Commit 9a5530c6 authored by Yan, Zheng's avatar Yan, Zheng Committed by Ilya Dryomov
Browse files

ceph: wait unsafe sync writes for evicting inode 



Otherwise ceph_sync_write_unsafe() may access/modify freed inode.

Signed-off-by: default avatarYan, Zheng <zyan@redhat.com>
parent fc8c3892

fs/ceph/caps.c

+2 −48
Original line number Diff line number Diff line
@@ -1927,53 +1927,6 @@ static int caps_are_flushed(struct inode *inode, u64 flush_tid) 
	return ret;

}



/*

 * Wait on any unsafe replies for the given inode.  First wait on the

 * newest request, and make that the upper bound.  Then, if there are

 * more requests, keep waiting on the oldest as long as it is still older

 * than the original request.

 */

static void sync_write_wait(struct inode *inode)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	struct list_head *head = &ci->i_unsafe_writes;

	struct ceph_osd_request *req;

	u64 last_tid;



	if (!S_ISREG(inode->i_mode))

		return;



	spin_lock(&ci->i_unsafe_lock);

	if (list_empty(head))

		goto out;



	/* set upper bound as _last_ entry in chain */

	req = list_last_entry(head, struct ceph_osd_request,

			      r_unsafe_item);

	last_tid = req->r_tid;



	do {

		ceph_osdc_get_request(req);

		spin_unlock(&ci->i_unsafe_lock);

		dout("sync_write_wait on tid %llu (until %llu)\n",

		     req->r_tid, last_tid);

		wait_for_completion(&req->r_safe_completion);

		spin_lock(&ci->i_unsafe_lock);

		ceph_osdc_put_request(req);



		/*

		 * from here on look at first entry in chain, since we

		 * only want to wait for anything older than last_tid

		 */

		if (list_empty(head))

			break;

		req = list_first_entry(head, struct ceph_osd_request,

				       r_unsafe_item);

	} while (req->r_tid < last_tid);

out:

	spin_unlock(&ci->i_unsafe_lock);

}



/*

 * wait for any unsafe requests to complete.

 */
@@ -2026,7 +1979,8 @@ int ceph_fsync(struct file *file, loff_t start, loff_t end, int datasync) 
	int dirty;



	dout("fsync %p%s\n", inode, datasync ? " datasync" : "");

	sync_write_wait(inode);



	ceph_sync_write_wait(inode);



	ret = filemap_write_and_wait_range(inode->i_mapping, start, end);

	if (ret < 0)

fs/ceph/file.c

+48 −0
Original line number Diff line number Diff line
@@ -821,6 +821,54 @@ static void ceph_sync_write_unsafe(struct ceph_osd_request *req, bool unsafe) 
	}

}



/*

 * Wait on any unsafe replies for the given inode.  First wait on the

 * newest request, and make that the upper bound.  Then, if there are

 * more requests, keep waiting on the oldest as long as it is still older

 * than the original request.

 */

void ceph_sync_write_wait(struct inode *inode)

{

	struct ceph_inode_info *ci = ceph_inode(inode);

	struct list_head *head = &ci->i_unsafe_writes;

	struct ceph_osd_request *req;

	u64 last_tid;



	if (!S_ISREG(inode->i_mode))

		return;



	spin_lock(&ci->i_unsafe_lock);

	if (list_empty(head))

		goto out;



	/* set upper bound as _last_ entry in chain */



	req = list_last_entry(head, struct ceph_osd_request,

			      r_unsafe_item);

	last_tid = req->r_tid;



	do {

		ceph_osdc_get_request(req);

		spin_unlock(&ci->i_unsafe_lock);



		dout("sync_write_wait on tid %llu (until %llu)\n",

		     req->r_tid, last_tid);

		wait_for_completion(&req->r_safe_completion);

		ceph_osdc_put_request(req);



		spin_lock(&ci->i_unsafe_lock);

		/*

		 * from here on look at first entry in chain, since we

		 * only want to wait for anything older than last_tid

		 */

		if (list_empty(head))

			break;

		req = list_first_entry(head, struct ceph_osd_request,

				       r_unsafe_item);

	} while (req->r_tid < last_tid);

out:

	spin_unlock(&ci->i_unsafe_lock);

}



static ssize_t

ceph_direct_read_write(struct kiocb *iocb, struct iov_iter *iter,

fs/ceph/inode.c

+8 −0
Original line number Diff line number Diff line
@@ -585,6 +585,14 @@ int ceph_drop_inode(struct inode *inode) 
	return 1;

}



void ceph_evict_inode(struct inode *inode)

{

	/* wait unsafe sync writes */

	ceph_sync_write_wait(inode);

	truncate_inode_pages_final(&inode->i_data);

	clear_inode(inode);

}



static inline blkcnt_t calc_inode_blocks(u64 size)

{

	return (size + (1<<9) - 1) >> 9;

fs/ceph/super.c

+1 −0
Original line number Diff line number Diff line
@@ -731,6 +731,7 @@ static const struct super_operations ceph_super_ops = { 
	.destroy_inode	= ceph_destroy_inode,

	.write_inode    = ceph_write_inode,

	.drop_inode	= ceph_drop_inode,

	.evict_inode	= ceph_evict_inode,

	.sync_fs        = ceph_sync_fs,

	.put_super	= ceph_put_super,

	.show_options   = ceph_show_options,

fs/ceph/super.h

+2 −0
Original line number Diff line number Diff line
@@ -749,6 +749,7 @@ extern const struct inode_operations ceph_file_iops; 
extern struct inode *ceph_alloc_inode(struct super_block *sb);

extern void ceph_destroy_inode(struct inode *inode);

extern int ceph_drop_inode(struct inode *inode);

extern void ceph_evict_inode(struct inode *inode);



extern struct inode *ceph_get_inode(struct super_block *sb,

				    struct ceph_vino vino);
@@ -927,6 +928,7 @@ extern int ceph_atomic_open(struct inode *dir, struct dentry *dentry, 
extern int ceph_release(struct inode *inode, struct file *filp);

extern void ceph_fill_inline_data(struct inode *inode, struct page *locked_page,

				  char *data, size_t len);

extern void ceph_sync_write_wait(struct inode *inode);

/* dir.c */

extern const struct file_operations ceph_dir_fops;

extern const struct file_operations ceph_snapdir_fops;