Re: [patch 00/13] devtmpfs patches

[Stephen Smalley <sds@tycho.nsa.gov>]

On Wed, 2009-05-13 at 16:20 +0200, Kay Sievers wrote:
> On Wed, May 13, 2009 at 15:34, Kay Sievers <kay.sievers@vrfy.org> wrote:
> 
> >>> +     kern_cred = prepare_kernel_cred(NULL);
> >>
> >> If you have no intention of altering the credentials you create, you might
> >> want to use &init_cred instead of kern_cred.  That said, you might want to
> >> allocate it and let the security module alter it before you use it.
> >
> > Ah, didn't know that it was exported. It's the one in
> > include/linux/init_task.h, right? I'll give that a try.
> 
> That seems to work fine here. Thanks a lot for the help.
> 
> Maybe we could do the same credential swap in sysfs, and get rid of:
> /**
>  * lookup_one_noperm - bad hack for sysfs
>  * @name:       pathname component to lookup
>  * @base:       base directory to lookup from
>  *
>  * This is a variant of lookup_one_len that doesn't perform any permission
>  * checks.   It's a horrible hack to work around the braindead sysfs
>  * architecture and should not be used anywhere else.
>  *
>  * DON'T USE THIS FUNCTION EVER, thanks.
>  */
> 
> in fs/namei.c?
> 
> Seems a bit odd to have a vfs function for a single filesystem, called
> from a single location, and annotated as "do not use". Christoph added
> the comment a while ago, so adding him to Cc:.

Yes, that makes sense to me as well - we didn't have the credentials
infrastructure in place at the time that lookup_one_noperm was
introduced, but switching the credentials around a normal lookup_one_len
call should work now.

-- 
Stephen Smalley
National Security Agency