mirror O_APPEND and O_DIRECT into iocb->ki_flags

... avoiding write_iter/fcntl races.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>

11 files changed, 38 insertions, 23 deletions

diff --git a/fs/aio.c b/fs/aio.c
index 5785c4b58fea..e976185c8e5b 100644
--- a/fs/aio.c
+++ b/fs/aio.c
@@ -1502,7 +1502,7 @@ static int io_submit_one(struct kioctx *ctx, struct iocb __user *user_iocb,
 	}
 	req->common.ki_pos = iocb->aio_offset;
 	req->common.ki_complete = aio_complete;
-	req->common.ki_flags = 0;
+	req->common.ki_flags = iocb_flags(req->common.ki_filp);
 
 	if (iocb->aio_flags & IOCB_FLAG_RESFD) {
 		/*
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index c64d11c41eeb..faa7d390841b 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1794,7 +1794,7 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	if (sync)
 		atomic_inc(&BTRFS_I(inode)->sync_writers);
 
-	if (file->f_flags & O_DIRECT) {
+	if (iocb->ki_flags & IOCB_DIRECT) {
 		num_written = __btrfs_direct_write(iocb, from, pos);
 	} else {
 		num_written = __btrfs_buffered_write(file, from, pos);
diff --git a/fs/ceph/file.c b/fs/ceph/file.c
index 3f0b9339d823..b9b8eb225f66 100644
--- a/fs/ceph/file.c
+++ b/fs/ceph/file.c
@@ -457,7 +457,7 @@ static ssize_t ceph_sync_read(struct kiocb *iocb, struct iov_iter *i,
 	if (ret < 0)
 		return ret;
 
-	if (file->f_flags & O_DIRECT) {
+	if (iocb->ki_flags & IOCB_DIRECT) {
 		while (iov_iter_count(i)) {
 			size_t start;
 			ssize_t n;
@@ -828,7 +828,7 @@ again:
 		return ret;
 
 	if ((got & (CEPH_CAP_FILE_CACHE|CEPH_CAP_FILE_LAZYIO)) == 0 ||
-	    (iocb->ki_filp->f_flags & O_DIRECT) ||
+	    (iocb->ki_flags & IOCB_DIRECT) ||
 	    (fi->flags & CEPH_F_SYNC)) {
 
 		dout("aio_sync_read %p %llx.%llx %llu~%u got cap refs on %s\n",
@@ -995,12 +995,12 @@ retry_snap:
 	     inode, ceph_vinop(inode), pos, count, ceph_cap_string(got));
 
 	if ((got & (CEPH_CAP_FILE_BUFFER|CEPH_CAP_FILE_LAZYIO)) == 0 ||
-	    (file->f_flags & O_DIRECT) || (fi->flags & CEPH_F_SYNC)) {
+	    (iocb->ki_flags & IOCB_DIRECT) || (fi->flags & CEPH_F_SYNC)) {
 		struct iov_iter data;
 		mutex_unlock(&inode->i_mutex);
 		/* we might need to revert back to that point */
 		data = *from;
-		if (file->f_flags & O_DIRECT)
+		if (iocb->ki_flags & IOCB_DIRECT)
 			written = ceph_sync_direct_write(iocb, &data, pos);
 		else
 			written = ceph_sync_write(iocb, &data, pos);
diff --git a/fs/ext4/file.c b/fs/ext4/file.c
index c10785f10d1d..53bbc0b1995f 100644
--- a/fs/ext4/file.c
+++ b/fs/ext4/file.c
@@ -95,7 +95,7 @@ ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	struct inode *inode = file_inode(iocb->ki_filp);
 	struct mutex *aio_mutex = NULL;
 	struct blk_plug plug;
-	int o_direct = io_is_direct(file);
+	int o_direct = iocb->ki_flags & IOCB_DIRECT;
 	int overwrite = 0;
 	ssize_t ret;
 
@@ -106,7 +106,7 @@ ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (o_direct &&
 	    ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
 	    !is_sync_kiocb(iocb) &&
-	    (file->f_flags & O_APPEND ||
+	    (iocb->ki_flags & IOCB_APPEND ||
 	     ext4_unaligned_aio(inode, from, iocb->ki_pos))) {
 		aio_mutex = ext4_aio_mutex(inode);
 		mutex_lock(aio_mutex);
diff --git a/fs/fuse/file.c b/fs/fuse/file.c
index b86c8e08399a..5ef05b5c4cff 100644
--- a/fs/fuse/file.c
+++ b/fs/fuse/file.c
@@ -1177,7 +1177,7 @@ static ssize_t fuse_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (err)
 		goto out;
 
-	if (file->f_flags & O_DIRECT) {
+	if (iocb->ki_flags & IOCB_DIRECT) {
 		loff_t pos = iocb->ki_pos;
 		written = generic_file_direct_write(iocb, from, pos);
 		if (written < 0 || !iov_iter_count(from))
diff --git a/fs/gfs2/file.c b/fs/gfs2/file.c
index 614bb42cb7e1..08329afa1339 100644
--- a/fs/gfs2/file.c
+++ b/fs/gfs2/file.c
@@ -709,7 +709,7 @@ static ssize_t gfs2_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 
 	gfs2_size_hint(file, iocb->ki_pos, iov_iter_count(from));
 
-	if (file->f_flags & O_APPEND) {
+	if (iocb->ki_flags & IOCB_APPEND) {
 		struct gfs2_holder gh;
 
 		ret = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, 0, &gh);
diff --git a/fs/nfs/file.c b/fs/nfs/file.c
index f6a3adedf027..14364dc001f7 100644
--- a/fs/nfs/file.c
+++ b/fs/nfs/file.c
@@ -170,7 +170,7 @@ nfs_file_read(struct kiocb *iocb, struct iov_iter *to)
 	struct inode *inode = file_inode(iocb->ki_filp);
 	ssize_t result;
 
-	if (iocb->ki_filp->f_flags & O_DIRECT)
+	if (iocb->ki_flags & IOCB_DIRECT)
 		return nfs_file_direct_read(iocb, to, iocb->ki_pos);
 
 	dprintk("NFS: read(%pD2, %zu@%lu)\n",
@@ -680,7 +680,7 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 	if (result)
 		return result;
 
-	if (file->f_flags & O_DIRECT)
+	if (iocb->ki_flags & IOCB_DIRECT)
 		return nfs_file_direct_write(iocb, from, pos);
 
 	dprintk("NFS: write(%pD2, %zu@%Ld)\n",
@@ -692,7 +692,7 @@ ssize_t nfs_file_write(struct kiocb *iocb, struct iov_iter *from)
 	/*
 	 * O_APPEND implies that we must revalidate the file length.
 	 */
-	if (file->f_flags & O_APPEND) {
+	if (iocb->ki_flags & IOCB_APPEND) {
 		result = nfs_revalidate_file_size(inode, file);
 		if (result)
 			goto out;
diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c
index b93919f50f0f..cd37f6cd4d51 100644
--- a/fs/ocfs2/file.c
+++ b/fs/ocfs2/file.c
@@ -2274,8 +2274,8 @@ static ssize_t ocfs2_file_write_iter(struct kiocb *iocb,
 	if (count == 0)
 		return 0;
 
-	appending = file->f_flags & O_APPEND ? 1 : 0;
-	direct_io = file->f_flags & O_DIRECT ? 1 : 0;
+	appending = iocb->ki_flags & IOCB_APPEND ? 1 : 0;
+	direct_io = iocb->ki_flags & IOCB_DIRECT ? 1 : 0;
 
 	mutex_lock(&inode->i_mutex);
 
@@ -2429,7 +2429,7 @@ relock:
 
 out_dio:
 	/* buffered aio wouldn't have proper lock coverage today */
-	BUG_ON(ret == -EIOCBQUEUED && !(file->f_flags & O_DIRECT));
+	BUG_ON(ret == -EIOCBQUEUED && !(iocb->ki_flags & IOCB_DIRECT));
 
 	if (unlikely(written <= 0))
 		goto no_sync;
@@ -2546,7 +2546,7 @@ static ssize_t ocfs2_file_read_iter(struct kiocb *iocb,
 	 * buffered reads protect themselves in ->readpage().  O_DIRECT reads
 	 * need locks to protect pending reads from racing with truncate.
 	 */
-	if (filp->f_flags & O_DIRECT) {
+	if (iocb->ki_flags & IOCB_DIRECT) {
 		have_alloc_sem = 1;
 		ocfs2_iocb_set_sem_locked(iocb);
 
@@ -2580,7 +2580,7 @@ static ssize_t ocfs2_file_read_iter(struct kiocb *iocb,
 	trace_generic_file_aio_read_ret(ret);
 
 	/* buffered aio wouldn't have proper lock coverage today */
-	BUG_ON(ret == -EIOCBQUEUED && !(filp->f_flags & O_DIRECT));
+	BUG_ON(ret == -EIOCBQUEUED && !(iocb->ki_flags & IOCB_DIRECT));
 
 	/* see ocfs2_file_write_iter */
 	if (ret == -EIOCBQUEUED || !ocfs2_iocb_is_rw_locked(iocb)) {
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 28d157807b42..1f12ad0a8585 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -279,7 +279,7 @@ xfs_file_read_iter(
 
 	XFS_STATS_INC(xs_read_calls);
 
-	if (unlikely(file->f_flags & O_DIRECT))
+	if (unlikely(iocb->ki_flags & IOCB_DIRECT))
 		ioflags |= XFS_IO_ISDIRECT;
 	if (file->f_mode & FMODE_NOCMTIME)
 		ioflags |= XFS_IO_INVIS;
@@ -804,7 +804,7 @@ xfs_file_write_iter(
 	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
 		return -EIO;
 
-	if (unlikely(file->f_flags & O_DIRECT))
+	if (unlikely(iocb->ki_flags & IOCB_DIRECT))
 		ret = xfs_file_dio_aio_write(iocb, from);
 	else
 		ret = xfs_file_buffered_aio_write(iocb, from);
diff --git a/include/linux/fs.h b/include/linux/fs.h
index b4aa400ac723..b1d7db28c13c 100644
--- a/include/linux/fs.h
+++ b/include/linux/fs.h
@@ -315,6 +315,8 @@ struct address_space;
 struct writeback_control;
 
 #define IOCB_EVENTFD		(1 << 0)
+#define IOCB_APPEND		(1 << 1)
+#define IOCB_DIRECT		(1 << 2)
 
 struct kiocb {
 	struct file		*ki_filp;
@@ -329,10 +331,13 @@ static inline bool is_sync_kiocb(struct kiocb *kiocb)
 	return kiocb->ki_complete == NULL;
 }
 
+static inline int iocb_flags(struct file *file);
+
 static inline void init_sync_kiocb(struct kiocb *kiocb, struct file *filp)
 {
 	*kiocb = (struct kiocb) {
 		.ki_filp = filp,
+		.ki_flags = iocb_flags(filp),
 	};
 }
 
@@ -2779,6 +2784,16 @@ static inline bool io_is_direct(struct file *filp)
 	return (filp->f_flags & O_DIRECT) || IS_DAX(file_inode(filp));
 }
 
+static inline int iocb_flags(struct file *file)
+{
+	int res = 0;
+	if (file->f_flags & O_APPEND)
+		res |= IOCB_APPEND;
+	if (io_is_direct(file))
+		res |= IOCB_DIRECT;
+	return res;
+}
+
 static inline ino_t parent_ino(struct dentry *dentry)
 {
 	ino_t res;
diff --git a/mm/filemap.c b/mm/filemap.c
index 243997a26e7c..405de370e657 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -1694,7 +1694,7 @@ generic_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
 	loff_t *ppos = &iocb->ki_pos;
 	loff_t pos = *ppos;
 
-	if (io_is_direct(file)) {
+	if (iocb->ki_flags & IOCB_DIRECT) {
 		struct address_space *mapping = file->f_mapping;
 		struct inode *inode = mapping->host;
 		size_t count = iov_iter_count(iter);
@@ -2271,7 +2271,7 @@ inline ssize_t generic_write_checks(struct kiocb *iocb, struct iov_iter *from)
 		return 0;
 
 	/* FIXME: this is for backwards compatibility with 2.4 */
-	if (file->f_flags & O_APPEND)
+	if (iocb->ki_flags & IOCB_APPEND)
 		iocb->ki_pos = i_size_read(inode);
 
 	pos = iocb->ki_pos;
@@ -2545,7 +2545,7 @@ ssize_t __generic_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
 	if (err)
 		goto out;
 
-	if (io_is_direct(file)) {
+	if (iocb->ki_flags & IOCB_DIRECT) {
 		loff_t pos, endbyte;
 
 		written = generic_file_direct_write(iocb, from, iocb->ki_pos);