[patch 1/8] 2.6.22-rc4-mm2 buffered write fixes

[patch 1/8] 2.6.22-rc4-mm2 buffered write fixes

[Nick Piggin]

Here are several fixes for issues that Dmitriy found, and also several
fixlets for non-compiling filesystems.

These only caused one trivial down-stack reject, so I won't worry about
sending you the fix for that.

These have had some testing with various filesystems, block sizes, and
journal modes with fsx-linux, fsstress and swapping-kbuilds.

--

Dmitriy noticed some weird code I had in __page_symlink that is wrong.

Fixes: fs-introduce-write_begin-write_end-and-perform_write-aops.patch

Signed-off-by: Nick Piggin <npiggin@suse.de>

Index: linux-2.6/fs/namei.c
===================================================================
--- linux-2.6.orig/fs/namei.c
+++ linux-2.6/fs/namei.c
@@ -2702,21 +2702,20 @@ int __page_symlink(struct inode *inode, 
 	char *kaddr;
 
 retry:
-	err = pagecache_write_begin(NULL, mapping, 0, PAGE_CACHE_SIZE,
+	err = pagecache_write_begin(NULL, mapping, 0, len-1,
 				AOP_FLAG_UNINTERRUPTIBLE, &page, &fsdata);
 	if (err)
 		goto fail;
 
 	kaddr = kmap_atomic(page, KM_USER0);
 	memcpy(kaddr, symname, len-1);
-	memset(kaddr+len-1, 0, PAGE_CACHE_SIZE-(len-1));
 	kunmap_atomic(kaddr, KM_USER0);
 
-	err = pagecache_write_end(NULL, mapping, 0, PAGE_CACHE_SIZE, PAGE_CACHE_SIZE,
+	err = pagecache_write_end(NULL, mapping, 0, len-1, len-1,
 							page, fsdata);
 	if (err < 0)
 		goto fail;
-	if (err < PAGE_CACHE_SIZE)
+	if (err < len-1)
 		goto retry;
 
 	mark_inode_dirty(inode);

[patch 2/8] 2.6.22-rc4-mm2 buffered write fixes

[Nick Piggin]

Dmitriy noticed that iov_iter_fault_in_readable could go past the end
of the first iov in a multi-iov situation, and that could be considered
an EFAULT by the caller. Fix and comment.

Fixes: fs-introduce-write_begin-write_end-and-perform_write-aops.patch

Signed-off-by: Nick Piggin <npiggin@suse.de>

Index: linux-2.6/mm/filemap.c
===================================================================
--- linux-2.6.orig/mm/filemap.c
+++ linux-2.6/mm/filemap.c
@@ -1794,9 +1794,19 @@ void iov_iter_advance(struct iov_iter *i
 	i->count -= bytes;
 }
 
+/*
+ * Fault in the first iovec of the given iov_iter, to a maximum length
+ * of bytes. Returns 0 on success, or non-zero if the memory could not be
+ * accessed (ie. because it is an invalid address).
+ *
+ * writev-intensive code may want this to prefault several iovecs -- that
+ * would be possible (callers must not rely on the fact that _only_ the
+ * first iovec will be faulted with the current implementation).
+ */
 int iov_iter_fault_in_readable(struct iov_iter *i, size_t bytes)
 {
 	char __user *buf = i->iov->iov_base + i->iov_offset;
+	bytes = min(bytes, i->iov->iov_len - i->iov_offset);
 	return fault_in_pages_readable(buf, bytes);
 }
 

[patch 3/8] 2.6.22-rc4-mm2 buffered write fixes

[Nick Piggin]

Don't move journal_stop from under page lock.

Fixes: ext3-convert-to-new-aops.patch
Signed-off-by: Nick Piggin <npiggin@suse.de>

Index: linux-2.6/fs/ext3/inode.c
===================================================================
--- linux-2.6.orig/fs/ext3/inode.c
+++ linux-2.6/fs/ext3/inode.c
@@ -36,6 +36,7 @@
 #include <linux/mpage.h>
 #include <linux/uio.h>
 #include <linux/bio.h>
+#include <linux/swap.h> /* mark_page_accessed */
 #include "xattr.h"
 #include "acl.h"
 
@@ -1217,6 +1218,31 @@ static int write_end_fn(handle_t *handle
 }
 
 /*
+ * Generic write_end handler for ordered and writeback ext3 journal modes.
+ * We can't use generic_write_end, because that unlocks the page and we need to
+ * unlock the page after ext3_journal_stop, but ext3_journal_stop must run
+ * after block_write_end.
+ */
+static int ext3_generic_write_end(struct file *file,
+				struct address_space *mapping,
+				loff_t pos, unsigned len, unsigned copied,
+				struct page *page, void *fsdata)
+{
+	struct inode *inode = file->f_mapping->host;
+
+	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);
+
+	mark_page_accessed(page);
+
+	if (pos+copied > inode->i_size) {
+		i_size_write(inode, pos+copied);
+		mark_inode_dirty(inode);
+	}
+
+	return copied;
+}
+
+/*
  * We need to pick up the new inode size which generic_commit_write gave us
  * `file' can be NULL - eg, when called from page_symlink().
  *
@@ -1250,17 +1276,17 @@ static int ext3_ordered_write_end(struct
 		new_i_size = pos + copied;
 		if (new_i_size > EXT3_I(inode)->i_disksize)
 			EXT3_I(inode)->i_disksize = new_i_size;
-		copied = generic_write_end(file, mapping, pos, len, copied,
+		copied = ext3_generic_write_end(file, mapping, pos, len, copied,
 							page, fsdata);
 		if (copied < 0)
 			ret = copied;
-	} else {
-		unlock_page(page);
-		page_cache_release(page);
 	}
 	ret2 = ext3_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
+	unlock_page(page);
+	page_cache_release(page);
+
 	return ret ? ret : copied;
 }
 
@@ -1278,7 +1304,7 @@ static int ext3_writeback_write_end(stru
 	if (new_i_size > EXT3_I(inode)->i_disksize)
 		EXT3_I(inode)->i_disksize = new_i_size;
 
-	copied = generic_write_end(file, mapping, pos, len, copied,
+	copied = ext3_generic_write_end(file, mapping, pos, len, copied,
 							page, fsdata);
 	if (copied < 0)
 		ret = copied;
@@ -1286,6 +1312,9 @@ static int ext3_writeback_write_end(stru
 	ret2 = ext3_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
+	unlock_page(page);
+	page_cache_release(page);
+
 	return ret ? ret : copied;
 }
 
@@ -1313,8 +1342,6 @@ static int ext3_journalled_write_end(str
 				to, &partial, write_end_fn);
 	if (!partial)
 		SetPageUptodate(page);
-	unlock_page(page);
-	page_cache_release(page);
 	if (pos+copied > inode->i_size)
 		i_size_write(inode, pos+copied);
 	EXT3_I(inode)->i_state |= EXT3_STATE_JDATA;
@@ -1328,6 +1355,9 @@ static int ext3_journalled_write_end(str
 	ret2 = ext3_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
+	unlock_page(page);
+	page_cache_release(page);
+
 	return ret ? ret : copied;
 }
 

[patch 4/8] 2.6.22-rc4-mm2 buffered write fixes

[Nick Piggin]

Don't move journal_stop from under page lock.

Fixes: ext4-convert-to-new-aops.patch
Signed-off-by: Nick Piggin <npiggin@suse.de>

Index: linux-2.6/fs/ext4/inode.c
===================================================================
--- linux-2.6.orig/fs/ext4/inode.c
+++ linux-2.6/fs/ext4/inode.c
@@ -36,6 +36,7 @@
 #include <linux/mpage.h>
 #include <linux/uio.h>
 #include <linux/bio.h>
+#include <linux/swap.h> /* mark_page_accessed */
 #include "xattr.h"
 #include "acl.h"
 
@@ -1215,6 +1216,31 @@ static int write_end_fn(handle_t *handle
 }
 
 /*
+ * Generic write_end handler for ordered and writeback ext4 journal modes.
+ * We can't use generic_write_end, because that unlocks the page and we need to
+ * unlock the page after ext4_journal_stop, but ext4_journal_stop must run
+ * after block_write_end.
+ */
+static int ext4_generic_write_end(struct file *file,
+				struct address_space *mapping,
+				loff_t pos, unsigned len, unsigned copied,
+				struct page *page, void *fsdata)
+{
+	struct inode *inode = file->f_mapping->host;
+
+	copied = block_write_end(file, mapping, pos, len, copied, page, fsdata);
+
+	mark_page_accessed(page);
+
+	if (pos+copied > inode->i_size) {
+		i_size_write(inode, pos+copied);
+		mark_inode_dirty(inode);
+	}
+
+	return copied;
+}
+
+/*
  * We need to pick up the new inode size which generic_commit_write gave us
  * `file' can be NULL - eg, when called from page_symlink().
  *
@@ -1248,17 +1274,17 @@ static int ext4_ordered_write_end(struct
 		new_i_size = pos + copied;
 		if (new_i_size > EXT4_I(inode)->i_disksize)
 			EXT4_I(inode)->i_disksize = new_i_size;
-		copied = generic_write_end(file, mapping, pos, len, copied,
+		copied = ext4_generic_write_end(file, mapping, pos, len, copied,
 							page, fsdata);
 		if (copied < 0)
 			ret = copied;
-	} else {
-		unlock_page(page);
-		page_cache_release(page);
 	}
 	ret2 = ext4_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
+	unlock_page(page);
+	page_cache_release(page);
+
 	return ret ? ret : copied;
 }
 
@@ -1276,7 +1302,7 @@ static int ext4_writeback_write_end(stru
 	if (new_i_size > EXT4_I(inode)->i_disksize)
 		EXT4_I(inode)->i_disksize = new_i_size;
 
-	copied = generic_write_end(file, mapping, pos, len, copied,
+	copied = ext4_generic_write_end(file, mapping, pos, len, copied,
 							page, fsdata);
 	if (copied < 0)
 		ret = copied;
@@ -1284,6 +1310,9 @@ static int ext4_writeback_write_end(stru
 	ret2 = ext4_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
+	unlock_page(page);
+	page_cache_release(page);
+
 	return ret ? ret : copied;
 }
 
@@ -1311,8 +1340,6 @@ static int ext4_journalled_write_end(str
 				to, &partial, write_end_fn);
 	if (!partial)
 		SetPageUptodate(page);
-	unlock_page(page);
-	page_cache_release(page);
 	if (pos+copied > inode->i_size)
 		i_size_write(inode, pos+copied);
 	EXT4_I(inode)->i_state |= EXT4_STATE_JDATA;
@@ -1326,6 +1353,9 @@ static int ext4_journalled_write_end(str
 	ret2 = ext4_journal_stop(handle);
 	if (!ret)
 		ret = ret2;
+	unlock_page(page);
+	page_cache_release(page);
+
 	return ret ? ret : copied;
 }
 

[patch 4/8] minix: convert to new aops fix

[Nick Piggin]

Index: linux-2.6/fs/minix/minix.h
===================================================================
--- linux-2.6.orig/fs/minix/minix.h
+++ linux-2.6/fs/minix/minix.h
@@ -54,6 +54,9 @@ extern int minix_new_block(struct inode 
 extern void minix_free_block(struct inode *inode, unsigned long block);
 extern unsigned long minix_count_free_blocks(struct minix_sb_info *sbi);
 extern int minix_getattr(struct vfsmount *, struct dentry *, struct kstat *);
+extern int __minix_write_begin(struct file *file, struct address_space *mapping,
+			loff_t pos, unsigned len, unsigned flags,
+			struct page **pagep, void **fsdata);
 
 extern void V1_minix_truncate(struct inode *);
 extern void V2_minix_truncate(struct inode *);

[patch 6/8] sysv: convert to new aops fix

[Nick Piggin]

Index: linux-2.6/fs/sysv/sysv.h
===================================================================
--- linux-2.6.orig/fs/sysv/sysv.h
+++ linux-2.6/fs/sysv/sysv.h
@@ -136,6 +136,9 @@ extern unsigned long sysv_count_free_blo
 
 /* itree.c */
 extern void sysv_truncate(struct inode *);
+extern int __sysv_write_begin(struct file *file, struct address_space *mapping,
+			loff_t pos, unsigned len, unsigned flags,
+			struct page **pagep, void **fsdata);
 
 /* inode.c */
 extern int sysv_write_inode(struct inode *, int);

[patch 7/8] ufs: convert to new aops fix

[Nick Piggin]

Index: linux-2.6/fs/ufs/util.h
===================================================================
--- linux-2.6.orig/fs/ufs/util.h
+++ linux-2.6/fs/ufs/util.h
@@ -231,6 +231,9 @@ ufs_set_inode_gid(struct super_block *sb
 
 extern dev_t ufs_get_inode_dev(struct super_block *, struct ufs_inode_info *);
 extern void ufs_set_inode_dev(struct super_block *, struct ufs_inode_info *, dev_t);
+extern int __ufs_write_begin(struct file *file, struct address_space *mapping,
+		loff_t pos, unsigned len, unsigned flags,
+		struct page **pagep, void **fsdata);
 
 /*
  * These functions manipulate ufs buffers

[patch 8/8] reiser4: fix for new aops patches

[Nick Piggin]

Index: linux-2.6/fs/reiser4/plugin/item/extent_file_ops.c
===================================================================
--- linux-2.6.orig/fs/reiser4/plugin/item/extent_file_ops.c
+++ linux-2.6/fs/reiser4/plugin/item/extent_file_ops.c
@@ -7,7 +7,6 @@
 
 #include <linux/quotaops.h>
 #include <linux/swap.h>
-#include "../../../../mm/filemap.h"
 
 static inline reiser4_extent *ext_by_offset(const znode *node, int offset)
 {
@@ -937,6 +936,31 @@ static int write_extent_reserve_space(st
 	return reiser4_grab_space(count, 0 /* flags */);
 }
 
+/*
+ * filemap_copy_from_user no longer exists in generic code, because it
+ * is deadlocky (copying from user while holding the page lock is bad).
+ * As a temporary fix for reiser4, just define it here.
+ */
+static inline size_t
+filemap_copy_from_user(struct page *page, unsigned long offset,
+			const char __user *buf, unsigned bytes)
+{
+	char *kaddr;
+	int left;
+
+	kaddr = kmap_atomic(page, KM_USER0);
+	left = __copy_from_user_inatomic_nocache(kaddr + offset, buf, bytes);
+	kunmap_atomic(kaddr, KM_USER0);
+
+	if (left != 0) {
+		/* Do it the slow way */
+		kaddr = kmap(page);
+		left = __copy_from_user_nocache(kaddr + offset, buf, bytes);
+		kunmap(page);
+	}
+	return bytes - left;
+}
+
 /**
  * reiser4_write_extent - write method of extent item plugin
  * @file: file to write to