Re: [PATCH hid v12 03/15] HID: initial BPF implementation

[Alexei Starovoitov <alexei.starovoitov@gmail.com>]

On Wed, Nov 23, 2022 at 6:53 AM Benjamin Tissoires
<benjamin.tissoires@redhat.com> wrote:
>
> Hi Jon,
>
> On Wed, Nov 23, 2022 at 2:25 PM Jon Hunter <jonathanh@nvidia.com> wrote:
> >
> >
> > On 03/11/2022 15:57, Benjamin Tissoires wrote:
> > > Declare an entry point that can use fmod_ret BPF programs, and
> > > also an API to access and change the incoming data.
> > >
> > > A simpler implementation would consist in just calling
> > > hid_bpf_device_event() for any incoming event and let users deal
> > > with the fact that they will be called for any event of any device.
> > >
> > > The goal of HID-BPF is to partially replace drivers, so this situation
> > > can be problematic because we might have programs which will step on
> > > each other toes.
> > >
> > > For that, we add a new API hid_bpf_attach_prog() that can be called
> > > from a syscall and we manually deal with a jump table in hid-bpf.
> > >
> > > Whenever we add a program to the jump table (in other words, when we
> > > attach a program to a HID device), we keep the number of time we added
> > > this program in the jump table so we can release it whenever there are
> > > no other users.
> > >
> > > HID devices have an RCU protected list of available programs in the
> > > jump table, and those programs are called one after the other thanks
> > > to bpf_tail_call().
> > >
> > > To achieve the detection of users losing their fds on the programs we
> > > attached, we add 2 tracing facilities on bpf_prog_release() (for when
> > > a fd is closed) and bpf_free_inode() (for when a pinned program gets
> > > unpinned).
> > >
> > > Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
> > > Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
> >
> > ...
> >
> > > +static int __init hid_bpf_init(void)
> > > +{
> > > +     int err;
> > > +
> > > +     /* Note: if we exit with an error any time here, we would entirely break HID, which
> > > +      * is probably not something we want. So we log an error and return success.
> > > +      *
> > > +      * This is not a big deal: the syscall allowing to attach a BPF program to a HID device
> > > +      * will not be available, so nobody will be able to use the functionality.
> > > +      */
> > > +
> > > +     err = register_btf_kfunc_id_set(BPF_PROG_TYPE_TRACING, &hid_bpf_kfunc_set);
> > > +     if (err) {
> > > +             pr_warn("error while setting HID BPF tracing kfuncs: %d", err);
> > > +             return 0;
> > > +     }
> > > +
> > > +     err = hid_bpf_preload_skel();
> > > +     if (err) {
> > > +             pr_warn("error while preloading HID BPF dispatcher: %d", err);
> > > +             return 0;
> > > +     }
> > > +
> > > +     /* register syscalls after we are sure we can load our preloaded bpf program */
> > > +     err = register_btf_kfunc_id_set(BPF_PROG_TYPE_SYSCALL, &hid_bpf_syscall_kfunc_set);
> > > +     if (err) {
> > > +             pr_warn("error while setting HID BPF syscall kfuncs: %d", err);
> > > +             return 0;
> > > +     }
> > > +
> > > +     return 0;
> > > +}
> >
> >
> > We have a kernel test that checks for new warning and error messages on
> > boot and with this change I am now seeing the following error message on
> > our Tegra platforms ...
> >
> >   WARNING KERN hid_bpf: error while preloading HID BPF dispatcher: -13
> >
> > I have a quick look at the code, but I can't say I am familiar with
> > this. So I wanted to ask if a way to fix this or avoid this? I see the
> > code returns 0, so one option would be to make this an informational or
> > debug print.
>
> I am not in favor of debug in that case, because I suspect it'll hide
> too much when getting a bug report. Informational could do, yes.
>
> However, before that, I'd like to dig a little bit more on why it is
> failing. I thought arm64 now has support of tracing bpf programs, so I
> would not expect this to fail.

Unfortunately the patches to add support for such tracing bpf progs got stuck.
Florent/Mark can probably share the latest status.