[V9fs-developer] fs/9p: Don't use O_TRUNC flag in TOPEN and TLOPEN request

Commit Message

Aneesh Kumar K.V Feb. 5, 2013, 5:56 a.m. UTC

From: "Aneesh Kumar K.V" <[e-mail]>

We do the truncate via setattr request, hence don't pass the O_TRUNC flag in
open request. Without this patch we end up sending zero sized write request
to server when we try to truncate. Some servers (VirtFS) were not handling that
properly.

Reported-by: M. Mohan Kumar <[e-mail]>
Signed-off-by: Aneesh Kumar K.V <[e-mail]>
---
 fs/9p/vfs_file.c       |    4 ----
 fs/9p/vfs_inode.c      |    3 ---
 fs/9p/vfs_inode_dotl.c |    1 -
 3 files changed, 8 deletions(-)

Patch

diff --git a/fs/9p/vfs_file.c b/fs/9p/vfs_file.c
index 3356e3e..9aa559d 100644
--- a/fs/9p/vfs_file.c
+++ b/fs/9p/vfs_file.c
@@ -80,10 +80,6 @@  int v9fs_file_open(struct inode *inode, struct file *file)
 			p9_client_clunk(fid);
 			return err;
 		}
-		if (file->f_flags & O_TRUNC) {
-			i_size_write(inode, 0);
-			inode->i_blocks = 0;
-		}
 		if ((file->f_flags & O_APPEND) &&
 			(!v9fs_proto_dotu(v9ses) && !v9fs_proto_dotl(v9ses)))
 			generic_file_llseek(file, 0, SEEK_END);
diff --git a/fs/9p/vfs_inode.c b/fs/9p/vfs_inode.c
index 890bed5..57d017a 100644
--- a/fs/9p/vfs_inode.c
+++ b/fs/9p/vfs_inode.c
@@ -192,9 +192,6 @@  int v9fs_uflags2omode(int uflags, int extended)
 		break;
 	}
 
-	if (uflags & O_TRUNC)
-		ret |= P9_OTRUNC;
-
 	if (extended) {
 		if (uflags & O_EXCL)
 			ret |= P9_OEXCL;
diff --git a/fs/9p/vfs_inode_dotl.c b/fs/9p/vfs_inode_dotl.c
index 7c29558..3eaca8b 100644
--- a/fs/9p/vfs_inode_dotl.c
+++ b/fs/9p/vfs_inode_dotl.c
@@ -186,7 +186,6 @@  static int v9fs_mapped_dotl_flags(int flags)
 		{ O_CREAT,	P9_DOTL_CREATE },
 		{ O_EXCL,	P9_DOTL_EXCL },
 		{ O_NOCTTY,	P9_DOTL_NOCTTY },
-		{ O_TRUNC,	P9_DOTL_TRUNC },
 		{ O_APPEND,	P9_DOTL_APPEND },
 		{ O_NONBLOCK,	P9_DOTL_NONBLOCK },
 		{ O_DSYNC,	P9_DOTL_DSYNC },