
if (btrfs_free_space_cachep)
kmem_cache_destroy(btrfs_free_space_cachep);
- if (btrfs_delalloc_work_cachep)
- kmem_cache_destroy(btrfs_delalloc_work_cachep);
}
int btrfs_init_cachep(void)
if (!btrfs_free_space_cachep)
goto fail;
- btrfs_delalloc_work_cachep = kmem_cache_create("btrfs_delalloc_work",
- sizeof(struct btrfs_delalloc_work), 0,
- SLAB_RECLAIM_ACCOUNT | SLAB_MEM_SPREAD,
- NULL);
- if (!btrfs_delalloc_work_cachep)
- goto fail;
-
return 0;
fail:
btrfs_destroy_cachep();
delalloc_work = container_of(work, struct btrfs_delalloc_work,
work);
inode = delalloc_work->inode;
- if (delalloc_work->wait) {
- btrfs_wait_ordered_range(inode, 0, (u64)-1);
- } else {
+ filemap_flush(inode->i_mapping);
+ if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
+ &BTRFS_I(inode)->runtime_flags))
filemap_flush(inode->i_mapping);
- if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
- &BTRFS_I(inode)->runtime_flags))
- filemap_flush(inode->i_mapping);
- }
if (delalloc_work->delay_iput)
btrfs_add_delayed_iput(inode);
}
struct btrfs_delalloc_work *btrfs_alloc_delalloc_work(struct inode *inode,
- int wait, int delay_iput)
+ int delay_iput)
{
struct btrfs_delalloc_work *work;
- work = kmem_cache_zalloc(btrfs_delalloc_work_cachep, GFP_NOFS);
+ work = kmalloc(sizeof(*work), GFP_NOFS);
if (!work)
return NULL;
init_completion(&work->completion);
INIT_LIST_HEAD(&work->list);
work->inode = inode;
- work->wait = wait;
work->delay_iput = delay_iput;
WARN_ON_ONCE(!inode);
btrfs_init_work(&work->work, btrfs_flush_delalloc_helper,
void btrfs_wait_and_free_delalloc_work(struct btrfs_delalloc_work *work)
{
wait_for_completion(&work->completion);
- kmem_cache_free(btrfs_delalloc_work_cachep, work);
+ kfree(work);
}
/*
}
spin_unlock(&root->delalloc_lock);
- work = btrfs_alloc_delalloc_work(inode, 0, delay_iput);
+ work = btrfs_alloc_delalloc_work(inode, delay_iput);
if (!work) {
if (delay_iput)
btrfs_add_delayed_iput(inode);
/*
* 2 items for inode item and ref
* 2 items for dir items
+ * 1 item for updating parent inode item
+ * 1 item for the inline extent item
* 1 item for xattr if selinux is on
*/
- trans = btrfs_start_transaction(root, 5);
+ trans = btrfs_start_transaction(root, 7);
if (IS_ERR(trans))
return PTR_ERR(trans);
if (err)
goto out_unlock_inode;
- err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
- if (err)
- goto out_unlock_inode;
-
path = btrfs_alloc_path();
if (!path) {
err = -ENOMEM;
inode_set_bytes(inode, name_len);
btrfs_i_size_write(inode, name_len);
err = btrfs_update_inode(trans, root, inode);
+ /*
+ * Last step, add directory indexes for our symlink inode. This is the
+ * last step to avoid extra cleanup of these indexes if an error happens
+ * elsewhere above.
+ */
+ if (!err)
+ err = btrfs_add_nondir(trans, dir, dentry, inode, 0, index);
if (err) {
drop_inode = 1;
goto out_unlock_inode;
}
}
- cur_bytes = min(num_bytes, 256ULL * 1024 * 1024);
+ cur_bytes = min_t(u64, num_bytes, SZ_256M);
cur_bytes = max(cur_bytes, min_size);
/*
* If we are severely fragmented we could end up with really
.fsync = btrfs_sync_file,
};
-static struct extent_io_ops btrfs_extent_io_ops = {
+static const struct extent_io_ops btrfs_extent_io_ops = {
.fill_delalloc = run_delalloc_range,
.submit_bio_hook = btrfs_submit_bio_hook,
.merge_bio_hook = btrfs_merge_bio_hook,
projects / cascardo / linux.git / blobdiff