On Sat, Mar 5, 2022 at 2:07 AM Benjamin Tissoires
<benjamin.tissoires@xxxxxxxxxx> wrote:
>
> On Sat, Mar 5, 2022 at 12:12 AM Song Liu <song@xxxxxxxxxx> wrote:
> >
> > On Fri, Mar 4, 2022 at 9:30 AM Benjamin Tissoires
> > <benjamin.tissoires@xxxxxxxxxx> wrote:
> > >
> > > LIRC_MODE2 does not really need net_admin capability, but only sys_admin.
> > >
> > > Extract a new helper for it, it will be also used for the HID bpf
> > > implementation.
> > >
> > > Cc: Sean Young <sean@xxxxxxxx>
> > > Signed-off-by: Benjamin Tissoires <benjamin.tissoires@xxxxxxxxxx>
> > >
> > > ---
> > >
> > > new in v2
> > > ---
> > > kernel/bpf/syscall.c | 14 +++++++++++++-
> > > 1 file changed, 13 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/kernel/bpf/syscall.c b/kernel/bpf/syscall.c
> > > index db402ebc5570..cc570891322b 100644
> > > --- a/kernel/bpf/syscall.c
> > > +++ b/kernel/bpf/syscall.c
> > > @@ -2165,7 +2165,6 @@ static bool is_net_admin_prog_type(enum bpf_prog_type prog_type)
> > > case BPF_PROG_TYPE_LWT_SEG6LOCAL:
> > > case BPF_PROG_TYPE_SK_SKB:
> > > case BPF_PROG_TYPE_SK_MSG:
> > > - case BPF_PROG_TYPE_LIRC_MODE2:
> > > case BPF_PROG_TYPE_FLOW_DISSECTOR:
> > > case BPF_PROG_TYPE_CGROUP_DEVICE:
> > > case BPF_PROG_TYPE_CGROUP_SOCK:
> > > @@ -2202,6 +2201,17 @@ static bool is_perfmon_prog_type(enum bpf_prog_type prog_type)
> > > }
> > > }
> > >
> > > +static bool is_sys_admin_prog_type(enum bpf_prog_type prog_type)
> > > +{
> > > + switch (prog_type) {
> > > + case BPF_PROG_TYPE_LIRC_MODE2:
> > > + case BPF_PROG_TYPE_EXT: /* extends any prog */
> > > + return true;
> > > + default:
> > > + return false;
> > > + }
> > > +}
> >
> > I am not sure whether we should do this. This is a behavior change, that may
> > break some user space. Also, BPF_PROG_TYPE_EXT is checked in
> > is_perfmon_prog_type(), and this change will make that case useless.
>
> Sure, I can drop it from v3 and make this function appear for HID only.
>
> Regarding BPF_PROG_TYPE_EXT, it was already in both
> is_net_admin_prog_type() and is_perfmon_prog_type(), so I duplicated
> it here, but I agree, given that it's already in the first function
> there, CPA_SYS_ADMIN is already checked.
I think with current code, a user with CAP_BPF, CAP_NET_ADMIN, and
CAP_PERFMON (but not CAP_SYS_ADMIN) can load programs of type
BPF_PROG_TYPE_EXT. But after the patch, the same user will not be
able to do it. Did I misread it? It is not a common case though.
