[PATCH] ceph: avoid pending truncate during nonblocking cap acquire

[Max Kellermann]

Readahead acquires caps while the folios are locked.  This can
deadlock when processing a pending truncate; truncate_pagecache() can
then attempt to a folio that is already locked.

Return -EAGAIN instead when `NON_BLOCKING` is set, causing the
readahead to be abandoned.

Fixes: 49870056005c ("ceph: convert ceph_readpages to ceph_readahead")
Cc: stable@xxxxxxxxxxxxxxx
Signed-off-by: Max Kellermann <max.kellermann@xxxxxxxxx>
---
 fs/ceph/caps.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/fs/ceph/caps.c b/fs/ceph/caps.c
index bcb04c6cb92c..86e195c5fd32 100644
--- a/fs/ceph/caps.c
+++ b/fs/ceph/caps.c
@@ -2889,6 +2889,10 @@ static int try_get_cap_refs(struct inode *inode, int need, int want,
 	}
 
 	/* finish pending truncate */
+	if (ci->i_truncate_pending && (flags & NON_BLOCKING)) {
+		ret = -EAGAIN;
+		goto out_unlock;
+	}
 	while (ci->i_truncate_pending) {
 		spin_unlock(&ci->i_ceph_lock);
 		if (snap_rwsem_locked) {
-- 
2.47.3