From: "Yan, Zheng" <zyan@redhat.com>
Date: Mon, 28 Aug 2017 15:41:28 +0800
Subject: ceph: handle race between vmtruncate and queuing cap snap
Git-commit: c8fd0d37f81dd38e3f319f4938b45a5aaf0dfc58
Patch-mainline: v4.14-rc1
References: FATE#324714
It's possible that we create a cap snap while there is pending
vmtruncate (truncate hasn't been processed by worker thread).
We should truncate dirty pages beyond capsnap->size in that case.
Signed-off-by: "Yan, Zheng" <zyan@redhat.com>
Signed-off-by: Ilya Dryomov <idryomov@gmail.com>
Acked-by: Luis Henriques <lhenriques@suse.com>
---
fs/ceph/inode.c | 13 ++++++++++++-
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c
index a19fafdf87f8..373dab5173ca 100644
--- a/fs/ceph/inode.c
+++ b/fs/ceph/inode.c
@@ -1833,9 +1833,20 @@ void __ceph_do_pending_vmtruncate(struct inode *inode)
* possibly truncate them.. so write AND block!
*/
if (ci->i_wrbuffer_ref_head < ci->i_wrbuffer_ref) {
+ struct ceph_cap_snap *capsnap;
+ to = ci->i_truncate_size;
+ list_for_each_entry(capsnap, &ci->i_cap_snaps, ci_item) {
+ // MDS should have revoked Frw caps
+ WARN_ON_ONCE(capsnap->writing);
+ if (capsnap->dirty_pages && capsnap->size > to)
+ to = capsnap->size;
+ }
+ spin_unlock(&ci->i_ceph_lock);
dout("__do_pending_vmtruncate %p flushing snaps first\n",
inode);
- spin_unlock(&ci->i_ceph_lock);
+
+ truncate_pagecache(inode, to);
+
filemap_write_and_wait_range(&inode->i_data, 0,
inode->i_sb->s_maxbytes);
goto retry;