On Fri, Oct 15, 2021 at 02:26:37AM +0530, Kumar Kartikeya Dwivedi wrote:
> This helper allows us to get the address of a kernel symbol from inside
> a BPF_PROG_TYPE_SYSCALL prog (used by gen_loader), so that we can
> relocate typeless ksym vars.
...
> +BPF_CALL_4(bpf_kallsyms_lookup_name, const char *, name, int, name_sz, int, flags, u64 *, res)
> +{
> + if (flags)
> + return -EINVAL;
> +
> + if (name_sz <= 1 || name[name_sz - 1])
> + return -EINVAL;
> +
> + *res = kallsyms_lookup_name(name);
> + return *res ? 0 : -ENOENT;
> +}
> +
> +const struct bpf_func_proto bpf_kallsyms_lookup_name_proto = {
> + .func = bpf_kallsyms_lookup_name,
> + .gpl_only = true,
When libbpf is processing of typeless ksyms it parses /proc/kallsyms.
There is no gpl-ness in the action of reading this file.
Hence above should be '= false' for consistency between
light and regular skeleton.
But different check is necessary.
This helper is available to syscall_bpf prog type (which is CAP_BPF)
and shouldn't be used to bypass kptr checks.
So bpf_kallsyms_lookup_name() should probably have:
if (!bpf_dump_raw_ok(current_cred())))
return -EPERM;
The rest of the patches look great.
Thank you for working on this!
