Re: nfsd_rename() multiple export problem

[Peter Staubach <staubach@redhat.com>]

Boris Zuckerman wrote:
> Hi,
> 
>  
> 
> This is probably a question for Peter Staubach. However, may be somebody
> else can help me too…
> 
>  
> 
> Once in a while under heavy load nfsd returns a NFSERR_XDEV from rename
> checking following condition:
> 
>  
> 
>       err = (rqstp->rq_vers == 2) ? nfserr_acces : nfserr_xdev;
> 
>       if (ffhp->fh_export != tfhp->fh_export)
> 
>             goto out;
> 
>  
> 
> Note that several lines later the similar check looks like this:
> 
>  
> 
>       host_err = -EXDEV;
> 
>       if (ffhp->fh_export->ex_path.mnt != tfhp->fh_export->ex_path.mnt)
> 
>             goto out_dput_new;
> 
>  
> 
> As you can see the code is not making any longer assumptions that
> fh_export must match, but rather is checking ex_path.mnt.
> 
> We reproduced the problem in-house and got following information: the
> source and the target directories are the same; Ibrix resolved them to
> the same dentry.
> 
> However, there are 2 different fh_exports. I noticed that one of them
> has expiry_time = 0x0 and  another has expiry_time = 0x4a6aee6d
> 
> Is it significant? Below is the complete dump of exports and svc_fh
> structurs. Please, let me know if you need more information.
> 
> I have the crash dump and can dig out anything you need.
> 

What version of the Linux kernel are you using?

	Thanx...

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