On 12/04/2023 11:45, Florian Westphal wrote:
> Quentin Deslandes <qde@xxxxxxxx> wrote:
>> On 05/04/2023 18:11, Florian Westphal wrote:
>>> Add minimal support to hook bpf programs to netfilter hooks, e.g.
>>> PREROUTING or FORWARD.
>>>
>>> For this the most relevant parts for registering a netfilter
>>> hook via the in-kernel api are exposed to userspace via bpf_link.
>>>
>>> The new program type is 'tracing style', i.e. there is no context
>>> access rewrite done by verifier, the function argument (struct bpf_nf_ctx)
>>> isn't stable.
>>> There is no support for direct packet access, dynptr api should be used
>>> instead.
>>
>> Does this mean the verifier will reject any program accessing ctx->skb
>> (e.g. ctx->skb + X)?
>
> Do you mean access to ctx->skb->data + X? If so, yes, that won't work.
>
> Otherwise, then no, it just means that programs might have to be recompiled
> if they lack needed relocation information, but only if bpf_nf_ctx structure is
> changed.
>
> Initial version used "__sk_buff *skb", like e.g. clsact. I was told
> to not do that and expose the real kernel-side structure instead and to
> not bother with direct packet access (skb->data access) support.
>
That's exactly what I meant, thanks.
>>> #include "vmlinux.h"
>>> extern int bpf_dynptr_from_skb(struct __sk_buff *skb, __u64 flags,
>>> struct bpf_dynptr *ptr__uninit) __ksym;
>>> extern void *bpf_dynptr_slice(const struct bpf_dynptr *ptr, uint32_t offset,
>>> void *buffer, uint32_t buffer__sz) __ksym;
>>> SEC("netfilter")
>>> int nf_test(struct bpf_nf_ctx *ctx)
>>> {
>>> struct nf_hook_state *state = ctx->state;
>>> struct sk_buff *skb = ctx->skb;
>
> ctx->skb is dereferenced...
>
>>> if (bpf_dynptr_from_skb(skb, 0, &ptr))
>>> return NF_DROP;
>
> ... dynptr is created ...
>
>>> iph = bpf_dynptr_slice(&ptr, 0, &_iph, sizeof(_iph));
>>> if (!iph)
>>> return NF_DROP;
>>> th = bpf_dynptr_slice(&ptr, iph->ihl << 2, &_th, sizeof(_th));
>
> ip header access.
