orangefs: fix deadlock; do not write i_size in read_iter

After do_readv_writev, the inode cache is invalidated anyway, so i_size
will never be read.  It will be fetched from the server which will also
know about updates from other machines.

Fixes deadlock on 32-bit SMP.

See https://marc.info/?l=linux-fsdevel&m=151268557427760&w=2

Signed-off-by: Martin Brandenburg <martin@omnibond.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Mike Marshall <hubcap@omnibond.com>
Cc: stable@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

fs/orangefs/file.c

@@ -452,7 +452,7 @@ ssize_t orangefs_inode_read(struct inode *inode,
 static ssize_t orangefs_file_read_iter(struct kiocb *iocb, struct iov_iter *iter)
 {
 	struct file *file = iocb->ki_filp;
-	loff_t pos = *(&iocb->ki_pos);
+	loff_t pos = iocb->ki_pos;
 	ssize_t rc = 0;
 
 	BUG_ON(iocb->private);
@@ -492,9 +492,6 @@ static ssize_t orangefs_file_write_iter(struct kiocb *iocb, struct iov_iter *ite
 		}
 	}
 
-	if (file->f_pos > i_size_read(file->f_mapping->host))
-		orangefs_i_size_write(file->f_mapping->host, file->f_pos);
-
 	rc = generic_write_checks(iocb, iter);
 
 	if (rc <= 0) {
@@ -508,7 +505,7 @@ static ssize_t orangefs_file_write_iter(struct kiocb *iocb, struct iov_iter *ite
 	 * pos to the end of the file, so we will wait till now to set
 	 * pos...
 	 */
-	pos = *(&iocb->ki_pos);
+	pos = iocb->ki_pos;
 
 	rc = do_readv_writev(ORANGEFS_IO_WRITE,
 			     file,

fs/orangefs/orangefs-kernel.h

@@ -533,17 +533,6 @@ do {									\
 	sys_attr.mask = ORANGEFS_ATTR_SYS_ALL_SETABLE;			\
 } while (0)
 
-static inline void orangefs_i_size_write(struct inode *inode, loff_t i_size)
-{
-#if BITS_PER_LONG == 32 && defined(CONFIG_SMP)
-	inode_lock(inode);
-#endif
-	i_size_write(inode, i_size);
-#if BITS_PER_LONG == 32 && defined(CONFIG_SMP)
-	inode_unlock(inode);
-#endif
-}
-
 static inline void orangefs_set_timeout(struct dentry *dentry)
 {
 	unsigned long time = jiffies + orangefs_dcache_timeout_msecs*HZ/1000;