Re: [OOPS] sysfs_hash_and_remove (was Re: What protection ....)

[Greg KH <greg@kroah.com>]

On Wed, Nov 23, 2005 at 06:22:13PM +0530, Maneesh Soni wrote:
> On Wed, Nov 23, 2005 at 09:18:45AM +0100, Ingo Molnar wrote:
> > 
> [..]
> > on a related note - i've been carrying the patch below in -rt for 2 
> > months (i.e. Steven's kernel has it too), as a workaround against the 
> > crash described below.
> > 
> [..]
> 
> > i'm occasionally getting the crash below on a PREEMPT_RT kernel. Might 
> > be a PREEMPT_RT bug, or might be some sysfs race only visible under 
> > PREEMPT_RT. Any ideas? The crash is at:
> > 
> > (gdb) list *0xc01a2095
> > 0xc01a2095 is in sysfs_hash_and_remove (fs/sysfs/inode.c:229).
> > 224     }
> > 225
> > 226     void sysfs_hash_and_remove(struct dentry * dir, const char * name)
> > 227     {
> > 228             struct sysfs_dirent * sd;
> > 229             struct sysfs_dirent * parent_sd = dir->d_fsdata;
> > 230
> > 231             if (dir->d_inode == NULL)
> > 232                     /* no inode means this hasn't been made visible yet */
> > 233                     return;
> > (gdb)
> > 
> Looks like here it is crashing due to bogus dentry pointer in the kobject 
> kobj->dentry. Could be some stale pointer? 

Did you ever figure anything out here?  I'm seeing a lot more reports of
this problem lately, especially if you enable slab debugging.  For
example:
	http://bugzilla.kernel.org/show_bug.cgi?id=5876

thanks,

greg k-h