Patchwork ext4: Save and restore state flags in EXT4_IOC_SETFLAGS.

login
register
mail settings
Submitter Theodore Ts'o
Date June 1, 2012, 3:48 a.m.
Message ID <20120601034820.GC7897@thunk.org>
Download mbox | patch
Permalink /patch/162264/
State Accepted
Headers show

Comments

Theodore Ts'o - June 1, 2012, 3:48 a.m.
On Fri, May 04, 2012 at 11:37:49PM +0800, Tao Ma wrote:
> From: Tao Ma <boyu.mt@taobao.com>
> 
> In commit 353eb83c we removes i_state_flags with 64-bit longs,
> But in case we call EXT4_IOC_SETFLAGS, we fail to save the
> high 32-bit state flags and only stores the low 32-bit back
> to ei->i_flags. So the state flags are missing now in 64-bit
> long architecture.

The problem with this approach is that we are still editing i_flags
and then replacing it with the new value.  So we're vulnerable to
races where some other process is modifies the i_state_flags between
when we sample it using ext4_save_state_flags() and when we restore
them.

This is a better way to fix the problem, and what I plan to commit
into the ext4 tree:

commit 79906964a187c405db72a3abc60eb9b50d804fbc
Author: Theodore Ts'o <tytso@mit.edu>
Date:   Thu May 31 23:46:01 2012 -0400

    ext4: don't trash state flags in EXT4_IOC_SETFLAGS
    
    In commit 353eb83c we removed i_state_flags with 64-bit longs, But
    when handling the EXT4_IOC_SETFLAGS ioctl, we replace i_flags
    directly, which trashes the state flags which are stored in the high
    32-bits of i_flags on 64-bit platforms.  So use the the
    ext4_{set,clear}_inode_flags() functions which use atomic bit
    manipulation functions instead.
    
    Reported-by: Tao Ma <boyu.mt@taobao.com>
    Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
    Cc: stable@kernel.org



--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Tao Ma - June 4, 2012, 3:22 a.m.
Hi Ted,
	Sorry for the delay.
On 06/01/2012 11:48 AM, Ted Ts'o wrote:
> On Fri, May 04, 2012 at 11:37:49PM +0800, Tao Ma wrote:
>> From: Tao Ma <boyu.mt@taobao.com>
>>
>> In commit 353eb83c we removes i_state_flags with 64-bit longs,
>> But in case we call EXT4_IOC_SETFLAGS, we fail to save the
>> high 32-bit state flags and only stores the low 32-bit back
>> to ei->i_flags. So the state flags are missing now in 64-bit
>> long architecture.
> 
> The problem with this approach is that we are still editing i_flags
> and then replacing it with the new value.  So we're vulnerable to
> races where some other process is modifies the i_state_flags between
> when we sample it using ext4_save_state_flags() and when we restore
> them.
> 
> This is a better way to fix the problem, and what I plan to commit
> into the ext4 tree:
> 
> commit 79906964a187c405db72a3abc60eb9b50d804fbc
> Author: Theodore Ts'o <tytso@mit.edu>
> Date:   Thu May 31 23:46:01 2012 -0400
> 
>     ext4: don't trash state flags in EXT4_IOC_SETFLAGS
>     
>     In commit 353eb83c we removed i_state_flags with 64-bit longs, But
>     when handling the EXT4_IOC_SETFLAGS ioctl, we replace i_flags
>     directly, which trashes the state flags which are stored in the high
>     32-bits of i_flags on 64-bit platforms.  So use the the
>     ext4_{set,clear}_inode_flags() functions which use atomic bit
>     manipulation functions instead.
>     
>     Reported-by: Tao Ma <boyu.mt@taobao.com>
>     Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
>     Cc: stable@kernel.org
> 
> diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
> index feba55a..8ad112a 100644
> --- a/fs/ext4/ioctl.c
> +++ b/fs/ext4/ioctl.c
> @@ -38,7 +38,7 @@ long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
>  		handle_t *handle = NULL;
>  		int err, migrate = 0;
>  		struct ext4_iloc iloc;
> -		unsigned int oldflags;
> +		unsigned int oldflags, mask, i;
>  		unsigned int jflag;
>  
>  		if (!inode_owner_or_capable(inode))
> @@ -115,8 +115,14 @@ long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
>  		if (err)
>  			goto flags_err;
>  
> -		flags = flags & EXT4_FL_USER_MODIFIABLE;
> -		flags |= oldflags & ~EXT4_FL_USER_MODIFIABLE;
> +		for (i = 0, mask = 1; i < 32; i++, mask <<= 1) {
> +			if (!(mask & EXT4_FL_USER_MODIFIABLE))
> +				continue;
> +			if (mask & flags)
> +				ext4_set_inode_flag(inode, i);
> +			else
> +				ext4_clear_inode_flag(inode, i);
> +		}
>  		ei->i_flags = flags;
You forget to remove this line and we can still get the problem.

I am busy with other stuff and when I got to it, it is already in linus'
tree. So I have sent out another patch to remove this line and after
that my test case passed. The patch subject is "[PATCH] ext4: Don't set
i_flags in EXT4_IOC_SETFLAGS."

Thanks
Tao
>  
>  		ext4_set_inode_flags(inode);
> 
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Patch

diff --git a/fs/ext4/ioctl.c b/fs/ext4/ioctl.c
index feba55a..8ad112a 100644
--- a/fs/ext4/ioctl.c
+++ b/fs/ext4/ioctl.c
@@ -38,7 +38,7 @@  long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 		handle_t *handle = NULL;
 		int err, migrate = 0;
 		struct ext4_iloc iloc;
-		unsigned int oldflags;
+		unsigned int oldflags, mask, i;
 		unsigned int jflag;
 
 		if (!inode_owner_or_capable(inode))
@@ -115,8 +115,14 @@  long ext4_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 		if (err)
 			goto flags_err;
 
-		flags = flags & EXT4_FL_USER_MODIFIABLE;
-		flags |= oldflags & ~EXT4_FL_USER_MODIFIABLE;
+		for (i = 0, mask = 1; i < 32; i++, mask <<= 1) {
+			if (!(mask & EXT4_FL_USER_MODIFIABLE))
+				continue;
+			if (mask & flags)
+				ext4_set_inode_flag(inode, i);
+			else
+				ext4_clear_inode_flag(inode, i);
+		}
 		ei->i_flags = flags;
 
 		ext4_set_inode_flags(inode);