Commit 3c4aa443 authored by Linus Torvalds's avatar Linus Torvalds
Browse files

Merge tag 'ceph-for-6.4-rc1' of https://github.com/ceph/ceph-client 

Pull ceph updates from Ilya Dryomov:
 "A few filesystem improvements, with a rather nasty use-after-free fix
  from Xiubo intended for stable"

* tag 'ceph-for-6.4-rc1' of https://github.com/ceph/ceph-client:
  ceph: reorder fields in 'struct ceph_snapid_map'
  ceph: pass ino# instead of old_dentry if it's disconnected
  ceph: fix potential use-after-free bug when trimming caps
  ceph: implement writeback livelock avoidance using page tagging
  ceph: do not print the whole xattr value if it's too long
parents 8e15605b db2993a4

fs/ceph/addr.c

+10 −1
Original line number Diff line number Diff line
@@ -808,6 +808,7 @@ static int ceph_writepages_start(struct address_space *mapping, 
	bool should_loop, range_whole = false;

	bool done = false;

	bool caching = ceph_is_cache_enabled(inode);

	xa_mark_t tag;



	if (wbc->sync_mode == WB_SYNC_NONE &&

	    fsc->write_congested)
@@ -834,6 +835,11 @@ static int ceph_writepages_start(struct address_space *mapping, 
	start_index = wbc->range_cyclic ? mapping->writeback_index : 0;

	index = start_index;



	if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages) {

		tag = PAGECACHE_TAG_TOWRITE;

	} else {

		tag = PAGECACHE_TAG_DIRTY;

	}

retry:

	/* find oldest snap context with dirty data */

	snapc = get_oldest_context(inode, &ceph_wbc, NULL);
@@ -872,6 +878,9 @@ static int ceph_writepages_start(struct address_space *mapping, 
		dout(" non-head snapc, range whole\n");

	}



	if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages)

		tag_pages_for_writeback(mapping, index, end);



	ceph_put_snap_context(last_snapc);

	last_snapc = snapc;
@@ -888,7 +897,7 @@ static int ceph_writepages_start(struct address_space *mapping, 


get_more_pages:

		nr_folios = filemap_get_folios_tag(mapping, &index,

				end, PAGECACHE_TAG_DIRTY, &fbatch);

						   end, tag, &fbatch);

		dout("pagevec_lookup_range_tag got %d\n", nr_folios);

		if (!nr_folios && !locked_pages)

			break;

fs/ceph/caps.c

+1 −1
Original line number Diff line number Diff line
@@ -431,7 +431,7 @@ void ceph_reservation_status(struct ceph_fs_client *fsc, 
 *

 * Called with i_ceph_lock held.

 */

static struct ceph_cap *__get_cap_for_mds(struct ceph_inode_info *ci, int mds)

struct ceph_cap *__get_cap_for_mds(struct ceph_inode_info *ci, int mds)

{

	struct ceph_cap *cap;

	struct rb_node *n = ci->i_caps.rb_node;

fs/ceph/debugfs.c

+12 −6
Original line number Diff line number Diff line
@@ -248,14 +248,20 @@ static int metrics_caps_show(struct seq_file *s, void *p) 
	return 0;

}



static int caps_show_cb(struct inode *inode, struct ceph_cap *cap, void *p)

static int caps_show_cb(struct inode *inode, int mds, void *p)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	struct seq_file *s = p;

	struct ceph_cap *cap;



	spin_lock(&ci->i_ceph_lock);

	cap = __get_cap_for_mds(ci, mds);

	if (cap)

		seq_printf(s, "0x%-17llx%-3d%-17s%-17s\n", ceph_ino(inode),

			   cap->session->s_mds,

			   ceph_cap_string(cap->issued),

			   ceph_cap_string(cap->implemented));

	spin_unlock(&ci->i_ceph_lock);

	return 0;

}

fs/ceph/dir.c

+11 −2
Original line number Diff line number Diff line
@@ -1050,6 +1050,9 @@ static int ceph_link(struct dentry *old_dentry, struct inode *dir, 
	struct ceph_mds_request *req;

	int err;



	if (dentry->d_flags & DCACHE_DISCONNECTED)

		return -EINVAL;



	err = ceph_wait_on_conflict_unlink(dentry);

	if (err)

		return err;
@@ -1057,8 +1060,8 @@ static int ceph_link(struct dentry *old_dentry, struct inode *dir, 
	if (ceph_snap(dir) != CEPH_NOSNAP)

		return -EROFS;



	dout("link in dir %p old_dentry %p dentry %p\n", dir,

	     old_dentry, dentry);

	dout("link in dir %p %llx.%llx old_dentry %p:'%pd' dentry %p:'%pd'\n",

	     dir, ceph_vinop(dir), old_dentry, old_dentry, dentry, dentry);

	req = ceph_mdsc_create_request(mdsc, CEPH_MDS_OP_LINK, USE_AUTH_MDS);

	if (IS_ERR(req)) {

		d_drop(dentry);
@@ -1067,6 +1070,12 @@ static int ceph_link(struct dentry *old_dentry, struct inode *dir, 
	req->r_dentry = dget(dentry);

	req->r_num_caps = 2;

	req->r_old_dentry = dget(old_dentry);

	/*

	 * The old_dentry maybe a DCACHE_DISCONNECTED dentry, then we

	 * will just pass the ino# to MDSs.

	 */

	if (old_dentry->d_flags & DCACHE_DISCONNECTED)

		req->r_ino2 = ceph_vino(d_inode(old_dentry));

	req->r_parent = dir;

	ihold(dir);

	set_bit(CEPH_MDS_R_PARENT_LOCKED, &req->r_req_flags);

fs/ceph/mds_client.c

+51 −27
Original line number Diff line number Diff line
@@ -1632,8 +1632,8 @@ static void cleanup_session_requests(struct ceph_mds_client *mdsc, 
 * Caller must hold session s_mutex.

 */

int ceph_iterate_session_caps(struct ceph_mds_session *session,

			      int (*cb)(struct inode *, struct ceph_cap *,

					void *), void *arg)

			      int (*cb)(struct inode *, int mds, void *),

			      void *arg)

{

	struct list_head *p;

	struct ceph_cap *cap;
@@ -1645,6 +1645,8 @@ int ceph_iterate_session_caps(struct ceph_mds_session *session, 
	spin_lock(&session->s_cap_lock);

	p = session->s_caps.next;

	while (p != &session->s_caps) {

		int mds;



		cap = list_entry(p, struct ceph_cap, session_caps);

		inode = igrab(&cap->ci->netfs.inode);

		if (!inode) {
@@ -1652,6 +1654,7 @@ int ceph_iterate_session_caps(struct ceph_mds_session *session, 
			continue;

		}

		session->s_cap_iterator = cap;

		mds = cap->mds;

		spin_unlock(&session->s_cap_lock);



		if (last_inode) {
@@ -1663,7 +1666,7 @@ int ceph_iterate_session_caps(struct ceph_mds_session *session, 
			old_cap = NULL;

		}



		ret = cb(inode, cap, arg);

		ret = cb(inode, mds, arg);

		last_inode = inode;



		spin_lock(&session->s_cap_lock);
@@ -1696,19 +1699,24 @@ int ceph_iterate_session_caps(struct ceph_mds_session *session, 
	return ret;

}



static int remove_session_caps_cb(struct inode *inode, struct ceph_cap *cap,

				  void *arg)

static int remove_session_caps_cb(struct inode *inode, int mds, void *arg)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	bool invalidate = false;

	int iputs;

	struct ceph_cap *cap;

	int iputs = 0;



	spin_lock(&ci->i_ceph_lock);

	cap = __get_cap_for_mds(ci, mds);

	if (cap) {

		dout(" removing cap %p, ci is %p, inode is %p\n",

		     cap, ci, &ci->netfs.inode);

	spin_lock(&ci->i_ceph_lock);



		iputs = ceph_purge_inode_cap(inode, cap, &invalidate);

	}

	spin_unlock(&ci->i_ceph_lock);



	if (cap)

		wake_up_all(&ci->i_cap_wq);

	if (invalidate)

		ceph_queue_invalidate(inode);
@@ -1780,8 +1788,7 @@ enum { 
 *

 * caller must hold s_mutex.

 */

static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap,

			      void *arg)

static int wake_up_session_cb(struct inode *inode, int mds, void *arg)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	unsigned long ev = (unsigned long)arg;
@@ -1792,12 +1799,14 @@ static int wake_up_session_cb(struct inode *inode, struct ceph_cap *cap, 
		ci->i_requested_max_size = 0;

		spin_unlock(&ci->i_ceph_lock);

	} else if (ev == RENEWCAPS) {

		if (cap->cap_gen < atomic_read(&cap->session->s_cap_gen)) {

			/* mds did not re-issue stale cap */

		struct ceph_cap *cap;



		spin_lock(&ci->i_ceph_lock);

		cap = __get_cap_for_mds(ci, mds);

		/* mds did not re-issue stale cap */

		if (cap && cap->cap_gen < atomic_read(&cap->session->s_cap_gen))

			cap->issued = cap->implemented = CEPH_CAP_PIN;

		spin_unlock(&ci->i_ceph_lock);

		}

	} else if (ev == FORCE_RO) {

	}

	wake_up_all(&ci->i_cap_wq);
@@ -1959,16 +1968,22 @@ static bool drop_negative_children(struct dentry *dentry) 
 * Yes, this is a bit sloppy.  Our only real goal here is to respond to

 * memory pressure from the MDS, though, so it needn't be perfect.

 */

static int trim_caps_cb(struct inode *inode, struct ceph_cap *cap, void *arg)

static int trim_caps_cb(struct inode *inode, int mds, void *arg)

{

	int *remaining = arg;

	struct ceph_inode_info *ci = ceph_inode(inode);

	int used, wanted, oissued, mine;

	struct ceph_cap *cap;



	if (*remaining <= 0)

		return -1;



	spin_lock(&ci->i_ceph_lock);

	cap = __get_cap_for_mds(ci, mds);

	if (!cap) {

		spin_unlock(&ci->i_ceph_lock);

		return 0;

	}

	mine = cap->issued | cap->implemented;

	used = __ceph_caps_used(ci);

	wanted = __ceph_caps_file_wanted(ci);
@@ -2555,6 +2570,7 @@ static struct ceph_msg *create_request_message(struct ceph_mds_session *session, 
	u64 ino1 = 0, ino2 = 0;

	int pathlen1 = 0, pathlen2 = 0;

	bool freepath1 = false, freepath2 = false;

	struct dentry *old_dentry = NULL;

	int len;

	u16 releases;

	void *p, *end;
@@ -2572,7 +2588,10 @@ static struct ceph_msg *create_request_message(struct ceph_mds_session *session, 
	}



	/* If r_old_dentry is set, then assume that its parent is locked */

	ret = set_request_path_attr(NULL, req->r_old_dentry,

	if (req->r_old_dentry &&

	    !(req->r_old_dentry->d_flags & DCACHE_DISCONNECTED))

		old_dentry = req->r_old_dentry;

	ret = set_request_path_attr(NULL, old_dentry,

			      req->r_old_dentry_dir,

			      req->r_path2, req->r_ino2.ino,

			      &path2, &pathlen2, &ino2, &freepath2, true);
@@ -3911,26 +3930,22 @@ static struct dentry* d_find_primary(struct inode *inode) 
/*

 * Encode information about a cap for a reconnect with the MDS.

 */

static int reconnect_caps_cb(struct inode *inode, struct ceph_cap *cap,

			  void *arg)

static int reconnect_caps_cb(struct inode *inode, int mds, void *arg)

{

	union {

		struct ceph_mds_cap_reconnect v2;

		struct ceph_mds_cap_reconnect_v1 v1;

	} rec;

	struct ceph_inode_info *ci = cap->ci;

	struct ceph_inode_info *ci = ceph_inode(inode);

	struct ceph_reconnect_state *recon_state = arg;

	struct ceph_pagelist *pagelist = recon_state->pagelist;

	struct dentry *dentry;

	struct ceph_cap *cap;

	char *path;

	int pathlen = 0, err;

	int pathlen = 0, err = 0;

	u64 pathbase;

	u64 snap_follows;



	dout(" adding %p ino %llx.%llx cap %p %lld %s\n",

	     inode, ceph_vinop(inode), cap, cap->cap_id,

	     ceph_cap_string(cap->issued));



	dentry = d_find_primary(inode);

	if (dentry) {

		/* set pathbase to parent dir when msg_version >= 2 */
@@ -3947,6 +3962,15 @@ static int reconnect_caps_cb(struct inode *inode, struct ceph_cap *cap, 
	}



	spin_lock(&ci->i_ceph_lock);

	cap = __get_cap_for_mds(ci, mds);

	if (!cap) {

		spin_unlock(&ci->i_ceph_lock);

		goto out_err;

	}

	dout(" adding %p ino %llx.%llx cap %p %lld %s\n",

	     inode, ceph_vinop(inode), cap, cap->cap_id,

	     ceph_cap_string(cap->issued));



	cap->seq = 0;        /* reset cap seq */

	cap->issue_seq = 0;  /* and issue_seq */

	cap->mseq = 0;       /* and migrate_seq */