lkml.org 
[lkml]   [2019]   [Oct]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH 5.3 084/112] btrfs: allocate new inode in NOFS context
Date
From: Josef Bacik <josef@toxicpanda.com>

commit 11a19a90870ea5496a8ded69b86f5b476b6d3355 upstream.

A user reported a lockdep splat

======================================================
WARNING: possible circular locking dependency detected
5.2.11-gentoo #2 Not tainted
------------------------------------------------------
kswapd0/711 is trying to acquire lock:
000000007777a663 (sb_internal){.+.+}, at: start_transaction+0x3a8/0x500

but task is already holding lock:
000000000ba86300 (fs_reclaim){+.+.}, at: __fs_reclaim_acquire+0x0/0x30

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #1 (fs_reclaim){+.+.}:
kmem_cache_alloc+0x1f/0x1c0
btrfs_alloc_inode+0x1f/0x260
alloc_inode+0x16/0xa0
new_inode+0xe/0xb0
btrfs_new_inode+0x70/0x610
btrfs_symlink+0xd0/0x420
vfs_symlink+0x9c/0x100
do_symlinkat+0x66/0xe0
do_syscall_64+0x55/0x1c0
entry_SYSCALL_64_after_hwframe+0x49/0xbe

-> #0 (sb_internal){.+.+}:
__sb_start_write+0xf6/0x150
start_transaction+0x3a8/0x500
btrfs_commit_inode_delayed_inode+0x59/0x110
btrfs_evict_inode+0x19e/0x4c0
evict+0xbc/0x1f0
inode_lru_isolate+0x113/0x190
__list_lru_walk_one.isra.4+0x5c/0x100
list_lru_walk_one+0x32/0x50
prune_icache_sb+0x36/0x80
super_cache_scan+0x14a/0x1d0
do_shrink_slab+0x131/0x320
shrink_node+0xf7/0x380
balance_pgdat+0x2d5/0x640
kswapd+0x2ba/0x5e0
kthread+0x147/0x160
ret_from_fork+0x24/0x30

other info that might help us debug this:

Possible unsafe locking scenario:

CPU0 CPU1
---- ----
lock(fs_reclaim);
lock(sb_internal);
lock(fs_reclaim);
lock(sb_internal);
---
fs/btrfs/inode.c | 3 +++
1 file changed, 3 insertions(+)

--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -6276,13 +6276,16 @@ static struct inode *btrfs_new_inode(str
u32 sizes[2];
int nitems = name ? 2 : 1;
unsigned long ptr;
+ unsigned int nofs_flag;
int ret;

path = btrfs_alloc_path();
if (!path)
return ERR_PTR(-ENOMEM);

+ nofs_flag = memalloc_nofs_save();
inode = new_inode(fs_info->sb);
+ memalloc_nofs_restore(nofs_flag);
if (!inode) {
btrfs_free_path(path);
return ERR_PTR(-ENOMEM);

\
 
 \ /
  Last update: 2019-10-17 00:00    [W:0.267 / U:4.724 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site