I need to be able to safely deal with refs in my next patch, so convert refs and pages_reading to ints and introduce an eb_lock spinlock so I can use this to safely manipulate the refs count when marking eb's as stale. Thanks,
Signed-off-by: Josef Bacik <[email protected]> --- fs/btrfs/backref.c | 4 +- fs/btrfs/disk-io.c | 8 +++++- fs/btrfs/extent_io.c | 49 ++++++++++++++++++++++------------------- fs/btrfs/extent_io.h | 22 ++++++++++++++++-- include/trace/events/btrfs.h | 2 +- 5 files changed, 54 insertions(+), 31 deletions(-) diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index b9a8432..04cf24d 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -904,7 +904,7 @@ static char *iref_to_path(struct btrfs_root *fs_root, struct btrfs_path *path, eb = path->nodes[0]; /* make sure we can use eb after releasing the path */ if (eb != eb_in) - atomic_inc(&eb->refs); + extent_buffer_get(eb); btrfs_release_path(path); iref = btrfs_item_ptr(eb, slot, struct btrfs_inode_ref); @@ -1251,7 +1251,7 @@ static int iterate_irefs(u64 inum, struct btrfs_root *fs_root, slot = path->slots[0]; eb = path->nodes[0]; /* make sure we can use eb after releasing the path */ - atomic_inc(&eb->refs); + extent_buffer_get(eb); btrfs_release_path(path); item = btrfs_item_nr(eb, slot); diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 6f97129..b8f2284 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -566,7 +566,9 @@ static int btree_readpage_end_io_hook(struct page *page, u64 start, u64 end, tree = &BTRFS_I(page->mapping->host)->io_tree; eb = (struct extent_buffer *)page->private; - reads_done = atomic_dec_and_test(&eb->pages_reading); + spin_lock(&eb->eb_lock); + reads_done = (--eb->pages_reading == 0); + spin_unlock(&eb->eb_lock); if (!reads_done) goto err; @@ -3391,7 +3393,9 @@ static int btrfs_destroy_marked_extents(struct btrfs_root *root, if (eb) { ret = test_and_clear_bit(EXTENT_BUFFER_DIRTY, &eb->bflags); - atomic_set(&eb->refs, 1); + spin_lock(&eb->eb_lock); + eb->refs = 1; + spin_unlock(&eb->eb_lock); } if (PageWriteback(page)) end_page_writeback(page); diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 6d948eb..f6f31b6 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -94,7 +94,7 @@ void extent_io_exit(void) eb = list_entry(buffers.next, struct extent_buffer, leak_list); printk(KERN_ERR "btrfs buffer leak start %llu len %lu " "refs %d\n", (unsigned long long)eb->start, - eb->len, atomic_read(&eb->refs)); + eb->len, eb->refs); list_del(&eb->leak_list); kmem_cache_free(extent_buffer_cache, eb); } @@ -3573,6 +3573,7 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree, eb->start = start; eb->len = len; eb->tree = tree; + spin_lock_init(&eb->eb_lock); rwlock_init(&eb->lock); atomic_set(&eb->write_locks, 0); atomic_set(&eb->read_locks, 0); @@ -3589,8 +3590,8 @@ static struct extent_buffer *__alloc_extent_buffer(struct extent_io_tree *tree, list_add(&eb->leak_list, &buffers); spin_unlock_irqrestore(&leak_lock, flags); #endif - atomic_set(&eb->refs, 1); - atomic_set(&eb->pages_reading, 0); + eb->refs = 1; + eb->pages_reading = 0; if (len > MAX_INLINE_EXTENT_BUFFER_SIZE) { struct page **pages; @@ -3677,7 +3678,7 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, rcu_read_lock(); eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT); - if (eb && atomic_inc_not_zero(&eb->refs)) { + if (eb && extent_buffer_inc_not_zero(eb)) { rcu_read_unlock(); mark_page_accessed(eb->pages[0]); return eb; @@ -3705,7 +3706,7 @@ struct extent_buffer *alloc_extent_buffer(struct extent_io_tree *tree, * overwrite page->private. */ exists = (struct extent_buffer *)p->private; - if (atomic_inc_not_zero(&exists->refs)) { + if (extent_buffer_inc_not_zero(exists)) { spin_unlock(&mapping->private_lock); unlock_page(p); goto free_eb; @@ -3742,7 +3743,7 @@ again: if (ret == -EEXIST) { exists = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT); - if (!atomic_inc_not_zero(&exists->refs)) { + if (!extent_buffer_inc_not_zero(exists)) { spin_unlock(&tree->buffer_lock); radix_tree_preload_end(); synchronize_rcu(); @@ -3754,7 +3755,7 @@ again: goto free_eb; } /* add one reference for the tree */ - atomic_inc(&eb->refs); + extent_buffer_get(eb); spin_unlock(&tree->buffer_lock); radix_tree_preload_end(); @@ -3782,8 +3783,7 @@ free_eb: unlock_page(eb->pages[i]); } - if (!atomic_dec_and_test(&eb->refs)) - return exists; + /* The ref count of eb can only be 1 at this point, just release it */ btrfs_release_extent_buffer(eb); return exists; } @@ -3795,7 +3795,7 @@ struct extent_buffer *find_extent_buffer(struct extent_io_tree *tree, rcu_read_lock(); eb = radix_tree_lookup(&tree->buffer, start >> PAGE_CACHE_SHIFT); - if (eb && atomic_inc_not_zero(&eb->refs)) { + if (eb && extent_buffer_inc_not_zero(eb)) { rcu_read_unlock(); mark_page_accessed(eb->pages[0]); return eb; @@ -3810,8 +3810,12 @@ void free_extent_buffer(struct extent_buffer *eb) if (!eb) return; - if (!atomic_dec_and_test(&eb->refs)) + spin_lock(&eb->eb_lock); + if (--eb->refs != 0) { + spin_unlock(&eb->eb_lock); return; + } + spin_unlock(&eb->eb_lock); WARN_ON(1); } @@ -4041,7 +4045,9 @@ int read_extent_buffer_pages(struct extent_io_tree *tree, goto unlock_exit; } - atomic_set(&eb->pages_reading, num_reads); + spin_lock(&eb->eb_lock); + eb->pages_reading = num_reads; + spin_unlock(&eb->eb_lock); for (i = start_i; i < num_pages; i++) { page = extent_buffer_page(eb, i); if (!PageUptodate(page)) { @@ -4434,32 +4440,29 @@ int try_release_extent_buffer(struct extent_io_tree *tree, struct page *page) u64 start = page_offset(page); struct extent_buffer *eb = (struct extent_buffer *)page->private; int ret = 1; + int release = 0; if (!PagePrivate(page) || !eb) return 1; spin_lock(&tree->buffer_lock); - if (atomic_read(&eb->refs) > 1 || - test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) { + spin_lock(&eb->eb_lock); + if (eb->refs > 1 || test_bit(EXTENT_BUFFER_DIRTY, &eb->bflags)) { + spin_unlock(&eb->eb_lock); ret = 0; goto out; } + eb->refs--; + release = 1; + spin_unlock(&eb->eb_lock); - /* - * set @eb->refs to 0 if it is already 1, and then release the @eb. - * Or go back. - */ - if (atomic_cmpxchg(&eb->refs, 1, 0) != 1) { - ret = 0; - goto out; - } radix_tree_delete(&tree->buffer, start >> PAGE_CACHE_SHIFT); btrfs_release_extent_buffer_page(eb, 0); out: spin_unlock(&tree->buffer_lock); /* at this point we can safely release the extent buffer */ - if (atomic_read(&eb->refs) == 0) + if (release) call_rcu(&eb->rcu_head, btrfs_release_extent_buffer_rcu); return ret; } diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index f030a2d..d11e872 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -128,8 +128,9 @@ struct extent_buffer { unsigned long map_len; unsigned long bflags; struct extent_io_tree *tree; - atomic_t refs; - atomic_t pages_reading; + spinlock_t eb_lock; + int refs; + int pages_reading; struct list_head leak_list; struct rcu_head rcu_head; pid_t lock_owner; @@ -272,7 +273,22 @@ struct page *extent_buffer_page(struct extent_buffer *eb, unsigned long i); static inline void extent_buffer_get(struct extent_buffer *eb) { - atomic_inc(&eb->refs); + spin_lock(&eb->eb_lock); + eb->refs++; + spin_unlock(&eb->eb_lock); +} + +static inline int extent_buffer_inc_not_zero(struct extent_buffer *eb) +{ + int ret = 0; + + spin_lock(&eb->eb_lock); + if (eb->refs) { + eb->refs++; + ret = 1; + } + spin_unlock(&eb->eb_lock); + return ret; } int memcmp_extent_buffer(struct extent_buffer *eb, const void *ptrv, diff --git a/include/trace/events/btrfs.h b/include/trace/events/btrfs.h index 84f3001..b4b04e7 100644 --- a/include/trace/events/btrfs.h +++ b/include/trace/events/btrfs.h @@ -618,7 +618,7 @@ TRACE_EVENT(btrfs_cow_block, TP_fast_assign( __entry->root_objectid = root->root_key.objectid; __entry->buf_start = buf->start; - __entry->refs = atomic_read(&buf->refs); + __entry->refs = buf->refs; __entry->cow_start = cow->start; __entry->buf_level = btrfs_header_level(buf); __entry->cow_level = btrfs_header_level(cow); -- 1.7.5.2 -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
