[PATCH 3/3] FS-Cache: refcount becomes corrupt under vma preasure.

From: Milosz Tanski
Date: Tue Jul 22 2014 - 11:51:17 EST


In rare cases under heavy VMA preasure the ref count for fscache cookies
becomes corrupt. In this case we decrement ref count even if fail before
incrememnt the refcount.

FS-Cache: Assertion failed bnode-eca5f9c6/syslog
0 > 0 is false
------------[ cut here ]------------
kernel BUG at fs/fscache/cookie.c:519!
invalid opcode: 0000 [#1] SMP
Call Trace:
[<ffffffffa01ba060>] __fscache_relinquish_cookie+0x50/0x220 [fscache]
[<ffffffffa02d64ce>] ceph_fscache_unregister_inode_cookie+0x3e/0x50 [ceph]
[<ffffffffa02ae1d3>] ceph_destroy_inode+0x33/0x200 [ceph]
[<ffffffff811cf67e>] ? __fsnotify_inode_delete+0xe/0x10
[<ffffffff811a9e0c>] destroy_inode+0x3c/0x70
[<ffffffff811a9f51>] evict+0x111/0x180
[<ffffffff811aa763>] iput+0x103/0x190
[<ffffffff811a5de8>] __dentry_kill+0x1c8/0x220
[<ffffffff811a5f31>] shrink_dentry_list+0xf1/0x250
[<ffffffff811a762c>] prune_dcache_sb+0x4c/0x60
[<ffffffff811930af>] super_cache_scan+0xff/0x170
[<ffffffff8113d7a0>] shrink_slab_node+0x140/0x2c0
[<ffffffff8113f2da>] shrink_slab+0x8a/0x130
[<ffffffff81142572>] balance_pgdat+0x3e2/0x5d0
[<ffffffff811428ca>] kswapd+0x16a/0x4a0
[<ffffffff810a43f0>] ? __wake_up_sync+0x20/0x20
[<ffffffff81142760>] ? balance_pgdat+0x5d0/0x5d0
[<ffffffff81083e09>] kthread+0xc9/0xe0
[<ffffffff81010000>] ? ftrace_raw_event_xen_mmu_release_ptpage+0x70/0x90
[<ffffffff81083d40>] ? flush_kthread_worker+0xb0/0xb0
[<ffffffff8159f63c>] ret_from_fork+0x7c/0xb0
[<ffffffff81083d40>] ? flush_kthread_worker+0xb0/0xb0
RIP [<ffffffffa01b984b>] __fscache_disable_cookie+0x1db/0x210 [fscache]
RSP <ffff8803bc85f9b8>
---[ end trace 254d0d7c74a01f25 ]---
---
fs/fscache/page.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/fs/fscache/page.c b/fs/fscache/page.c
index 84e11aa..c83dd89 100644
--- a/fs/fscache/page.c
+++ b/fs/fscache/page.c
@@ -195,7 +195,7 @@ int __fscache_attr_changed(struct fscache_cookie *cookie)
{
struct fscache_operation *op;
struct fscache_object *object;
- bool wake_cookie;
+ bool wake_cookie = false;

_enter("%p", cookie);

@@ -225,15 +225,16 @@ int __fscache_attr_changed(struct fscache_cookie *cookie)

__fscache_use_cookie(cookie);
if (fscache_submit_exclusive_op(object, op) < 0)
- goto nobufs;
+ goto nobufs_dec;
spin_unlock(&cookie->lock);
fscache_stat(&fscache_n_attr_changed_ok);
fscache_put_operation(op);
_leave(" = 0");
return 0;

-nobufs:
+nobufs_dec:
wake_cookie = __fscache_unuse_cookie(cookie);
+nobufs:
spin_unlock(&cookie->lock);
kfree(op);
if (wake_cookie)
--
1.7.9.5



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/