aboutsummaryrefslogtreecommitdiff
path: root/fs
diff options
context:
space:
mode:
authorTao Ma <tao.ma@oracle.com>2008-11-12 08:26:58 +0800
committerMark Fasheh <mfasheh@suse.com>2009-01-05 08:34:18 -0800
commit2891d290aa6eee0821f7e4ad0b1c4ae4d964b0f1 (patch)
tree2f06b8d3afcf1986d0f041319a8ec08395acdf83 /fs
parent976331d8789d4d84a11b45b87c520ade83715343 (diff)
ocfs2: Add clusters free in dealloc_ctxt.
Now in ocfs2 xattr set, the whole process are divided into many small parts and they are wrapped into diffrent transactions and it make the set doesn't look like a real transaction. So we want to integrate it into a real one. In some cases we will allocate some clusters and free some in just one transaction. e.g, one xattr is larger than inline size, so it and its value root is stored within the inode while the value is outside in a cluster. Then we try to update it with a smaller value(larger than the size of root but smaller than inline size), we may need to free the outside cluster while allocate a new bucket(one cluster) since now the inode may be full. The old solution will lock the global_bitmap(if the local alloc failed in stress test) and then the truncate log. This will cause a ABBA lock with truncate log flush. This patch add the clusters free in dealloc_ctxt, so that we can record the free clusters during the transaction and then free it after we release the global_bitmap in xattr set. Signed-off-by: Tao Ma <tao.ma@oracle.com> Signed-off-by: Mark Fasheh <mfasheh@suse.com>
Diffstat (limited to 'fs')
-rw-r--r--fs/ocfs2/alloc.c106
-rw-r--r--fs/ocfs2/alloc.h4
2 files changed, 103 insertions, 7 deletions
diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c
index 0cc2deb9394..4614614084d 100644
--- a/fs/ocfs2/alloc.c
+++ b/fs/ocfs2/alloc.c
@@ -5800,7 +5800,10 @@ int ocfs2_truncate_log_init(struct ocfs2_super *osb)
*/
/*
- * Describes a single block free from a suballocator
+ * Describe a single bit freed from a suballocator. For the block
+ * suballocators, it represents one block. For the global cluster
+ * allocator, it represents some clusters and free_bit indicates
+ * clusters number.
*/
struct ocfs2_cached_block_free {
struct ocfs2_cached_block_free *free_next;
@@ -5815,10 +5818,10 @@ struct ocfs2_per_slot_free_list {
struct ocfs2_cached_block_free *f_first;
};
-static int ocfs2_free_cached_items(struct ocfs2_super *osb,
- int sysfile_type,
- int slot,
- struct ocfs2_cached_block_free *head)
+static int ocfs2_free_cached_blocks(struct ocfs2_super *osb,
+ int sysfile_type,
+ int slot,
+ struct ocfs2_cached_block_free *head)
{
int ret;
u64 bg_blkno;
@@ -5893,6 +5896,82 @@ out:
return ret;
}
+int ocfs2_cache_cluster_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt,
+ u64 blkno, unsigned int bit)
+{
+ int ret = 0;
+ struct ocfs2_cached_block_free *item;
+
+ item = kmalloc(sizeof(*item), GFP_NOFS);
+ if (item == NULL) {
+ ret = -ENOMEM;
+ mlog_errno(ret);
+ return ret;
+ }
+
+ mlog(0, "Insert clusters: (bit %u, blk %llu)\n",
+ bit, (unsigned long long)blkno);
+
+ item->free_blk = blkno;
+ item->free_bit = bit;
+ item->free_next = ctxt->c_global_allocator;
+
+ ctxt->c_global_allocator = item;
+ return ret;
+}
+
+static int ocfs2_free_cached_clusters(struct ocfs2_super *osb,
+ struct ocfs2_cached_block_free *head)
+{
+ struct ocfs2_cached_block_free *tmp;
+ struct inode *tl_inode = osb->osb_tl_inode;
+ handle_t *handle;
+ int ret = 0;
+
+ mutex_lock(&tl_inode->i_mutex);
+
+ while (head) {
+ if (ocfs2_truncate_log_needs_flush(osb)) {
+ ret = __ocfs2_flush_truncate_log(osb);
+ if (ret < 0) {
+ mlog_errno(ret);
+ break;
+ }
+ }
+
+ handle = ocfs2_start_trans(osb, OCFS2_TRUNCATE_LOG_UPDATE);
+ if (IS_ERR(handle)) {
+ ret = PTR_ERR(handle);
+ mlog_errno(ret);
+ break;
+ }
+
+ ret = ocfs2_truncate_log_append(osb, handle, head->free_blk,
+ head->free_bit);
+
+ ocfs2_commit_trans(osb, handle);
+ tmp = head;
+ head = head->free_next;
+ kfree(tmp);
+
+ if (ret < 0) {
+ mlog_errno(ret);
+ break;
+ }
+ }
+
+ mutex_unlock(&tl_inode->i_mutex);
+
+ while (head) {
+ /* Premature exit may have left some dangling items. */
+ tmp = head;
+ head = head->free_next;
+ kfree(tmp);
+ }
+
+ return ret;
+}
+
int ocfs2_run_deallocs(struct ocfs2_super *osb,
struct ocfs2_cached_dealloc_ctxt *ctxt)
{
@@ -5908,8 +5987,10 @@ int ocfs2_run_deallocs(struct ocfs2_super *osb,
if (fl->f_first) {
mlog(0, "Free items: (type %u, slot %d)\n",
fl->f_inode_type, fl->f_slot);
- ret2 = ocfs2_free_cached_items(osb, fl->f_inode_type,
- fl->f_slot, fl->f_first);
+ ret2 = ocfs2_free_cached_blocks(osb,
+ fl->f_inode_type,
+ fl->f_slot,
+ fl->f_first);
if (ret2)
mlog_errno(ret2);
if (!ret)
@@ -5920,6 +6001,17 @@ int ocfs2_run_deallocs(struct ocfs2_super *osb,
kfree(fl);
}
+ if (ctxt->c_global_allocator) {
+ ret2 = ocfs2_free_cached_clusters(osb,
+ ctxt->c_global_allocator);
+ if (ret2)
+ mlog_errno(ret2);
+ if (!ret)
+ ret = ret2;
+
+ ctxt->c_global_allocator = NULL;
+ }
+
return ret;
}
diff --git a/fs/ocfs2/alloc.h b/fs/ocfs2/alloc.h
index 70257c84cfb..c301cf225f0 100644
--- a/fs/ocfs2/alloc.h
+++ b/fs/ocfs2/alloc.h
@@ -167,11 +167,15 @@ int __ocfs2_flush_truncate_log(struct ocfs2_super *osb);
*/
struct ocfs2_cached_dealloc_ctxt {
struct ocfs2_per_slot_free_list *c_first_suballocator;
+ struct ocfs2_cached_block_free *c_global_allocator;
};
static inline void ocfs2_init_dealloc_ctxt(struct ocfs2_cached_dealloc_ctxt *c)
{
c->c_first_suballocator = NULL;
+ c->c_global_allocator = NULL;
}
+int ocfs2_cache_cluster_dealloc(struct ocfs2_cached_dealloc_ctxt *ctxt,
+ u64 blkno, unsigned int bit);
int ocfs2_run_deallocs(struct ocfs2_super *osb,
struct ocfs2_cached_dealloc_ctxt *ctxt);