In this patch, we make three things:
a) skiplist -> rcu-skiplist
This is quite direct, since in skiplist each level is a list,
any modification to the skiplist refers to "pointers change",
which fits RCU's sematic.
b) use rcu lock to protect extent_map instead of rwlock.
c) make extent_map reclaim after dropping the updater side lock.
Signed-off-by: Liu Bo <liubo2009@xxxxxxxxxxxxxx>
---
fs/btrfs/compression.c | 8 +++---
fs/btrfs/disk-io.c | 15 ++++++----
fs/btrfs/extent_io.c | 13 ++++-----
fs/btrfs/extent_map.c | 39 +++++++++++++++++---------
fs/btrfs/extent_map.h | 7 +++--
fs/btrfs/file.c | 23 +++++++++++-----
fs/btrfs/inode.c | 69 ++++++++++++++++++++++++++++++++---------------
fs/btrfs/ioctl.c | 8 +++---
fs/btrfs/relocation.c | 9 ++++--
fs/btrfs/scrub.c | 4 +-
fs/btrfs/skiplist.c | 6 ++--
fs/btrfs/skiplist.h | 25 +++++++++++------
fs/btrfs/volumes.c | 46 ++++++++++++++++++--------------
13 files changed, 168 insertions(+), 104 deletions(-)
diff --git a/fs/btrfs/compression.c b/fs/btrfs/compression.c
index 14f1c5a..bb4ac31 100644
--- a/fs/btrfs/compression.c
+++ b/fs/btrfs/compression.c
@@ -498,10 +498,10 @@ static noinline int add_ra_bio_pages(struct inode *inode,
*/
set_page_extent_mapped(page);
lock_extent(tree, last_offset, end, GFP_NOFS);
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, last_offset,
PAGE_CACHE_SIZE);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
if (!em || last_offset < em->start ||
(last_offset + PAGE_CACHE_SIZE > extent_map_end(em)) ||
@@ -583,11 +583,11 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
em_tree = &BTRFS_I(inode)->extent_tree;
/* we need the actual starting offset of this extent in the file */
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree,
page_offset(bio->bi_io_vec->bv_page),
PAGE_CACHE_SIZE);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
compressed_len = em->block_len;
cb = kmalloc(compressed_bio_size(root, compressed_len), GFP_NOFS);
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 3f9d555..2dbc969 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -189,17 +189,17 @@ static struct extent_map *btree_get_extent(struct inode *inode,
{
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
int ret;
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, start, len);
+ rcu_read_unlock();
if (em) {
em->bdev =
BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
- read_unlock(&em_tree->lock);
goto out;
}
- read_unlock(&em_tree->lock);
em = alloc_extent_map();
if (!em) {
@@ -212,8 +212,12 @@ static struct extent_map *btree_get_extent(struct inode *inode,
em->block_start = 0;
em->bdev = BTRFS_I(inode)->root->fs_info->fs_devices->latest_bdev;
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1, &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
+
if (ret == -EEXIST) {
u64 failed_start = em->start;
u64 failed_len = em->len;
@@ -231,7 +235,6 @@ static struct extent_map *btree_get_extent(struct inode *inode,
free_extent_map(em);
em = NULL;
}
- write_unlock(&em_tree->lock);
if (ret)
em = ERR_PTR(ret);
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 49f3c9d..30a8270 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -2013,10 +2013,10 @@ static int bio_readpage_error(struct bio *failed_bio, struct page *page,
failrec->bio_flags = 0;
failrec->in_validation = 0;
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, start, failrec->len);
+ rcu_read_unlock();
if (!em) {
- read_unlock(&em_tree->lock);
kfree(failrec);
return -EIO;
}
@@ -2025,7 +2025,6 @@ static int bio_readpage_error(struct bio *failed_bio, struct page *page,
free_extent_map(em);
em = NULL;
}
- read_unlock(&em_tree->lock);
if (!em || IS_ERR(em)) {
kfree(failrec);
@@ -3286,15 +3285,15 @@ int try_release_extent_mapping(struct extent_map_tree *map,
u64 len;
while (start <= end) {
len = end - start + 1;
- write_lock(&map->lock);
+ spin_lock(&map->lock);
em = lookup_extent_mapping(map, start, len);
if (IS_ERR_OR_NULL(em)) {
- write_unlock(&map->lock);
+ spin_unlock(&map->lock);
break;
}
if (test_bit(EXTENT_FLAG_PINNED, &em->flags) ||
em->start != start) {
- write_unlock(&map->lock);
+ spin_unlock(&map->lock);
free_extent_map(em);
break;
}
@@ -3307,7 +3306,7 @@ int try_release_extent_mapping(struct extent_map_tree *map,
free_extent_map(em);
}
start = extent_map_end(em);
- write_unlock(&map->lock);
+ spin_unlock(&map->lock);
/* once for us */
free_extent_map(em);
diff --git a/fs/btrfs/extent_map.c b/fs/btrfs/extent_map.c
index 746084c..e2e8af0 100644
--- a/fs/btrfs/extent_map.c
+++ b/fs/btrfs/extent_map.c
@@ -67,7 +67,7 @@ void extent_map_tree_init(struct extent_map_tree *tree)
{
tree->head.start = (-1ULL);
sl_init_list(&tree->map, &tree->head.sl_node);
- rwlock_init(&tree->lock);
+ spin_lock_init(&tree->lock);
}
/**
@@ -100,8 +100,11 @@ struct extent_map *alloc_extent_map(void)
return em;
}
-static inline void __free_extent_map(struct extent_map *em)
+static inline void __free_extent_map(struct rcu_head *head)
{
+ struct sl_node *node = container_of(head, struct sl_node, rcu_head);
+ struct extent_map *em = sl_entry(node, struct extent_map, sl_node);
+
#if MAP_LEAK_DEBUG
unsigned long flags;
@@ -129,7 +132,7 @@ void free_extent_map(struct extent_map *em)
WARN_ON(atomic_read(&em->refs) == 0);
if (atomic_dec_and_test(&em->refs))
- __free_extent_map(em);
+ call_rcu(&em->sl_node.rcu_head, __free_extent_map);
}
static inline int in_entry(struct sl_node *node, u64 offset)
@@ -166,14 +169,14 @@ static struct sl_node *sl_search(struct sl_list *list, u64 offset,
BUG_ON(!list);
level = list->level;
- p = list->head;
+ p = rcu_dereference(list->head);
BUG_ON(!p);
if (sl_empty(p))
return NULL;
do {
while (entry = next_entry(p, level, &q), entry->start <= offset)
- p = q;
+ p = rcu_dereference(q);
if (in_entry(p, offset))
return p;
@@ -262,7 +265,9 @@ static int mergable_maps(struct extent_map *prev, struct extent_map *next)
return 0;
}
-static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
+static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em,
+ struct extent_map **to_free1,
+ struct extent_map **to_free2)
{
struct extent_map *merge = NULL;
struct sl_node *sl;
@@ -278,7 +283,8 @@ static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
em->block_start = merge->block_start;
merge->in_tree = 0;
sl_erase(&merge->sl_node, &tree->map);
- free_extent_map(merge);
+ if (merge)
+ *to_free1 = merge;
}
}
@@ -290,7 +296,8 @@ static void try_merge_map(struct extent_map_tree *tree, struct extent_map *em)
em->block_len += merge->len;
merge->in_tree = 0;
sl_erase(&merge->sl_node, &tree->map);
- free_extent_map(merge);
+ if (merge)
+ *to_free2 = merge;
}
}
@@ -298,8 +305,9 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
{
int ret = 0;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
- write_lock(&tree->lock);
+ spin_lock(&tree->lock);
em = lookup_extent_mapping(tree, start, len);
WARN_ON(!em || em->start != start);
@@ -308,11 +316,13 @@ int unpin_extent_cache(struct extent_map_tree *tree, u64 start, u64 len)
clear_bit(EXTENT_FLAG_PINNED, &em->flags);
- try_merge_map(tree, em);
+ try_merge_map(tree, em, &to_free1, &to_free2);
free_extent_map(em);
out:
- write_unlock(&tree->lock);
+ spin_unlock(&tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
return ret;
}
@@ -326,8 +336,9 @@ out:
* into the tree directly, with an additional reference taken, or a
* reference dropped if the merge attempt was successful.
*/
-int add_extent_mapping(struct extent_map_tree *tree,
- struct extent_map *em)
+int add_extent_mapping(struct extent_map_tree *tree, struct extent_map *em,
+ struct extent_map **to_free1,
+ struct extent_map **to_free2)
{
int ret = 0;
struct sl_node *sl_node;
@@ -340,7 +351,7 @@ int add_extent_mapping(struct extent_map_tree *tree,
atomic_inc(&em->refs);
- try_merge_map(tree, em);
+ try_merge_map(tree, em, to_free1, to_free2);
out:
return ret;
}
diff --git a/fs/btrfs/extent_map.h b/fs/btrfs/extent_map.h
index 6d2c247..c61a105 100644
--- a/fs/btrfs/extent_map.h
+++ b/fs/btrfs/extent_map.h
@@ -41,7 +41,7 @@ struct map_head {
struct extent_map_tree {
struct sl_list map;
- rwlock_t lock;
+ spinlock_t lock;
struct map_head head;
};
@@ -62,8 +62,9 @@ static inline u64 extent_map_block_end(struct extent_map *em)
void extent_map_tree_init(struct extent_map_tree *tree);
struct extent_map *lookup_extent_mapping(struct extent_map_tree *tree,
u64 start, u64 len);
-int add_extent_mapping(struct extent_map_tree *tree,
- struct extent_map *em);
+int add_extent_mapping(struct extent_map_tree *tree, struct extent_map *em,
+ struct extent_map **to_free1,
+ struct extent_map **to_free2);
int remove_extent_mapping(struct extent_map_tree *tree, struct extent_map *em);
struct extent_map *alloc_extent_map(void);
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index cc7492c..8284202 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -435,10 +435,12 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
struct extent_map *em;
struct extent_map *split = NULL;
struct extent_map *split2 = NULL;
+ struct extent_map *to_free[4];
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
u64 len = end - start + 1;
int ret;
int testend = 1;
+ int i;
unsigned long flags;
int compressed = 0;
@@ -454,24 +456,27 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
split2 = alloc_extent_map();
BUG_ON(!split || !split2);
- write_lock(&em_tree->lock);
+ for (i = 0; i < 4; i++)
+ to_free[i] = NULL;
+ spin_lock(&em_tree->lock);
em = lookup_extent_mapping(em_tree, start, len);
if (!em) {
- write_unlock(&em_tree->lock);
+ spin_unlock(&em_tree->lock);
break;
}
+
flags = em->flags;
if (skip_pinned && test_bit(EXTENT_FLAG_PINNED, &em->flags)) {
if (testend && em->start + em->len >= start + len) {
free_extent_map(em);
- write_unlock(&em_tree->lock);
+ spin_unlock(&em_tree->lock);
break;
}
start = em->start + em->len;
if (testend)
len = start + len - (em->start + em->len);
free_extent_map(em);
- write_unlock(&em_tree->lock);
+ spin_unlock(&em_tree->lock);
continue;
}
compressed = test_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
@@ -493,7 +498,8 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
split->bdev = em->bdev;
split->flags = flags;
split->compress_type = em->compress_type;
- ret = add_extent_mapping(em_tree, split);
+ ret = add_extent_mapping(em_tree, split, &to_free[0],
+ &to_free[1]);
BUG_ON(ret);
free_extent_map(split);
split = split2;
@@ -519,12 +525,15 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
split->orig_start = split->start;
}
- ret = add_extent_mapping(em_tree, split);
+ ret = add_extent_mapping(em_tree, split, &to_free[2],
+ &to_free[3]);
BUG_ON(ret);
free_extent_map(split);
split = NULL;
}
- write_unlock(&em_tree->lock);
+ spin_unlock(&em_tree->lock);
+ for (i = 0; i < 4; i++)
+ free_extent_map(to_free[i]);
/* once for us */
free_extent_map(em);
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 13b0542..d896b39 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -573,6 +573,7 @@ static noinline int submit_compressed_extents(struct inode *inode,
struct btrfs_trans_handle *trans;
struct btrfs_key ins;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
struct extent_io_tree *io_tree;
@@ -675,9 +676,12 @@ retry:
set_bit(EXTENT_FLAG_COMPRESSED, &em->flags);
while (1) {
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1,
+ &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
if (ret != -EEXIST) {
free_extent_map(em);
break;
@@ -732,8 +736,9 @@ static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
struct extent_map *em;
u64 alloc_hint = 0;
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = search_extent_mapping(em_tree, start, num_bytes);
+ rcu_read_unlock();
if (em) {
/*
* if block start isn't an actual block number then find the
@@ -752,7 +757,6 @@ static u64 get_extent_allocation_hint(struct inode *inode, u64 start,
free_extent_map(em);
}
}
- read_unlock(&em_tree->lock);
return alloc_hint;
}
@@ -786,6 +790,7 @@ static noinline int cow_file_range(struct inode *inode,
u64 blocksize = root->sectorsize;
struct btrfs_key ins;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
int ret = 0;
@@ -854,9 +859,12 @@ static noinline int cow_file_range(struct inode *inode,
set_bit(EXTENT_FLAG_PINNED, &em->flags);
while (1) {
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1,
+ &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
if (ret != -EEXIST) {
free_extent_map(em);
break;
@@ -1195,6 +1203,7 @@ out_check:
if (extent_type == BTRFS_FILE_EXTENT_PREALLOC) {
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
struct extent_map_tree *em_tree;
em_tree = &BTRFS_I(inode)->extent_tree;
em = alloc_extent_map();
@@ -1207,9 +1216,12 @@ out_check:
em->bdev = root->fs_info->fs_devices->latest_bdev;
set_bit(EXTENT_FLAG_PINNED, &em->flags);
while (1) {
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1,
+ &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
if (ret != -EEXIST) {
free_extent_map(em);
break;
@@ -4862,7 +4874,9 @@ out_fail:
static int merge_extent_mapping(struct extent_map_tree *em_tree,
struct extent_map *existing,
struct extent_map *em,
- u64 map_start, u64 map_len)
+ u64 map_start, u64 map_len,
+ struct extent_map **to_free1,
+ struct extent_map **to_free2)
{
u64 start_diff;
@@ -4875,7 +4889,7 @@ static int merge_extent_mapping(struct extent_map_tree *em_tree,
em->block_start += start_diff;
em->block_len -= start_diff;
}
- return add_extent_mapping(em_tree, em);
+ return add_extent_mapping(em_tree, em, to_free1, to_free2);
}
static noinline int uncompress_inline(struct btrfs_path *path,
@@ -4944,17 +4958,19 @@ struct extent_map *btrfs_get_extent(struct inode *inode, struct page *page,
struct extent_buffer *leaf;
struct btrfs_key found_key;
struct extent_map *em = NULL;
+ struct extent_map *to_free[4];
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
struct btrfs_trans_handle *trans = NULL;
int compress_type;
+ int i;
again:
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, start, len);
if (em)
em->bdev = root->fs_info->fs_devices->latest_bdev;
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
if (em) {
if (em->start > start || em->start + em->len <= start)
@@ -5166,8 +5182,10 @@ insert:
}
err = 0;
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
+ for (i = 0; i < 4; i++)
+ to_free[i] = NULL;
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free[0], &to_free[1]);
/* it is possible that someone inserted the extent into the tree
* while we had the lock dropped. It is also possible that
* an overlapping map exists in the tree
@@ -5189,7 +5207,9 @@ insert:
if (existing) {
err = merge_extent_mapping(em_tree, existing,
em, start,
- root->sectorsize);
+ root->sectorsize,
+ &to_free[2],
+ &to_free[3]);
free_extent_map(existing);
if (err) {
free_extent_map(em);
@@ -5206,7 +5226,9 @@ insert:
err = 0;
}
}
- write_unlock(&em_tree->lock);
+ spin_unlock(&em_tree->lock);
+ for (i = 0; i < 4; i++)
+ free_extent_map(to_free[i]);
out:
trace_btrfs_get_extent(root, em);
@@ -5414,9 +5436,12 @@ static struct extent_map *btrfs_new_extent_direct(struct inode *inode,
set_bit(EXTENT_FLAG_PINNED, &em->flags);
while (insert) {
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1, &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
if (ret != -EEXIST)
break;
btrfs_drop_extent_cache(inode, start, start + em->len - 1, 0);
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index c04f02c..83fc601 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -673,9 +673,9 @@ static int check_defrag_in_cache(struct inode *inode, u64 offset, int thresh)
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
u64 end;
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, offset, PAGE_CACHE_SIZE);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
if (em) {
end = extent_map_end(em);
@@ -782,9 +782,9 @@ static int should_defrag_range(struct inode *inode, u64 start, u64 len,
* hopefully we have this extent in the tree already, try without
* the full extent lock
*/
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, start, len);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
if (!em) {
/* get the big lock and read metadata off disk */
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index cfb5543..b92d207 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -2884,6 +2884,7 @@ int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
struct btrfs_root *root = BTRFS_I(inode)->root;
struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
int ret = 0;
em = alloc_extent_map();
@@ -2899,9 +2900,11 @@ int setup_extent_mapping(struct inode *inode, u64 start, u64 end,
lock_extent(&BTRFS_I(inode)->io_tree, start, end, GFP_NOFS);
while (1) {
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1, &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
if (ret != -EEXIST) {
free_extent_map(em);
break;
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index ddf2c90..5aec748 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -1374,9 +1374,9 @@ static noinline_for_stack int scrub_chunk(struct scrub_dev *sdev,
int i;
int ret = -EINVAL;
- read_lock(&map_tree->map_tree.lock);
+ rcu_read_lock();
em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
- read_unlock(&map_tree->map_tree.lock);
+ rcu_read_unlock();
if (!em)
return -EINVAL;
diff --git a/fs/btrfs/skiplist.c b/fs/btrfs/skiplist.c
index c803478..1069922 100644
--- a/fs/btrfs/skiplist.c
+++ b/fs/btrfs/skiplist.c
@@ -62,7 +62,7 @@ inline void sl_link_node(struct sl_node *node, struct sl_node **backlook,
node->next[i] = q;
node->prev[i] = p;
- p->next[i] = node;
+ rcu_assign_pointer(p->next[i], node);
q->prev[i] = node;
i++;
@@ -78,11 +78,11 @@ void sl_erase(struct sl_node *node, struct sl_list *list)
level = node->level;
- for (i = 0; i <= level; i++) {
+ for (i = level; i >= 0; i--) {
prev = node->prev[i];
next = node->next[i];
- prev->next[i] = next;
+ rcu_assign_pointer(prev->next[i], next);
next->prev[i] = prev;
node->next[i] = node;
node->prev[i] = node;
diff --git a/fs/btrfs/skiplist.h b/fs/btrfs/skiplist.h
index 3e414b5..2ae997d 100644
--- a/fs/btrfs/skiplist.h
+++ b/fs/btrfs/skiplist.h
@@ -102,41 +102,48 @@ struct sl_node *sl_insert_node(struct sl_list *list, u64 offset,
#define _SKIPLIST_H
#include <linux/random.h>
+#include <linux/rcupdate.h>
#define MAXLEVEL 16
/* double p = 0.25; */
struct sl_node {
- struct sl_node **next;
- struct sl_node **prev;
+ struct sl_node __rcu **next;
+ struct sl_node __rcu **prev;
+ struct rcu_head rcu_head;
unsigned int level;
unsigned int head:1;
};
struct sl_list {
- struct sl_node *head;
- struct sl_node *h_next[MAXLEVEL];
- struct sl_node *h_prev[MAXLEVEL];
+ struct sl_node __rcu *head;
+ struct sl_node __rcu *h_next[MAXLEVEL];
+ struct sl_node __rcu *h_prev[MAXLEVEL];
unsigned int level;
};
-#define sl_entry(ptr, type, member) container_of(ptr, type, member)
+#define sl_entry(ptr, type, member) \
+ ({ \
+ typeof(*ptr) __rcu *__ptr = (typeof(*ptr) __rcu __force *)ptr; \
+ container_of((typeof(ptr))rcu_dereference(__ptr), \
+ type, member); \
+ })
static inline int sl_empty(const struct sl_node *head)
{
- return head->next[0] == head;
+ return (rcu_dereference(head->next[0]) == head);
}
static inline struct sl_node *__sl_next_with_level(struct sl_node *node,
int level)
{
- return node->next[level];
+ return rcu_dereference(node->next[level]);
}
static inline struct sl_node *__sl_prev_with_level(struct sl_node *node,
int level)
{
- return node->prev[level];
+ return rcu_dereference(node->prev[level]);
}
static inline struct sl_node *sl_next(struct sl_node *node)
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index adaac9e..c41502d 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -1955,9 +1955,9 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
* step two, delete the device extents and the
* chunk tree entries
*/
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, chunk_offset, 1);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
BUG_ON(em->start > chunk_offset ||
em->start + em->len < chunk_offset);
@@ -1988,9 +1988,9 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
ret = btrfs_remove_block_group(trans, extent_root, chunk_offset);
BUG_ON(ret);
- write_lock(&em_tree->lock);
+ spin_lock(&em_tree->lock);
remove_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ spin_unlock(&em_tree->lock);
kfree(map);
em->bdev = NULL;
@@ -2378,6 +2378,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
struct map_lookup *map = NULL;
struct extent_map_tree *em_tree;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
struct btrfs_device_info *devices_info = NULL;
u64 total_avail;
int num_stripes; /* total number of stripes to allocate */
@@ -2589,9 +2590,11 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
em->block_len = em->len;
em_tree = &extent_root->fs_info->mapping_tree.map_tree;
- write_lock(&em_tree->lock);
- ret = add_extent_mapping(em_tree, em);
- write_unlock(&em_tree->lock);
+ spin_lock(&em_tree->lock);
+ ret = add_extent_mapping(em_tree, em, &to_free1, &to_free2);
+ spin_unlock(&em_tree->lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
BUG_ON(ret);
free_extent_map(em);
@@ -2800,9 +2803,9 @@ int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset)
int readonly = 0;
int i;
- read_lock(&map_tree->map_tree.lock);
+ rcu_read_lock();
em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
- read_unlock(&map_tree->map_tree.lock);
+ rcu_read_unlock();
if (!em)
return 1;
@@ -2854,9 +2857,9 @@ int btrfs_num_copies(struct btrfs_mapping_tree *map_tree, u64 logical, u64 len)
struct extent_map_tree *em_tree = &map_tree->map_tree;
int ret;
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, logical, len);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
BUG_ON(!em);
BUG_ON(em->start > logical || em->start + em->len < logical);
@@ -2921,9 +2924,9 @@ again:
atomic_set(&bbio->error, 0);
}
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, logical, *length);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
if (!em) {
printk(KERN_CRIT "unable to find logical %llu len %llu\n",
@@ -3187,9 +3190,9 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
u64 stripe_nr;
int i, j, nr = 0;
- read_lock(&em_tree->lock);
+ rcu_read_lock();
em = lookup_extent_mapping(em_tree, chunk_start, 1);
- read_unlock(&em_tree->lock);
+ rcu_read_unlock();
BUG_ON(!em || em->start != chunk_start);
map = (struct map_lookup *)em->bdev;
@@ -3461,6 +3464,7 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
struct btrfs_mapping_tree *map_tree = &root->fs_info->mapping_tree;
struct map_lookup *map;
struct extent_map *em;
+ struct extent_map *to_free1 = NULL, *to_free2 = NULL;
u64 logical;
u64 length;
u64 devid;
@@ -3472,9 +3476,9 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
logical = key->offset;
length = btrfs_chunk_length(leaf, chunk);
- read_lock(&map_tree->map_tree.lock);
+ rcu_read_lock();
em = lookup_extent_mapping(&map_tree->map_tree, logical, 1);
- read_unlock(&map_tree->map_tree.lock);
+ rcu_read_unlock();
/* already mapped? */
if (em && em->start <= logical && em->start + em->len > logical) {
@@ -3533,9 +3537,11 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
map->stripes[i].dev->in_fs_metadata = 1;
}
- write_lock(&map_tree->map_tree.lock);
- ret = add_extent_mapping(&map_tree->map_tree, em);
- write_unlock(&map_tree->map_tree.lock);
+ spin_lock(&map_tree->map_tree.lock);
+ ret = add_extent_mapping(&map_tree->map_tree, em, &to_free1, &to_free2);
+ spin_unlock(&map_tree->map_tree.lock);
+ free_extent_map(to_free1);
+ free_extent_map(to_free2);
BUG_ON(ret);
free_extent_map(em);
--
1.6.5.2
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
