summaryrefslogtreecommitdiff
path: root/fs/ocfs2/journal.c
diff options
context:
space:
mode:
authorSunil Mushran <sunil.mushran@oracle.com>2011-07-24 21:35:54 +0400
committerSunil Mushran <sunil.mushran@oracle.com>2011-07-24 21:35:54 +0400
commita035bff6b82aca89c1223e2c614adc2d17ec8aa2 (patch)
treeedc27e463547909288f4f28438cc7e6104c125f6 /fs/ocfs2/journal.c
parent619c200de144b44f5e405305241bcd7edbb8c6cf (diff)
downloadlinux-a035bff6b82aca89c1223e2c614adc2d17ec8aa2.tar.xz
ocfs2: Add comment about orphan scanning
Add a comment that explains the reason as to why orphan scan scans all the slots. Signed-off-by: Sunil Mushran <sunil.mushran@oracle.com>
Diffstat (limited to 'fs/ocfs2/journal.c')
-rw-r--r--fs/ocfs2/journal.c14
1 files changed, 14 insertions, 0 deletions
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index 8eaaa7854409..0a42ae96dca7 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -1811,6 +1811,20 @@ static inline unsigned long ocfs2_orphan_scan_timeout(void)
* every slot, queuing a recovery of the slot on the ocfs2_wq thread. This
* is done to catch any orphans that are left over in orphan directories.
*
+ * It scans all slots, even ones that are in use. It does so to handle the
+ * case described below:
+ *
+ * Node 1 has an inode it was using. The dentry went away due to memory
+ * pressure. Node 1 closes the inode, but it's on the free list. The node
+ * has the open lock.
+ * Node 2 unlinks the inode. It grabs the dentry lock to notify others,
+ * but node 1 has no dentry and doesn't get the message. It trylocks the
+ * open lock, sees that another node has a PR, and does nothing.
+ * Later node 2 runs its orphan dir. It igets the inode, trylocks the
+ * open lock, sees the PR still, and does nothing.
+ * Basically, we have to trigger an orphan iput on node 1. The only way
+ * for this to happen is if node 1 runs node 2's orphan dir.
+ *
* ocfs2_queue_orphan_scan gets called every ORPHAN_SCAN_SCHEDULE_TIMEOUT
* seconds. It gets an EX lock on os_lockres and checks sequence number
* stored in LVB. If the sequence number has changed, it means some other