Attachment 11002 Details for Bug 11243 – Fix for master

[patch] Fix for master

streams.patch (text/plain), 2.18 KB, created by Ralph Böhme on 2015-04-29 15:41:29 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Ralph Böhme

Created: 2015-04-29 15:41:29 UTC

Size: 2.18 KB

patch

obsolete

>From 7daf165561107684e2890732bce81c6d89639ebc Mon Sep 17 00:00:00 2001
>From: Ralph Boehme <slow@samba.org>
>Date: Wed, 29 Apr 2015 16:53:04 +0200
>Subject: [PATCH] vfs: kernel_flock and named streams
>
>Streams implementing VFS modules may implement streams in a way that the
>fsp will have the basefile open in the fsp fd, so lacking a distinct fd
>for the stream, kernel_flock will apply on the basefile which is
>wrong. The actual check is deffered to the VFS module implementing the
>kernel_flock call.
>
>Bug: https://bugzilla.samba.org/show_bug.cgi?id=11243
>
>Signed-off-by: Ralph Boehme <slow@samba.org>
>---
> source3/modules/vfs_gpfs.c | 10 ++++++++++
> source3/smbd/open.c        |  9 +++++++++
> 2 files changed, 19 insertions(+)
>
>diff --git a/source3/modules/vfs_gpfs.c b/source3/modules/vfs_gpfs.c
>index aa07060..aa61d35 100644
>--- a/source3/modules/vfs_gpfs.c
>+++ b/source3/modules/vfs_gpfs.c
>@@ -135,6 +135,16 @@ static int vfs_gpfs_kernel_flock(vfs_handle_struct *handle, files_struct *fsp,
> 		return 0;
> 	}
> 
>+	/*
>+	 * A named stream fsp will have the basefile open in the fsp
>+	 * fd, so lacking a distinct fd for the stream we have to skip
>+	 * kernel_flock and set_gpfs_sharemode for stream.
>+	 */
>+	if (!is_ntfs_default_stream_smb_fname(fsp->fsp_name)) {
>+		DEBUG(2,("%s: kernel_flock on stream\n", fsp_str_dbg(fsp)));
>+		return 0;
>+	}
>+
> 	kernel_flock(fsp->fh->fd, share_mode, access_mask);
> 
> 	if (!set_gpfs_sharemode(fsp, access_mask, fsp->share_access)) {
>diff --git a/source3/smbd/open.c b/source3/smbd/open.c
>index 773b146..86c1c31 100644
>--- a/source3/smbd/open.c
>+++ b/source3/smbd/open.c
>@@ -3064,6 +3064,15 @@ static NTSTATUS open_file_ntcreate(connection_struct *conn,
> 
> 	if (fsp->fh->fd != -1 && lp_kernel_share_modes(SNUM(conn))) {
> 		int ret_flock;
>+		/*
>+		 * Beware: streams implementing VFS modules may
>+		 * implement streams in a way that fsp will have the
>+		 * basefile open in the fsp fd, so lacking a distinct
>+		 * fd for the stream kernel_flock will apply on the
>+		 * basefile which is wrong. The actual check is
>+		 * deffered to the VFS module implementing the
>+		 * kernel_flock call.
>+		 */
> 		ret_flock = SMB_VFS_KERNEL_FLOCK(fsp, share_access, access_mask);
> 		if(ret_flock == -1 ){
> 
>-- 
>2.1.0
>

Flags:

vl: review+
jra: review+

Actions: View

Attachments on bug 11243: 11002 | 11003 | 11004 | 11579 | 11585