Merge tag 'ceph-for-6.5-rc1' of https://github.com/ceph/ceph-client

Pull ceph updates from Ilya Dryomov:
 "A bunch of CephFS fixups from Xiubo, mostly around dropping caps,
  along with a fix for a regression in the readahead handling code which
  sneaked in with the switch to netfs helpers"

* tag 'ceph-for-6.5-rc1' of https://github.com/ceph/ceph-client:
  ceph: don't let check_caps skip sending responses for revoke msgs
  ceph: issue a cap release immediately if no cap exists
  ceph: trigger to flush the buffer when making snapshot
  ceph: fix blindly expanding the readahead windows
  ceph: add a dedicated private data for netfs rreq
  ceph: voluntarily drop Xx caps for requests those touch parent mtime
  ceph: try to dump the msgs when decoding fails
  ceph: only send metrics when the MDS rank is ready

1 files changed, 67 insertions, 18 deletions

diff --git a/fs/ceph/addr.c b/fs/ceph/addr.c
index 6bb251a4d613..59cbfb80edbd 100644
--- a/fs/ceph/addr.c
+++ b/fs/ceph/addr.c
@@ -187,16 +187,42 @@ static void ceph_netfs_expand_readahead(struct netfs_io_request *rreq)
 	struct inode *inode = rreq->inode;
 	struct ceph_inode_info *ci = ceph_inode(inode);
 	struct ceph_file_layout *lo = &ci->i_layout;
+	unsigned long max_pages = inode->i_sb->s_bdi->ra_pages;
+	loff_t end = rreq->start + rreq->len, new_end;
+	struct ceph_netfs_request_data *priv = rreq->netfs_priv;
+	unsigned long max_len;
 	u32 blockoff;
-	u64 blockno;
 
-	/* Expand the start downward */
-	blockno = div_u64_rem(rreq->start, lo->stripe_unit, &blockoff);
-	rreq->start = blockno * lo->stripe_unit;
-	rreq->len += blockoff;
+	if (priv) {
+		/* Readahead is disabled by posix_fadvise POSIX_FADV_RANDOM */
+		if (priv->file_ra_disabled)
+			max_pages = 0;
+		else
+			max_pages = priv->file_ra_pages;
+
+	}
 
-	/* Now, round up the length to the next block */
-	rreq->len = roundup(rreq->len, lo->stripe_unit);
+	/* Readahead is disabled */
+	if (!max_pages)
+		return;
+
+	max_len = max_pages << PAGE_SHIFT;
+
+	/*
+	 * Try to expand the length forward by rounding up it to the next
+	 * block, but do not exceed the file size, unless the original
+	 * request already exceeds it.
+	 */
+	new_end = min(round_up(end, lo->stripe_unit), rreq->i_size);
+	if (new_end > end && new_end <= rreq->start + max_len)
+		rreq->len = new_end - rreq->start;
+
+	/* Try to expand the start downward */
+	div_u64_rem(rreq->start, lo->stripe_unit, &blockoff);
+	if (rreq->len + blockoff <= max_len) {
+		rreq->start -= blockoff;
+		rreq->len += blockoff;
+	}
 }
 
 static bool ceph_netfs_clamp_length(struct netfs_io_subrequest *subreq)
@@ -362,18 +388,28 @@ static int ceph_init_request(struct netfs_io_request *rreq, struct file *file)
 {
 	struct inode *inode = rreq->inode;
 	int got = 0, want = CEPH_CAP_FILE_CACHE;
+	struct ceph_netfs_request_data *priv;
 	int ret = 0;
 
 	if (rreq->origin != NETFS_READAHEAD)
 		return 0;
 
+	priv = kzalloc(sizeof(*priv), GFP_NOFS);
+	if (!priv)
+		return -ENOMEM;
+
 	if (file) {
 		struct ceph_rw_context *rw_ctx;
 		struct ceph_file_info *fi = file->private_data;
 
+		priv->file_ra_pages = file->f_ra.ra_pages;
+		priv->file_ra_disabled = file->f_mode & FMODE_RANDOM;
+
 		rw_ctx = ceph_find_rw_context(fi);
-		if (rw_ctx)
+		if (rw_ctx) {
+			rreq->netfs_priv = priv;
 			return 0;
+		}
 	}
 
 	/*
@@ -383,27 +419,40 @@ static int ceph_init_request(struct netfs_io_request *rreq, struct file *file)
 	ret = ceph_try_get_caps(inode, CEPH_CAP_FILE_RD, want, true, &got);
 	if (ret < 0) {
 		dout("start_read %p, error getting cap\n", inode);
-		return ret;
+		goto out;
 	}
 
 	if (!(got & want)) {
 		dout("start_read %p, no cache cap\n", inode);
-		return -EACCES;
+		ret = -EACCES;
+		goto out;
+	}
+	if (ret == 0) {
+		ret = -EACCES;
+		goto out;
 	}
-	if (ret == 0)
-		return -EACCES;
 
-	rreq->netfs_priv = (void *)(uintptr_t)got;
-	return 0;
+	priv->caps = got;
+	rreq->netfs_priv = priv;
+
+out:
+	if (ret < 0)
+		kfree(priv);
+
+	return ret;
 }
 
 static void ceph_netfs_free_request(struct netfs_io_request *rreq)
 {
-	struct ceph_inode_info *ci = ceph_inode(rreq->inode);
-	int got = (uintptr_t)rreq->netfs_priv;
+	struct ceph_netfs_request_data *priv = rreq->netfs_priv;
+
+	if (!priv)
+		return;
 
-	if (got)
-		ceph_put_cap_refs(ci, got);
+	if (priv->caps)
+		ceph_put_cap_refs(ceph_inode(rreq->inode), priv->caps);
+	kfree(priv);
+	rreq->netfs_priv = NULL;
 }
 
 const struct netfs_request_ops ceph_netfs_ops = {