Re: 2.6.22-rc2-mm1 NTFS & SLUB related fix

[Andrew Morton <akpm@linux-foundation.org>]

On Fri, 25 May 2007 05:22:50 +0000 "young dave" <hidave.darkstar@gmail.com> wrote:

> Hi,
> 
> > Is this ntfs_init_locked_inode?
> 
> Yes, it is.
> 
> > >  Bytes b4 0xc2959e28:  00 00 00 00 00 00 00 00 5a 5a 5a 5a 5a 5a 5a
> > >    Object 0xc2959e38:  24 00 51 00 00 00 6b a5
> > >   Redzone 0xc2959e40:  00 00 cc cc
> >
> > First two bytes after the object overwritten. The allocation for this
> > object should have been two bytes longer.
> >
> > > Last alloc: ntfs_init_locked_inode+0x9e/0x110 jiffies_ago=5140 cpu=0 pid=1604
> >
> > This is the function that allocated a too short object.
> >
> 
> Only the last one byte of  the string  is zeroed, but It malloced 2
> more byte appended the string because size of thentfschar type is 2
> bytes , is this the reason? But why?
> 

Thing is, ntfs_inode.name[] is an array of le16's.  But local variable `i'
in there is a byte index, not an le16 index.  We end up writing that 0x0000
at twice the intended offset.

So I think this was meant:

--- a/fs/ntfs/inode.c~a
+++ a/fs/ntfs/inode.c
@@ -140,7 +140,7 @@ static int ntfs_init_locked_inode(struct
 		if (!ni->name)
 			return -ENOMEM;
 		memcpy(ni->name, na->name, i);
-		ni->name[i] = 0;
+		ni->name[na->name_len] = 0;
 	}
 	return 0;
 }
_