Re: [PATCHv17 bpf-next 5/6] selftests/bpf: Add verifier tests for bpf arg ARG_CONST_MAP_PTR_OR_NULL

[Hangbin Liu]

On Wed, Jan 27, 2021 at 02:24:47PM -0800, John Fastabend wrote:
> [...]
> 
> > +{
> > +   "ARG_CONST_MAP_PTR_OR_NULL: null pointer for ex_map",
> > +   .insns = {
> > +           BPF_MOV64_IMM(BPF_REG_1, 0),
> > +           /* bpf_redirect_map_multi arg1 (in_map) */
> > +           BPF_LD_MAP_FD(BPF_REG_1, 0),
> > +           /* bpf_redirect_map_multi arg2 (ex_map) */
> > +           BPF_MOV64_IMM(BPF_REG_2, 0),
> > +           /* bpf_redirect_map_multi arg3 (flags) */
> > +           BPF_MOV64_IMM(BPF_REG_3, 0),
> > +           BPF_EMIT_CALL(BPF_FUNC_redirect_map_multi),
> > +           BPF_EXIT_INSN(),
> > +   },
> > +   .fixup_map_devmap = { 1 },
> > +   .result = ACCEPT,
> > +   .prog_type = BPF_PROG_TYPE_XDP,
> > +   .retval = 4,
> 
> Do we need one more case where this is map_or_null? In above
> ex_map will be scalar tnum_const=0 and be exactly a null. This
> will push verifier here,
> 
>   meta->map_ptr = register_is_null(reg) ? NULL : reg->map_ptr;
> 
> In the below case it is known to be not null.
> 
> Is it also interesting to have a case where register_is_null(reg)
> check fails and reg->map_ptr is set, but may be null.

Hi John,

I'm not familiar with the test_verifier syntax. Doesn't
BPF_LD_MAP_FD(BPF_REG_1, 0) just assign the register with map NULL?

Thanks
hangbin