aufs: white-out 1/2

The whiteout represents a logical deletion.
Although the document in this commit mentioned about rmdir(2) and
rename(2) for dir, this commit doesn't contain such functions. They will
be added in later commits.
Signed-off-by: J. R. Okajima <hooanon05g@gmail.com>

Documentation/filesystems/aufs/design/02struct.txt

@@ -129,3 +129,27 @@ whiteout base and etc. This is unrelated to a privilege.
 Most of aufs operation tries acquiring a rw_semaphore for aufs
 superblock at the beginning, at the same time waits for the completion
 of all queued asynchronous tasks.
+
+
+Whiteout
+----------------------------------------------------------------------
+The whiteout in aufs is very similar to Unionfs's. That is represented
+by its filename. UnionMount takes an approach of a file mode, but I am
+afraid several utilities (find(1) or something) will have to support it.
+
+Basically the whiteout represents "logical deletion" which stops aufs to
+lookup further, but also it represents "dir is opaque" which also stop
+further lookup.
+
+In aufs, rmdir(2) and rename(2) for dir uses whiteout alternatively.
+In order to make several functions in a single systemcall to be
+revertible, aufs adopts an approach to rename a directory to a temporary
+unique whiteouted name.
+For example, in rename(2) dir where the target dir already existed, aufs
+renames the target dir to a temporary unique whiteouted name before the
+actual rename on a branch, and then handles other actions (make it opaque,
+update the attributes, etc). If an error happens in these actions, aufs
+simply renames the whiteouted name back and returns an error. If all are
+succeeded, aufs registers a function to remove the whiteouted unique
+temporary name completely and asynchronously to the system global
+workqueue.

fs/aufs/Makefile

@@ -12,7 +12,7 @@ ccflags-y += -include ${srctree}/include/uapi/linux/aufs_type.h
 obj-$(CONFIG_AUFS_FS) += aufs.o
 aufs-y := module.o sbinfo.o super.o branch.o xino.o sysaufs.o opts.o \
 	wkq.o vfsub.o dcsub.o \
-	cpup.o \
+	cpup.o whout.o \
 	dinfo.o \
 	iinfo.o inode.o
 

fs/aufs/aufs.h

@@ -36,6 +36,7 @@
 #include "super.h"
 #include "sysaufs.h"
 #include "vfsub.h"
+#include "whout.h"
 #include "wkq.h"
 
 #endif /* __KERNEL__ */

fs/aufs/debug.c

@@ -138,9 +138,9 @@ void au_dpri_dentry(struct dentry *dentry)
 	dinfo = au_di(dentry);
 	if (!dinfo)
 		return;
-	dpri("d-1: btop %d, bbot %d, gen %d\n",
+	dpri("d-1: btop %d, bbot %d, bwh %d, gen %d\n",
 	     dinfo->di_btop, dinfo->di_bbot,
-	     au_digen(dentry));
+	     dinfo->di_bwh, au_digen(dentry));
 	if (dinfo->di_btop < 0)
 		return;
 	for (bindex = dinfo->di_btop; bindex <= dinfo->di_bbot; bindex++)

fs/aufs/dentry.h

@@ -24,7 +24,7 @@ struct au_dinfo {
 	atomic_t		di_generation;
 
 	struct au_rwsem		di_rwsem;
-	aufs_bindex_t		di_btop, di_bbot;
+	aufs_bindex_t		di_btop, di_bbot, di_bwh;
 	struct au_hdentry	*di_hdentry;
 	struct rcu_head		rcu;
 } ____cacheline_aligned_in_smp;
@@ -147,6 +147,12 @@ static inline aufs_bindex_t au_dbbot(struct dentry *dentry)
 	return au_di(dentry)->di_bbot;
 }
 
+static inline aufs_bindex_t au_dbwh(struct dentry *dentry)
+{
+	DiMustAnyLock(dentry);
+	return au_di(dentry)->di_bwh;
+}
+
 /* todo: hard/soft set? */
 static inline void au_set_dbtop(struct dentry *dentry, aufs_bindex_t bindex)
 {
@@ -160,5 +166,12 @@ static inline void au_set_dbbot(struct dentry *dentry, aufs_bindex_t bindex)
 	au_di(dentry)->di_bbot = bindex;
 }
 
+static inline void au_set_dbwh(struct dentry *dentry, aufs_bindex_t bindex)
+{
+	DiMustWriteLock(dentry);
+	/* dbwh can be outside of btop - bbot range */
+	au_di(dentry)->di_bwh = bindex;
+}
+
 #endif /* __KERNEL__ */
 #endif /* __AUFS_DENTRY_H__ */

fs/aufs/dinfo.c

@@ -33,6 +33,7 @@ struct au_dinfo *au_di_alloc(struct super_block *sb, unsigned int lsc)
 		au_rw_write_lock_nested(&dinfo->di_rwsem, lsc);
 		dinfo->di_btop = -1;
 		dinfo->di_bbot = -1;
+		dinfo->di_bwh = -1;
 		for (i = 0; i < nbr; i++)
 			dinfo->di_hdentry[i].hd_id = -1;
 		goto out;

fs/aufs/vfsub.h

@@ -43,6 +43,12 @@ int vfsub_kern_path(const char *name, unsigned int flags, struct path *path);
 struct dentry *vfsub_lookup_one_len(const char *name, struct dentry *parent,
 				    int len);
 
+static inline struct dentry *vfsub_lkup_one(struct qstr *name,
+					    struct dentry *parent)
+{
+	return vfsub_lookup_one_len(name->name, parent, name->len);
+}
+
 /* ---------------------------------------------------------------------- */
 
 static inline loff_t vfsub_f_size_read(struct file *file)

fs/aufs/whout.c

+// SPDX-License-Identifier: GPL-2.0
+/*
+ * Copyright (C) 2005-2019 Junjiro R. Okajima
+ */
+
+/*
+ * whiteout for logical deletion and opaque directory
+ */
+
+#include "aufs.h"
+
+#define WH_MASK			0444
+
+/*
+ * generate whiteout name, which is NOT terminated by NULL.
+ * @name: original d_name.name
+ * @len: original d_name.len
+ * @wh: whiteout qstr
+ * returns zero when succeeds, otherwise error.
+ * succeeded value as wh->name should be freed by kfree().
+ */
+int au_wh_name_alloc(struct qstr *wh, const struct qstr *name)
+{
+	char *p;
+
+	if (unlikely(name->len > PATH_MAX - AUFS_WH_PFX_LEN))
+		return -ENAMETOOLONG;
+
+	wh->len = name->len + AUFS_WH_PFX_LEN;
+	p = kmalloc(wh->len, GFP_NOFS);
+	wh->name = p;
+	if (p) {
+		memcpy(p, AUFS_WH_PFX, AUFS_WH_PFX_LEN);
+		memcpy(p + AUFS_WH_PFX_LEN, name->name, name->len);
+		/* smp_mb(); */
+		return 0;
+	}
+	return -ENOMEM;
+}
+
+/* ---------------------------------------------------------------------- */
+
+/*
+ * test if the @wh_name exists under @h_parent.
+ * @try_sio specifies the necessary of super-io.
+ */
+int au_wh_test(struct dentry *h_parent, struct qstr *wh_name, int try_sio)
+{
+	int err;
+	struct dentry *wh_dentry;
+
+#if 0 /* re-commit later */
+	if (!try_sio)
+		wh_dentry = vfsub_lkup_one(wh_name, h_parent);
+	else
+		wh_dentry = au_sio_lkup_one(wh_name, h_parent);
+#else
+	wh_dentry = ERR_PTR(-ENOENT);
+#endif
+	err = PTR_ERR(wh_dentry);
+	if (IS_ERR(wh_dentry)) {
+		if (err == -ENAMETOOLONG)
+			err = 0;
+		goto out;
+	}
+
+	err = 0;
+	if (d_is_negative(wh_dentry))
+		goto out_wh; /* success */
+
+	err = 1;
+	if (d_is_reg(wh_dentry))
+		goto out_wh; /* success */
+
+	err = -EIO;
+	AuIOErr("%pd Invalid whiteout entry type 0%o.\n",
+		wh_dentry, d_inode(wh_dentry)->i_mode);
+
+out_wh:
+	dput(wh_dentry);
+out:
+	return err;
+}
+
+/* ---------------------------------------------------------------------- */
+
+/*
+ * lookup whiteout dentry.
+ * @h_parent: lower parent dentry which must exist and be locked
+ * @base_name: name of dentry which will be whiteouted
+ * returns dentry for whiteout.
+ */
+struct dentry *au_wh_lkup(struct dentry *h_parent, struct qstr *base_name,
+			  struct au_branch *br)
+{
+	int err;
+	struct qstr wh_name;
+	struct dentry *wh_dentry;
+
+	err = au_wh_name_alloc(&wh_name, base_name);
+	wh_dentry = ERR_PTR(err);
+	if (!err) {
+		wh_dentry = vfsub_lkup_one(&wh_name, h_parent);
+		au_kfree_try_rcu(wh_name.name);
+	}
+	return wh_dentry;
+}
+
+/*
+ * link/create a whiteout for @dentry on @bindex.
+ */
+struct dentry *au_wh_create(struct dentry *dentry, aufs_bindex_t bindex,
+			    struct dentry *h_parent)
+{
+	struct dentry *wh_dentry;
+	struct super_block *sb;
+	int err;
+
+	sb = dentry->d_sb;
+	wh_dentry = au_wh_lkup(h_parent, &dentry->d_name, au_sbr(sb, bindex));
+	if (!IS_ERR(wh_dentry) && d_is_negative(wh_dentry)) {
+		/* err = link_or_create_wh(sb, bindex, wh_dentry); re-commit later */
+		err = 0;
+		if (!err)
+			au_set_dbwh(dentry, bindex);
+		else {
+			dput(wh_dentry);
+			wh_dentry = ERR_PTR(err);
+		}
+	}
+
+	return wh_dentry;
+}

fs/aufs/whout.h

+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2005-2019 Junjiro R. Okajima
+ */
+
+/*
+ * whiteout for logical deletion and opaque directory
+ */
+
+#ifndef __AUFS_WHOUT_H__
+#define __AUFS_WHOUT_H__
+
+#ifdef __KERNEL__
+
+/* whout.c */
+struct qstr;
+int au_wh_name_alloc(struct qstr *wh, const struct qstr *name);
+struct dentry;
+int au_wh_test(struct dentry *h_parent, struct qstr *wh_name, int try_sio);
+struct au_branch;
+struct dentry *au_wh_lkup(struct dentry *h_parent, struct qstr *base_name,
+			  struct au_branch *br);
+struct dentry *au_wh_create(struct dentry *dentry, aufs_bindex_t bindex,
+			    struct dentry *h_parent);
+
+#endif /* __KERNEL__ */
+#endif /* __AUFS_WHOUT_H__ */

include/uapi/linux/aufs_type.h

@@ -61,6 +61,9 @@ typedef int16_t aufs_bindex_t;
 #define AUFS_ROOT_INO		2
 #define AUFS_FIRST_INO		11
 
+#define AUFS_WH_PFX		".wh."
+#define AUFS_WH_PFX_LEN		((int)sizeof(AUFS_WH_PFX) - 1)
+#define AUFS_WH_TMP_LEN		4
 #define AUFS_XINO_FNAME		"." AUFS_NAME ".xino"
 #define AUFS_XINO_DEFPATH	"/tmp/" AUFS_XINO_FNAME
 #define AUFS_XINO_DEF_SEC	30 /* seconds */