Re: [PATCH V3 1/8] sysfs: Add sysfs_emit and sysfs_emit_at to format sysfs output

From: Kees Cook
Date: Sat Oct 03 2020 - 18:15:35 EST


On Sat, Oct 03, 2020 at 03:55:51PM +0200, Greg Kroah-Hartman wrote:
> On Fri, Oct 02, 2020 at 03:30:30PM -0700, Kees Cook wrote:
> > On Wed, Sep 30, 2020 at 09:22:19PM -0700, Joe Perches wrote:
> > > On Wed, 2020-09-30 at 21:17 -0700, Kees Cook wrote:
> > > > On Wed, Sep 30, 2020 at 01:57:40PM +0200, Greg Kroah-Hartman wrote:
> > > > > Kees, and Rafael, I don't know if you saw this proposal from Joe for
> > > > > sysfs files, questions below:
> > > >
> > > > I'm a fan. I think the use of sprintf() in sysfs might have been one of
> > > > my earliest complaints about unsafe code patterns in the kernel. ;)
> > > []
> > > > > > + if (WARN(!buf || offset_in_page(buf),
> > > > > > + "invalid sysfs_emit: buf:%p\n", buf))
> > >
> > > The dump_stack() is also going to emit pointers
> > > so I don't see how it does anything but help
> > > show where the buffer was. It is hashed...
> >
> > dump_stack() is going to report symbols and register contents.
> >
> > I was just pointing out that %p has no value here[1]. The interesting
> > states are: "was it NULL?" "how offset was it?". Its actual content
> > won't matter.
>
> Ok, suggestions for a better error message are always welcome :)

... I did. :P

https://lore.kernel.org/lkml/202009302108.18B05CA38@keescook/

But it doesn't need to hold up the series; %p is hashed, so I don't
care. It's just that the mandate from Linus was to not add new %p uses.
:P

--
Kees Cook