Re: [PATCH bpf-next v3 02/17] bpf: introduce hid program type

[Song Liu <song@kernel.org>]

On Mon, Mar 21, 2022 at 9:07 AM Benjamin Tissoires
<benjamin.tissoires@redhat.com> wrote:
>
> Hi Song,
>
> many thanks for the quick response.
>
> On Fri, Mar 18, 2022 at 9:48 PM Song Liu <song@kernel.org> wrote:
[...]
> >
> > > diff --git a/include/uapi/linux/bpf.h b/include/uapi/linux/bpf.h
> >
> > We need to mirror these changes to tools/include/uapi/linux/bpf.h.
>
> OK. I did that in patch 4/17 but I can bring in the changes there too.

Let's keep changes to the two files in the same patch. This will make
sure they are back ported together.

[...]
> > > +enum hid_bpf_event {
> > > +       HID_BPF_UNDEF = 0,
> > > +       HID_BPF_DEVICE_EVENT,           /* when attach type is BPF_HID_DEVICE_EVENT */
> > > +       HID_BPF_RDESC_FIXUP,            /* ................... BPF_HID_RDESC_FIXUP */
> > > +       HID_BPF_USER_EVENT,             /* ................... BPF_HID_USER_EVENT */
> >
> > Why don't we have a DRIVER_EVENT type here?
>
> For driver event, I want to have a little bit more of information
> which tells which event we have:
> - HID_BPF_DRIVER_PROBE
> - HID_BPF_DRIVER_SUSPEND
> - HID_BPF_DRIVER_RAW_REQUEST
> - HID_BPF_DRIVER_RAW_REQUEST_ANSWER
> - etc...
>
> However, I am not entirely sure on the implementation of all of those,
> so I left them aside for now.
>
> I'll work on that for v4.

This set is already pretty big. I guess we can add them in a follow-up set.

>
> >
> > >
> > [...]
> > > +
> > > +BPF_CALL_3(bpf_hid_get_data, struct hid_bpf_ctx_kern*, ctx, u64, offset, u64, size)
> > > +{
> > > +       if (!size)
> > > +               return 0UL;
> > > +
> > > +       if (offset + size > ctx->allocated_size)
> > > +               return 0UL;
> > > +
> > > +       return (unsigned long)(ctx->data + offset);
> > > +}
> > > +
> > > +static const struct bpf_func_proto bpf_hid_get_data_proto = {
> > > +       .func      = bpf_hid_get_data,
> > > +       .gpl_only  = true,
> > > +       .ret_type  = RET_PTR_TO_ALLOC_MEM_OR_NULL,
> > > +       .arg1_type = ARG_PTR_TO_CTX,
> > > +       .arg2_type = ARG_ANYTHING,
> > > +       .arg3_type = ARG_CONST_ALLOC_SIZE_OR_ZERO,
> >
> > I think we should use ARG_CONST_SIZE or ARG_CONST_SIZE_OR_ZERO?
>
> I initially tried this with ARG_CONST_SIZE_OR_ZERO but it doesn't work
> for 2 reasons:
> - we need to pair the argument ARG_CONST_SIZE_* with a pointer to a
> memory just before, which doesn't really make sense here
> - ARG_CONST_SIZE_* isn't handled in the same way
> ARG_CONST_ALLOC_SIZE_OR_ZERO is. The latter tells the verifier that
> the given size is the available size of the returned
> PTR_TO_ALLOC_MEM_OR_NULL, which is exactly what we want.

I misread the logic initially. It makes sense now.

>
> >
> > > +};
> > > +
> > > +static const struct bpf_func_proto *
> > > +hid_func_proto(enum bpf_func_id func_id, const struct bpf_prog *prog)
> > > +{
> > > +       switch (func_id) {
> > > +       case BPF_FUNC_hid_get_data:
> > > +               return &bpf_hid_get_data_proto;
> > > +       default:
> > > +               return bpf_base_func_proto(func_id);
> > > +       }
> > > +}
> > [...]
> > > +
> > > +static int hid_bpf_prog_test_run(struct bpf_prog *prog,
> > > +                                const union bpf_attr *attr,
> > > +                                union bpf_attr __user *uattr)
> > > +{
> > > +       struct hid_device *hdev = NULL;
> > > +       struct bpf_prog_array *progs;
> > > +       bool valid_prog = false;
> > > +       int i;
> > > +       int target_fd, ret;
> > > +       void __user *data_out = u64_to_user_ptr(attr->test.data_out);
> > > +       void __user *data_in = u64_to_user_ptr(attr->test.data_in);
> > > +       u32 user_size_in = attr->test.data_size_in;
> > > +       u32 user_size_out = attr->test.data_size_out;
> > > +       u32 allocated_size = max(user_size_in, user_size_out);
> > > +       struct hid_bpf_ctx_kern ctx = {
> > > +               .type = HID_BPF_USER_EVENT,
> > > +               .allocated_size = allocated_size,
> > > +       };
> > > +
> > > +       if (!hid_hooks.hdev_from_fd)
> > > +               return -EOPNOTSUPP;
> > > +
> > > +       if (attr->test.ctx_size_in != sizeof(int))
> > > +               return -EINVAL;
> >
> > ctx_size_in is always 4 bytes?
>
> Yes. Basically what I had in mind is that the "ctx" for
> user_prog_test_run is the file descriptor to the sysfs that represent
> the HID device.
> This seemed to me to be the easiest to handle for users.
>
> I'm open to suggestions though.

How about we use data_in? ctx for test_run usually means the program ctx,
which is struct hid_bpf_ctx here.

Thanks,
Song