Re: [PATCH] soc: qcom: socinfo: Open read access to all for debugfs

[Doug Anderson]

Hi,

On Mon, Jan 11, 2021 at 6:04 PM Stephen Boyd <swboyd@xxxxxxxxxxxx> wrote:
>
> Quoting Doug Anderson (2021-01-05 15:06:35)
> > Hi,
> >
> > On Tue, Dec 15, 2020 at 11:19 PM Stephen Boyd <swboyd@xxxxxxxxxxxx> wrote:
> > >
> > > There doesn't seem to be any reason to limit this to only root user
> > > readable. Let's make it readable by all so that random programs can
> > > read the debugfs files in here instead of just root. The information is
> > > just that, informational, so this is fine.
> > >
> > > Cc: Sai Prakash Ranjan <saiprakash.ranjan@xxxxxxxxxxxxxx>
> > > Cc: Douglas Anderson <dianders@xxxxxxxxxxxx>
> > > Cc: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
> > > Signed-off-by: Stephen Boyd <swboyd@xxxxxxxxxxxx>
> > > ---
> > >  drivers/soc/qcom/socinfo.c | 40 +++++++++++++++++++-------------------
> > >  1 file changed, 20 insertions(+), 20 deletions(-)
> >
> > One worry I'd have is whether there would ever be any PII (personally
> > identifiable information) here, like maybe a chip serial number.  If
> > so, is that something that is OK to provide to any random process?
> > ...maybe I'm just being paranoid though, since presumably there are
> > enough random HW characteristics that could be strung together and
> > hashed to make roughly a unique ID anyway and hiding every HW
> > characteristic would be a bit extreme...
> >
>
> I suppose if that's a problem then the process that mounts debugfs can
> change the access to restrict it. I'm not aware of this needing to be a
> kernel policy though, so I'd rather leave that up to userspace to decide
> if it should be restricted further.

OK, fair enough.

Reviewed-by: Douglas Anderson <dianders@xxxxxxxxxxxx>