Merge branch 'work.sysv' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs

Pull sysv updates from Al Viro:
 "Fabio's 'switch to kmap_local_page()' patchset (originally after the
  ext2 counterpart, with a lot of cleaning up done to it; as the matter
  of fact, ext2 side is in need of similar cleanups - calling
  conventions there are bloody awful).

  Plus the equivalents of minix stuff..."

* 'work.sysv' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  sysv: fix handling of delete_entry and set_link failures
  fs/sysv: Replace kmap() with kmap_local_page()
  fs/sysv: Use dir_put_page() in sysv_rename()
  fs/sysv: Change the signature of dir_get_page()
  fs/sysv: Use the offset_in_page() helper
  sysv: don't flush page immediately for DIRSYNC directories

3 files changed, 111 insertions, 88 deletions

diff --git a/fs/sysv/dir.c b/fs/sysv/dir.c
index 88e38cd8f5c9..999bceb99974 100644
--- a/fs/sysv/dir.c
+++ b/fs/sysv/dir.c
@@ -28,37 +28,50 @@ const struct file_operations sysv_dir_operations = {
 	.fsync		= generic_file_fsync,
 };
 
-static inline void dir_put_page(struct page *page)
+inline void dir_put_page(struct page *page, void *page_addr)
 {
-	kunmap(page);
+	kunmap_local((void *)((unsigned long)page_addr & PAGE_MASK));
 	put_page(page);
 }
 
-static int dir_commit_chunk(struct page *page, loff_t pos, unsigned len)
+static void dir_commit_chunk(struct page *page, loff_t pos, unsigned len)
 {
 	struct address_space *mapping = page->mapping;
 	struct inode *dir = mapping->host;
-	int err = 0;
 
 	block_write_end(NULL, mapping, pos, len, len, page, NULL);
 	if (pos+len > dir->i_size) {
 		i_size_write(dir, pos+len);
 		mark_inode_dirty(dir);
 	}
-	if (IS_DIRSYNC(dir))
-		err = write_one_page(page);
-	else
-		unlock_page(page);
+	unlock_page(page);
+}
+
+static int sysv_handle_dirsync(struct inode *dir)
+{
+	int err;
+
+	err = filemap_write_and_wait(dir->i_mapping);
+	if (!err)
+		err = sync_inode_metadata(dir, 1);
 	return err;
 }
 
-static struct page * dir_get_page(struct inode *dir, unsigned long n)
+/*
+ * Calls to dir_get_page()/dir_put_page() must be nested according to the
+ * rules documented in mm/highmem.rst.
+ *
+ * NOTE: sysv_find_entry() and sysv_dotdot() act as calls to dir_get_page()
+ * and must be treated accordingly for nesting purposes.
+ */
+static void *dir_get_page(struct inode *dir, unsigned long n, struct page **p)
 {
 	struct address_space *mapping = dir->i_mapping;
 	struct page *page = read_mapping_page(mapping, n, NULL);
-	if (!IS_ERR(page))
-		kmap(page);
-	return page;
+	if (IS_ERR(page))
+		return ERR_CAST(page);
+	*p = page;
+	return kmap_local_page(page);
 }
 
 static int sysv_readdir(struct file *file, struct dir_context *ctx)
@@ -80,11 +93,11 @@ static int sysv_readdir(struct file *file, struct dir_context *ctx)
 	for ( ; n < npages; n++, offset = 0) {
 		char *kaddr, *limit;
 		struct sysv_dir_entry *de;
-		struct page *page = dir_get_page(inode, n);
+		struct page *page;
 
-		if (IS_ERR(page))
+		kaddr = dir_get_page(inode, n, &page);
+		if (IS_ERR(kaddr))
 			continue;
-		kaddr = (char *)page_address(page);
 		de = (struct sysv_dir_entry *)(kaddr+offset);
 		limit = kaddr + PAGE_SIZE - SYSV_DIRSIZE;
 		for ( ;(char*)de <= limit; de++, ctx->pos += sizeof(*de)) {
@@ -96,11 +109,11 @@ static int sysv_readdir(struct file *file, struct dir_context *ctx)
 			if (!dir_emit(ctx, name, strnlen(name,SYSV_NAMELEN),
 					fs16_to_cpu(SYSV_SB(sb), de->inode),
 					DT_UNKNOWN)) {
-				dir_put_page(page);
+				dir_put_page(page, kaddr);
 				return 0;
 			}
 		}
-		dir_put_page(page);
+		dir_put_page(page, kaddr);
 	}
 	return 0;
 }
@@ -123,6 +136,11 @@ static inline int namecompare(int len, int maxlen,
  * returns the cache buffer in which the entry was found, and the entry
  * itself (as a parameter - res_dir). It does NOT read the inode of the
  * entry - you'll have to do that yourself if you want to.
+ *
+ * On Success dir_put_page() should be called on *res_page.
+ *
+ * sysv_find_entry() acts as a call to dir_get_page() and must be treated
+ * accordingly for nesting purposes.
  */
 struct sysv_dir_entry *sysv_find_entry(struct dentry *dentry, struct page **res_page)
 {
@@ -142,11 +160,10 @@ struct sysv_dir_entry *sysv_find_entry(struct dentry *dentry, struct page **res_
 	n = start;
 
 	do {
-		char *kaddr;
-		page = dir_get_page(dir, n);
-		if (!IS_ERR(page)) {
-			kaddr = (char*)page_address(page);
-			de = (struct sysv_dir_entry *) kaddr;
+		char *kaddr = dir_get_page(dir, n, &page);
+
+		if (!IS_ERR(kaddr)) {
+			de = (struct sysv_dir_entry *)kaddr;
 			kaddr += PAGE_SIZE - SYSV_DIRSIZE;
 			for ( ; (char *) de <= kaddr ; de++) {
 				if (!de->inode)
@@ -155,7 +172,7 @@ struct sysv_dir_entry *sysv_find_entry(struct dentry *dentry, struct page **res_
 							name, de->name))
 					goto found;
 			}
-			dir_put_page(page);
+			dir_put_page(page, kaddr);
 		}
 
 		if (++n >= npages)
@@ -185,11 +202,9 @@ int sysv_add_link(struct dentry *dentry, struct inode *inode)
 
 	/* We take care of directory expansion in the same loop */
 	for (n = 0; n <= npages; n++) {
-		page = dir_get_page(dir, n);
-		err = PTR_ERR(page);
-		if (IS_ERR(page))
-			goto out;
-		kaddr = (char*)page_address(page);
+		kaddr = dir_get_page(dir, n, &page);
+		if (IS_ERR(kaddr))
+			return PTR_ERR(kaddr);
 		de = (struct sysv_dir_entry *)kaddr;
 		kaddr += PAGE_SIZE - SYSV_DIRSIZE;
 		while ((char *)de <= kaddr) {
@@ -200,14 +215,13 @@ int sysv_add_link(struct dentry *dentry, struct inode *inode)
 				goto out_page;
 			de++;
 		}
-		dir_put_page(page);
+		dir_put_page(page, kaddr);
 	}
 	BUG();
 	return -EINVAL;
 
 got_it:
-	pos = page_offset(page) +
-			(char*)de - (char*)page_address(page);
+	pos = page_offset(page) + offset_in_page(de);
 	lock_page(page);
 	err = sysv_prepare_chunk(page, pos, SYSV_DIRSIZE);
 	if (err)
@@ -215,12 +229,12 @@ got_it:
 	memcpy (de->name, name, namelen);
 	memset (de->name + namelen, 0, SYSV_DIRSIZE - namelen - 2);
 	de->inode = cpu_to_fs16(SYSV_SB(inode->i_sb), inode->i_ino);
-	err = dir_commit_chunk(page, pos, SYSV_DIRSIZE);
+	dir_commit_chunk(page, pos, SYSV_DIRSIZE);
 	dir->i_mtime = dir->i_ctime = current_time(dir);
 	mark_inode_dirty(dir);
+	err = sysv_handle_dirsync(dir);
 out_page:
-	dir_put_page(page);
-out:
+	dir_put_page(page, kaddr);
 	return err;
 out_unlock:
 	unlock_page(page);
@@ -230,19 +244,20 @@ out_unlock:
 int sysv_delete_entry(struct sysv_dir_entry *de, struct page *page)
 {
 	struct inode *inode = page->mapping->host;
-	char *kaddr = (char*)page_address(page);
-	loff_t pos = page_offset(page) + (char *)de - kaddr;
+	loff_t pos = page_offset(page) + offset_in_page(de);
 	int err;
 
 	lock_page(page);
 	err = sysv_prepare_chunk(page, pos, SYSV_DIRSIZE);
-	BUG_ON(err);
+	if (err) {
+		unlock_page(page);
+		return err;
+	}
 	de->inode = 0;
-	err = dir_commit_chunk(page, pos, SYSV_DIRSIZE);
-	dir_put_page(page);
+	dir_commit_chunk(page, pos, SYSV_DIRSIZE);
 	inode->i_ctime = inode->i_mtime = current_time(inode);
 	mark_inode_dirty(inode);
-	return err;
+	return sysv_handle_dirsync(inode);
 }
 
 int sysv_make_empty(struct inode *inode, struct inode *dir)
@@ -259,9 +274,7 @@ int sysv_make_empty(struct inode *inode, struct inode *dir)
 		unlock_page(page);
 		goto fail;
 	}
-	kmap(page);
-
-	base = (char*)page_address(page);
+	base = kmap_local_page(page);
 	memset(base, 0, PAGE_SIZE);
 
 	de = (struct sysv_dir_entry *) base;
@@ -271,8 +284,9 @@ int sysv_make_empty(struct inode *inode, struct inode *dir)
 	de->inode = cpu_to_fs16(SYSV_SB(inode->i_sb), dir->i_ino);
 	strcpy(de->name,"..");
 
-	kunmap(page);
-	err = dir_commit_chunk(page, 0, 2 * SYSV_DIRSIZE);
+	kunmap_local(base);
+	dir_commit_chunk(page, 0, 2 * SYSV_DIRSIZE);
+	err = sysv_handle_dirsync(inode);
 fail:
 	put_page(page);
 	return err;
@@ -286,16 +300,15 @@ int sysv_empty_dir(struct inode * inode)
 	struct super_block *sb = inode->i_sb;
 	struct page *page = NULL;
 	unsigned long i, npages = dir_pages(inode);
+	char *kaddr;
 
 	for (i = 0; i < npages; i++) {
-		char *kaddr;
-		struct sysv_dir_entry * de;
-		page = dir_get_page(inode, i);
+		struct sysv_dir_entry *de;
 
-		if (IS_ERR(page))
+		kaddr = dir_get_page(inode, i, &page);
+		if (IS_ERR(kaddr))
 			continue;
 
-		kaddr = (char *)page_address(page);
 		de = (struct sysv_dir_entry *)kaddr;
 		kaddr += PAGE_SIZE-SYSV_DIRSIZE;
 
@@ -314,44 +327,51 @@ int sysv_empty_dir(struct inode * inode)
 			if (de->name[1] != '.' || de->name[2])
 				goto not_empty;
 		}
-		dir_put_page(page);
+		dir_put_page(page, kaddr);
 	}
 	return 1;
 
 not_empty:
-	dir_put_page(page);
+	dir_put_page(page, kaddr);
 	return 0;
 }
 
 /* Releases the page */
-void sysv_set_link(struct sysv_dir_entry *de, struct page *page,
+int sysv_set_link(struct sysv_dir_entry *de, struct page *page,
 	struct inode *inode)
 {
 	struct inode *dir = page->mapping->host;
-	loff_t pos = page_offset(page) +
-			(char *)de-(char*)page_address(page);
+	loff_t pos = page_offset(page) + offset_in_page(de);
 	int err;
 
 	lock_page(page);
 	err = sysv_prepare_chunk(page, pos, SYSV_DIRSIZE);
-	BUG_ON(err);
+	if (err) {
+		unlock_page(page);
+		return err;
+	}
 	de->inode = cpu_to_fs16(SYSV_SB(inode->i_sb), inode->i_ino);
-	err = dir_commit_chunk(page, pos, SYSV_DIRSIZE);
-	dir_put_page(page);
+	dir_commit_chunk(page, pos, SYSV_DIRSIZE);
 	dir->i_mtime = dir->i_ctime = current_time(dir);
 	mark_inode_dirty(dir);
+	return sysv_handle_dirsync(inode);
 }
 
-struct sysv_dir_entry * sysv_dotdot (struct inode *dir, struct page **p)
+/*
+ * Calls to dir_get_page()/dir_put_page() must be nested according to the
+ * rules documented in mm/highmem.rst.
+ *
+ * sysv_dotdot() acts as a call to dir_get_page() and must be treated
+ * accordingly for nesting purposes.
+ */
+struct sysv_dir_entry *sysv_dotdot(struct inode *dir, struct page **p)
 {
-	struct page *page = dir_get_page(dir, 0);
-	struct sysv_dir_entry *de = NULL;
+	struct sysv_dir_entry *de = dir_get_page(dir, 0, p);
 
-	if (!IS_ERR(page)) {
-		de = (struct sysv_dir_entry*) page_address(page) + 1;
-		*p = page;
-	}
-	return de;
+	if (IS_ERR(de))
+		return NULL;
+	/* ".." is the second directory entry */
+	return de + 1;
 }
 
 ino_t sysv_inode_by_name(struct dentry *dentry)
@@ -362,7 +382,7 @@ ino_t sysv_inode_by_name(struct dentry *dentry)
 	
 	if (de) {
 		res = fs16_to_cpu(SYSV_SB(dentry->d_sb), de->inode);
-		dir_put_page(page);
+		dir_put_page(page, de);
 	}
 	return res;
 }
diff --git a/fs/sysv/namei.c b/fs/sysv/namei.c
index ecd424461511..a25862773d82 100644
--- a/fs/sysv/namei.c
+++ b/fs/sysv/namei.c
@@ -153,19 +153,18 @@ static int sysv_unlink(struct inode * dir, struct dentry * dentry)
 	struct inode * inode = d_inode(dentry);
 	struct page * page;
 	struct sysv_dir_entry * de;
-	int err = -ENOENT;
+	int err;
 
 	de = sysv_find_entry(dentry, &page);
 	if (!de)
-		goto out;
+		return -ENOENT;
 
-	err = sysv_delete_entry (de, page);
-	if (err)
-		goto out;
-
-	inode->i_ctime = dir->i_ctime;
-	inode_dec_link_count(inode);
-out:
+	err = sysv_delete_entry(de, page);
+	if (!err) {
+		inode->i_ctime = dir->i_ctime;
+		inode_dec_link_count(inode);
+	}
+	dir_put_page(page, de);
 	return err;
 }
 
@@ -227,7 +226,10 @@ static int sysv_rename(struct mnt_idmap *idmap, struct inode *old_dir,
 		new_de = sysv_find_entry(new_dentry, &new_page);
 		if (!new_de)
 			goto out_dir;
-		sysv_set_link(new_de, new_page, old_inode);
+		err = sysv_set_link(new_de, new_page, old_inode);
+		dir_put_page(new_page, new_de);
+		if (err)
+			goto out_dir;
 		new_inode->i_ctime = current_time(new_inode);
 		if (dir_de)
 			drop_nlink(new_inode);
@@ -240,23 +242,23 @@ static int sysv_rename(struct mnt_idmap *idmap, struct inode *old_dir,
 			inode_inc_link_count(new_dir);
 	}
 
-	sysv_delete_entry(old_de, old_page);
+	err = sysv_delete_entry(old_de, old_page);
+	if (err)
+		goto out_dir;
+
 	mark_inode_dirty(old_inode);
 
 	if (dir_de) {
-		sysv_set_link(dir_de, dir_page, new_dir);
-		inode_dec_link_count(old_dir);
+		err = sysv_set_link(dir_de, dir_page, new_dir);
+		if (!err)
+			inode_dec_link_count(old_dir);
 	}
-	return 0;
 
 out_dir:
-	if (dir_de) {
-		kunmap(dir_page);
-		put_page(dir_page);
-	}
+	if (dir_de)
+		dir_put_page(dir_page, dir_de);
 out_old:
-	kunmap(old_page);
-	put_page(old_page);
+	dir_put_page(old_page, old_de);
 out:
 	return err;
 }
diff --git a/fs/sysv/sysv.h b/fs/sysv/sysv.h
index 5e122a5673c1..f2c36ea42df6 100644
--- a/fs/sysv/sysv.h
+++ b/fs/sysv/sysv.h
@@ -148,12 +148,13 @@ extern void sysv_destroy_icache(void);
 
 
 /* dir.c */
+extern void dir_put_page(struct page *page, void *vaddr);
 extern struct sysv_dir_entry *sysv_find_entry(struct dentry *, struct page **);
 extern int sysv_add_link(struct dentry *, struct inode *);
 extern int sysv_delete_entry(struct sysv_dir_entry *, struct page *);
 extern int sysv_make_empty(struct inode *, struct inode *);
 extern int sysv_empty_dir(struct inode *);
-extern void sysv_set_link(struct sysv_dir_entry *, struct page *,
+extern int sysv_set_link(struct sysv_dir_entry *, struct page *,
 			struct inode *);
 extern struct sysv_dir_entry *sysv_dotdot(struct inode *, struct page **);
 extern ino_t sysv_inode_by_name(struct dentry *);