xfs
[Top] [All Lists]

Re: [PATCH] use proper mnt_want_write calls for handle ioctls

To: xfs@xxxxxxxxxxx
Subject: Re: [PATCH] use proper mnt_want_write calls for handle ioctls
From: Christoph Hellwig <hch@xxxxxx>
Date: Fri, 12 Sep 2008 10:44:31 +0200
In-reply-to: <20080814203952.GB18704@xxxxxx>
References: <20080814203952.GB18704@xxxxxx>
User-agent: Mutt/1.3.28i
ping?  Without this patch the CVS tree won't properly work with
read-only bind mounts.

On Thu, Aug 14, 2008 at 10:39:52PM +0200, Christoph Hellwig wrote:
> Since 2.6.26 all writes to filesystems need to be enclosed by a
> mnt_want_write / mnt_drop_write pair instead of checking for IS_RDONLY.
> 
> XFs was updated for this in mainline but the changes to xfs_ioctl.c
> were never megred back into the CVS tree.
> 
> The original commit introducing this was:
> 
> commit 42a74f206b914db13ee1f5ae932dcd91a77c8579
> Author: Dave Hansen <haveblue@xxxxxxxxxx>
> Date:   Fri Feb 15 14:37:46 2008 -0800
> 
>     [PATCH] r/o bind mounts: elevate write count for ioctls()
> 
> Index: linux-2.6-xfs/fs/xfs/linux-2.6/xfs_ioctl.c
> ===================================================================
> --- linux-2.6-xfs.orig/fs/xfs/linux-2.6/xfs_ioctl.c   2008-08-14 
> 14:54:53.000000000 -0300
> +++ linux-2.6-xfs/fs/xfs/linux-2.6/xfs_ioctl.c        2008-08-14 
> 15:15:49.000000000 -0300
> @@ -543,8 +543,6 @@ xfs_attrmulti_attr_set(
>       char                    *kbuf;
>       int                     error = EFAULT;
>  
> -     if (IS_RDONLY(inode))
> -             return -EROFS;
>       if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
>               return EPERM;
>       if (len > XATTR_SIZE_MAX)
> @@ -570,8 +568,6 @@ xfs_attrmulti_attr_remove(
>       char                    *name,
>       __uint32_t              flags)
>  {
> -     if (IS_RDONLY(inode))
> -             return -EROFS;
>       if (IS_IMMUTABLE(inode) || IS_APPEND(inode))
>               return EPERM;
>       return xfs_attr_remove(XFS_I(inode), name, flags);
> @@ -581,6 +577,7 @@ STATIC int
>  xfs_attrmulti_by_handle(
>       xfs_mount_t             *mp,
>       void                    __user *arg,
> +     struct file             *parfilp,
>       struct inode            *parinode)
>  {
>       int                     error;
> @@ -634,13 +631,21 @@ xfs_attrmulti_by_handle(
>                                       &ops[i].am_length, ops[i].am_flags);
>                       break;
>               case ATTR_OP_SET:
> +                     ops[i].am_error = mnt_want_write(parfilp->f_path.mnt);
> +                     if (ops[i].am_error)
> +                             break;
>                       ops[i].am_error = xfs_attrmulti_attr_set(inode,
>                                       attr_name, ops[i].am_attrvalue,
>                                       ops[i].am_length, ops[i].am_flags);
> +                     mnt_drop_write(parfilp->f_path.mnt);
>                       break;
>               case ATTR_OP_REMOVE:
> +                     ops[i].am_error = mnt_want_write(parfilp->f_path.mnt);
> +                     if (ops[i].am_error)
> +                             break;
>                       ops[i].am_error = xfs_attrmulti_attr_remove(inode,
>                                       attr_name, ops[i].am_flags);
> +                     mnt_drop_write(parfilp->f_path.mnt);
>                       break;
>               default:
>                       ops[i].am_error = EINVAL;
> @@ -1431,7 +1436,7 @@ xfs_ioctl(
>               return xfs_attrlist_by_handle(mp, arg, inode);
>  
>       case XFS_IOC_ATTRMULTI_BY_HANDLE:
> -             return xfs_attrmulti_by_handle(mp, arg, inode);
> +             return xfs_attrmulti_by_handle(mp, arg, filp, inode);
>  
>       case XFS_IOC_SWAPEXT: {
>               error = xfs_swapext((struct xfs_swapext __user *)arg);
---end quoted text---

<Prev in Thread] Current Thread [Next in Thread>