1 files changed, 97 insertions, 42 deletions

diff --git a/fs/overlayfs/inode.c b/fs/overlayfs/inode.c
index 79e8994e3bc1..8be6cd264f66 100644
--- a/fs/overlayfs/inode.c
+++ b/fs/overlayfs/inode.c
@@ -10,6 +10,7 @@
 #include <linux/xattr.h>
 #include <linux/posix_acl.h>
 #include <linux/ratelimit.h>
+#include <linux/fiemap.h>
 #include "overlayfs.h"
 
 
@@ -58,6 +59,24 @@ int ovl_setattr(struct dentry *dentry, struct iattr *attr)
 		if (attr->ia_valid & (ATTR_KILL_SUID|ATTR_KILL_SGID))
 			attr->ia_valid &= ~ATTR_MODE;
 
+		/*
+		 * We might have to translate ovl file into real file object
+		 * once use cases emerge.  For now, simply don't let underlying
+		 * filesystem rely on attr->ia_file
+		 */
+		attr->ia_valid &= ~ATTR_FILE;
+
+		/*
+		 * If open(O_TRUNC) is done, VFS calls ->setattr with ATTR_OPEN
+		 * set.  Overlayfs does not pass O_TRUNC flag to underlying
+		 * filesystem during open -> do not pass ATTR_OPEN.  This
+		 * disables optimization in fuse which assumes open(O_TRUNC)
+		 * already set file size to 0.  But we never passed O_TRUNC to
+		 * fuse.  So by clearing ATTR_OPEN, fuse will be forced to send
+		 * setattr request to server.
+		 */
+		attr->ia_valid &= ~ATTR_OPEN;
+
 		inode_lock(upperdentry->d_inode);
 		old_cred = ovl_override_creds(dentry->d_sb);
 		err = notify_change(upperdentry, attr, NULL);
@@ -79,6 +98,7 @@ static int ovl_map_dev_ino(struct dentry *dentry, struct kstat *stat, int fsid)
 {
 	bool samefs = ovl_same_fs(dentry->d_sb);
 	unsigned int xinobits = ovl_xino_bits(dentry->d_sb);
+	unsigned int xinoshift = 64 - xinobits;
 
 	if (samefs) {
 		/*
@@ -89,22 +109,22 @@ static int ovl_map_dev_ino(struct dentry *dentry, struct kstat *stat, int fsid)
 		stat->dev = dentry->d_sb->s_dev;
 		return 0;
 	} else if (xinobits) {
-		unsigned int shift = 64 - xinobits;
 		/*
 		 * All inode numbers of underlying fs should not be using the
 		 * high xinobits, so we use high xinobits to partition the
 		 * overlay st_ino address space. The high bits holds the fsid
-		 * (upper fsid is 0). This way overlay inode numbers are unique
-		 * and all inodes use overlay st_dev. Inode numbers are also
-		 * persistent for a given layer configuration.
+		 * (upper fsid is 0). The lowest xinobit is reserved for mapping
+		 * the non-peresistent inode numbers range in case of overflow.
+		 * This way all overlay inode numbers are unique and use the
+		 * overlay st_dev.
 		 */
-		if (stat->ino >> shift) {
-			pr_warn_ratelimited("inode number too big (%pd2, ino=%llu, xinobits=%d)\n",
-					    dentry, stat->ino, xinobits);
-		} else {
-			stat->ino |= ((u64)fsid) << shift;
+		if (likely(!(stat->ino >> xinoshift))) {
+			stat->ino |= ((u64)fsid) << (xinoshift + 1);
 			stat->dev = dentry->d_sb->s_dev;
 			return 0;
+		} else if (ovl_xino_warn(dentry->d_sb)) {
+			pr_warn_ratelimited("inode number too big (%pd2, ino=%llu, xinobits=%d)\n",
+					    dentry, stat->ino, xinobits);
 		}
 	}
 
@@ -437,7 +457,7 @@ int ovl_update_time(struct inode *inode, struct timespec64 *ts, int flags)
 	if (flags & S_ATIME) {
 		struct ovl_fs *ofs = inode->i_sb->s_fs_info;
 		struct path upperpath = {
-			.mnt = ofs->upper_mnt,
+			.mnt = ovl_upper_mnt(ofs),
 			.dentry = ovl_upperdentry_dereference(OVL_I(inode)),
 		};
 
@@ -460,10 +480,6 @@ static int ovl_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 		return -EOPNOTSUPP;
 
 	old_cred = ovl_override_creds(inode->i_sb);
-
-	if (fieinfo->fi_flags & FIEMAP_FLAG_SYNC)
-		filemap_write_and_wait(realinode->i_mapping);
-
 	err = realinode->i_op->fiemap(realinode, fieinfo, start, len);
 	revert_creds(old_cred);
 
@@ -504,7 +520,7 @@ static const struct address_space_operations ovl_aops = {
 
 /*
  * It is possible to stack overlayfs instance on top of another
- * overlayfs instance as lower layer. We need to annonate the
+ * overlayfs instance as lower layer. We need to annotate the
  * stackable i_mutex locks according to stack level of the super
  * block instance. An overlayfs instance can never be in stack
  * depth 0 (there is always a real fs below it).  An overlayfs
@@ -561,27 +577,73 @@ static inline void ovl_lockdep_annotate_inode_mutex_key(struct inode *inode)
 #endif
 }
 
-static void ovl_fill_inode(struct inode *inode, umode_t mode, dev_t rdev,
-			   unsigned long ino, int fsid)
+static void ovl_next_ino(struct inode *inode)
+{
+	struct ovl_fs *ofs = inode->i_sb->s_fs_info;
+
+	inode->i_ino = atomic_long_inc_return(&ofs->last_ino);
+	if (unlikely(!inode->i_ino))
+		inode->i_ino = atomic_long_inc_return(&ofs->last_ino);
+}
+
+static void ovl_map_ino(struct inode *inode, unsigned long ino, int fsid)
 {
 	int xinobits = ovl_xino_bits(inode->i_sb);
+	unsigned int xinoshift = 64 - xinobits;
 
 	/*
 	 * When d_ino is consistent with st_ino (samefs or i_ino has enough
 	 * bits to encode layer), set the same value used for st_ino to i_ino,
 	 * so inode number exposed via /proc/locks and a like will be
 	 * consistent with d_ino and st_ino values. An i_ino value inconsistent
-	 * with d_ino also causes nfsd readdirplus to fail.  When called from
-	 * ovl_new_inode(), ino arg is 0, so i_ino will be updated to real
-	 * upper inode i_ino on ovl_inode_init() or ovl_inode_update().
+	 * with d_ino also causes nfsd readdirplus to fail.
 	 */
-	if (ovl_same_dev(inode->i_sb)) {
-		inode->i_ino = ino;
-		if (xinobits && fsid && !(ino >> (64 - xinobits)))
-			inode->i_ino |= (unsigned long)fsid << (64 - xinobits);
-	} else {
-		inode->i_ino = get_next_ino();
+	inode->i_ino = ino;
+	if (ovl_same_fs(inode->i_sb)) {
+		return;
+	} else if (xinobits && likely(!(ino >> xinoshift))) {
+		inode->i_ino |= (unsigned long)fsid << (xinoshift + 1);
+		return;
 	}
+
+	/*
+	 * For directory inodes on non-samefs with xino disabled or xino
+	 * overflow, we allocate a non-persistent inode number, to be used for
+	 * resolving st_ino collisions in ovl_map_dev_ino().
+	 *
+	 * To avoid ino collision with legitimate xino values from upper
+	 * layer (fsid 0), use the lowest xinobit to map the non
+	 * persistent inode numbers to the unified st_ino address space.
+	 */
+	if (S_ISDIR(inode->i_mode)) {
+		ovl_next_ino(inode);
+		if (xinobits) {
+			inode->i_ino &= ~0UL >> xinobits;
+			inode->i_ino |= 1UL << xinoshift;
+		}
+	}
+}
+
+void ovl_inode_init(struct inode *inode, struct ovl_inode_params *oip,
+		    unsigned long ino, int fsid)
+{
+	struct inode *realinode;
+
+	if (oip->upperdentry)
+		OVL_I(inode)->__upperdentry = oip->upperdentry;
+	if (oip->lowerpath && oip->lowerpath->dentry)
+		OVL_I(inode)->lower = igrab(d_inode(oip->lowerpath->dentry));
+	if (oip->lowerdata)
+		OVL_I(inode)->lowerdata = igrab(d_inode(oip->lowerdata));
+
+	realinode = ovl_inode_real(inode);
+	ovl_copyattr(realinode, inode);
+	ovl_copyflags(realinode, inode);
+	ovl_map_ino(inode, ino, fsid);
+}
+
+static void ovl_fill_inode(struct inode *inode, umode_t mode, dev_t rdev)
+{
 	inode->i_mode = mode;
 	inode->i_flags |= S_NOCMTIME;
 #ifdef CONFIG_FS_POSIX_ACL
@@ -719,7 +781,7 @@ struct inode *ovl_new_inode(struct super_block *sb, umode_t mode, dev_t rdev)
 
 	inode = new_inode(sb);
 	if (inode)
-		ovl_fill_inode(inode, mode, rdev, 0, 0);
+		ovl_fill_inode(inode, mode, rdev);
 
 	return inode;
 }
@@ -843,7 +905,7 @@ struct inode *ovl_get_trap_inode(struct super_block *sb, struct dentry *dir)
  * Does overlay inode need to be hashed by lower inode?
  */
 static bool ovl_hash_bylower(struct super_block *sb, struct dentry *upper,
-			     struct dentry *lower, struct dentry *index)
+			     struct dentry *lower, bool index)
 {
 	struct ovl_fs *ofs = sb->s_fs_info;
 
@@ -856,7 +918,7 @@ static bool ovl_hash_bylower(struct super_block *sb, struct dentry *upper,
 		return true;
 
 	/* Yes, if won't be copied up */
-	if (!ofs->upper_mnt)
+	if (!ovl_upper_mnt(ofs))
 		return true;
 
 	/* No, if lower hardlink is or will be broken on copy up */
@@ -891,8 +953,8 @@ struct inode *ovl_get_inode(struct super_block *sb,
 	struct dentry *lowerdentry = lowerpath ? lowerpath->dentry : NULL;
 	bool bylower = ovl_hash_bylower(sb, upperdentry, lowerdentry,
 					oip->index);
-	int fsid = bylower ? oip->lowerpath->layer->fsid : 0;
-	bool is_dir, metacopy = false;
+	int fsid = bylower ? lowerpath->layer->fsid : 0;
+	bool is_dir;
 	unsigned long ino = 0;
 	int err = oip->newinode ? -EEXIST : -ENOMEM;
 
@@ -941,9 +1003,11 @@ struct inode *ovl_get_inode(struct super_block *sb,
 			err = -ENOMEM;
 			goto out_err;
 		}
+		ino = realinode->i_ino;
+		fsid = lowerpath->layer->fsid;
 	}
-	ovl_fill_inode(inode, realinode->i_mode, realinode->i_rdev, ino, fsid);
-	ovl_inode_init(inode, upperdentry, lowerdentry, oip->lowerdata);
+	ovl_fill_inode(inode, realinode->i_mode, realinode->i_rdev);
+	ovl_inode_init(inode, oip, ino, fsid);
 
 	if (upperdentry && ovl_is_impuredir(upperdentry))
 		ovl_set_flag(OVL_IMPURE, inode);
@@ -951,15 +1015,6 @@ struct inode *ovl_get_inode(struct super_block *sb,
 	if (oip->index)
 		ovl_set_flag(OVL_INDEX, inode);
 
-	if (upperdentry) {
-		err = ovl_check_metacopy_xattr(upperdentry);
-		if (err < 0)
-			goto out_err;
-		metacopy = err;
-		if (!metacopy)
-			ovl_set_flag(OVL_UPPERDATA, inode);
-	}
-
 	OVL_I(inode)->redirect = oip->redirect;
 
 	if (bylower)