linux-kernel - Re: NFS infinite loop in filemap

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 8 May 2008 09:07:59 -0400
From:	Josef Bacik <jbacik@...hat.com>
To:	Miklos Szeredi <miklos@...redi.hu>
Cc:	trond.myklebust@....uio.no, npiggin@...e.de,
	linux-kernel@...r.kernel.org, linux-fsdevel@...r.kernel.org,
	linux-mm@...ck.org
Subject: Re: NFS infinite loop in filemap_fault()

On Thu, May 08, 2008 at 08:47:09AM +0200, Miklos Szeredi wrote:
> > Page fault on NFS apparently goes into an infinite loop if the read on
> > the server fails.
> > 
> > I don't understand the NFS readpage code, but the filemap_fault() code
> > looks somewhat suspicious:
> > 
> > 	/*
> > 	 * Umm, take care of errors if the page isn't up-to-date.
> > 	 * Try to re-read it _once_. We do this synchronously,
> > 	 * because there really aren't any performance issues here
> > 	 * and we need to check for errors.
> > 	 */
> > 	ClearPageError(page);
> > 	error = mapping->a_ops->readpage(file, page);
> > 	page_cache_release(page);
> > 
> > 	if (!error || error == AOP_TRUNCATED_PAGE)
> > 		goto retry_find;
> > 
> > The comment doesn't seem to match what the it actually does: if
> > ->readpage() is asynchronous, then this will just repeat everything,
> > without any guarantee that it will re-read once.
> 
> This patch fixes it.  It's probably wrong in some subtle way though...
> 
> Miklos
> 
> 
> ---
>  mm/filemap.c |    6 ++++++
>  1 file changed, 6 insertions(+)
> 
> Index: linux.git/mm/filemap.c
> ===================================================================
> --- linux.git.orig/mm/filemap.c	2008-05-08 08:17:22.000000000 +0200
> +++ linux.git/mm/filemap.c	2008-05-08 08:19:59.000000000 +0200
> @@ -1461,6 +1461,12 @@ page_not_uptodate:
>  	 */
>  	ClearPageError(page);
>  	error = mapping->a_ops->readpage(file, page);
> +	if (!error && !PageUptodate(page)) {

Shouldn't you have (!error || error != AOP_TRUNCATED_PAGE), since the fs can
return AOP_TRUNCATED_PAGE if it needs vfs to try the readpage again?  Things
like OCFS2/GFS2 do this, they send of a lock request and return
AOP_TRUNCATED_PAGE so that when we come back into readpage we are already
holding our lock without blocking while holding the page lock.  Thanks,

Josef
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/