ext4: optimize ext4_should_retry_alloc() to improve ENOSPC performance

If there are no pending blocks to be released after a commit, forcing
a journal commit has no hope of helping.  It's possible that a commit
had just completed, so if there are now free blocks available for
allocation, it's worth retrying the commit.

Reported-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>

4 files changed, 23 insertions, 4 deletions

diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c
index 3020fd70c392..0b8105b3293d 100644
--- a/fs/ext4/balloc.c
+++ b/fs/ext4/balloc.c
@@ -610,7 +610,9 @@ int ext4_should_retry_alloc(struct super_block *sb, int *retries)
 
         jbd_debug(1, "%s: retrying operation after ENOSPC\n", sb->s_id);
 
-        jbd2_journal_force_commit_nested(EXT4_SB(sb)->s_journal);
+        smp_mb();
+        if (EXT4_SB(sb)->s_mb_free_pending)
+                jbd2_journal_force_commit_nested(EXT4_SB(sb)->s_journal);
         return 1;
 }
 
diff --git a/fs/ext4/ext4.h b/fs/ext4/ext4.h
index b84aa1ca480a..96c73e6fec6e 100644
--- a/fs/ext4/ext4.h
+++ b/fs/ext4/ext4.h
@@ -1430,6 +1430,7 @@ struct ext4_sb_info {
         unsigned short *s_mb_offsets;
         unsigned int *s_mb_maxs;
         unsigned int s_group_info_size;
+        unsigned int s_mb_free_pending;
 
         /* tunables */
         unsigned long s_stripe;
diff --git a/fs/ext4/ext4_jbd2.h b/fs/ext4/ext4_jbd2.h
index 09c1ef38cbe6..b1d52c14098e 100644
--- a/fs/ext4/ext4_jbd2.h
+++ b/fs/ext4/ext4_jbd2.h
@@ -175,6 +175,13 @@ struct ext4_journal_cb_entry {
  * There is no guaranteed calling order of multiple registered callbacks on
  * the same transaction.
  */
+static inline void _ext4_journal_callback_add(handle_t *handle,
+                        struct ext4_journal_cb_entry *jce)
+{
+        /* Add the jce to transaction's private list */
+        list_add_tail(&jce->jce_list, &handle->h_transaction->t_private_list);
+}
+
 static inline void ext4_journal_callback_add(handle_t *handle,
                         void (*func)(struct super_block *sb,
                                      struct ext4_journal_cb_entry *jce,
@@ -187,10 +194,11 @@ static inline void ext4_journal_callback_add(handle_t *handle,
         /* Add the jce to transaction's private list */
         jce->jce_func = func;
         spin_lock(&sbi->s_md_lock);
-        list_add_tail(&jce->jce_list, &handle->h_transaction->t_private_list);
+        _ext4_journal_callback_add(handle, jce);
         spin_unlock(&sbi->s_md_lock);
 }
 
+
 /**
  * ext4_journal_callback_del: delete a registered callback
  * @handle: active journal transaction handle on which callback was registered
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index c1ab3ec30423..77249e1f5c3a 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2627,6 +2627,7 @@ int ext4_mb_init(struct super_block *sb)
 
         spin_lock_init(&sbi->s_md_lock);
         spin_lock_init(&sbi->s_bal_lock);
+        sbi->s_mb_free_pending = 0;
 
         sbi->s_mb_max_to_scan = MB_DEFAULT_MAX_TO_SCAN;
         sbi->s_mb_min_to_scan = MB_DEFAULT_MIN_TO_SCAN;
@@ -2814,6 +2815,9 @@ static void ext4_free_data_callback(struct super_block *sb,
         /* we expect to find existing buddy because it's pinned */
         BUG_ON(err != 0);
 
+        spin_lock(&EXT4_SB(sb)->s_md_lock);
+        EXT4_SB(sb)->s_mb_free_pending -= entry->efd_count;
+        spin_unlock(&EXT4_SB(sb)->s_md_lock);
 
         db = e4b.bd_info;
         /* there are blocks to put in buddy to make them really free */
@@ -4583,6 +4587,7 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
 {
         ext4_group_t group = e4b->bd_group;
         ext4_grpblk_t cluster;
+        ext4_grpblk_t clusters = new_entry->efd_count;
         struct ext4_free_data *entry;
         struct ext4_group_info *db = e4b->bd_info;
         struct super_block *sb = e4b->bd_sb;
@@ -4649,8 +4654,11 @@ ext4_mb_free_metadata(handle_t *handle, struct ext4_buddy *e4b,
                 }
         }
         /* Add the extent to transaction's private list */
-        ext4_journal_callback_add(handle, ext4_free_data_callback,
-                                  &new_entry->efd_jce);
+        new_entry->efd_jce.jce_func = ext4_free_data_callback;
+        spin_lock(&sbi->s_md_lock);
+        _ext4_journal_callback_add(handle, &new_entry->efd_jce);
+        sbi->s_mb_free_pending += clusters;
+        spin_unlock(&sbi->s_md_lock);
         return 0;
 }