ext4_free_blocks is looping around the allocation request and mimics
__GFP_NOFAIL behavior without any allocation fallback strategy. Let's
remove the open coded loop and replace it with __GFP_NOFAIL. Without
the flag the allocator has no way to find out never-fail requirement
and cannot help in any way.
Signed-off-by: Michal Hocko <[email protected]>
---
Hi Ted,
I am sorry for seding these changes one at the time but I haven't found
a proper way to find all of them automatically. So I am discovering
them as I check the code due to other reasons.
Thanks!
fs/ext4/mballoc.c | 16 +++++-----------
1 file changed, 5 insertions(+), 11 deletions(-)
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index 8d1e60214ef0..41260489d3bc 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -4800,18 +4800,12 @@ void ext4_free_blocks(handle_t *handle, struct inode *inode,
/*
* blocks being freed are metadata. these blocks shouldn't
* be used until this transaction is committed
+ *
+ * We use __GFP_NOFAIL because ext4_free_blocks() is not allowed
+ * to fail.
*/
- retry:
- new_entry = kmem_cache_alloc(ext4_free_data_cachep, GFP_NOFS);
- if (!new_entry) {
- /*
- * We use a retry loop because
- * ext4_free_blocks() is not allowed to fail.
- */
- cond_resched();
- congestion_wait(BLK_RW_ASYNC, HZ/50);
- goto retry;
- }
+ new_entry = kmem_cache_alloc(ext4_free_data_cachep,
+ GFP_NOFS|__GFP_NOFAIL);
new_entry->efd_start_cluster = bit;
new_entry->efd_group = block_group;
new_entry->efd_count = count_clusters;
--
2.1.4
On Tue, Jun 23, 2015 at 11:50:37AM +0200, Michal Hocko wrote:
> ext4_free_blocks is looping around the allocation request and mimics
> __GFP_NOFAIL behavior without any allocation fallback strategy. Let's
> remove the open coded loop and replace it with __GFP_NOFAIL. Without
> the flag the allocator has no way to find out never-fail requirement
> and cannot help in any way.
>
> Signed-off-by: Michal Hocko <[email protected]>
Thanks, applied.
- Ted