On Nov 14, 2019, at 3:01 PM, Alexei Starovoitov [off-list ref] wrote:

On Thu, Nov 14, 2019 at 10:55:37PM +0000, Song Liu wrote:

quoted

quoted

On Nov 14, 2019, at 10:57 AM, Alexei Starovoitov [off-list ref] wrote:

Annotate BPF program context types with program-side type and kernel-side type.
This type information is used by the verifier. btf_get_prog_ctx_type() is
used in the later patches to verify that BTF type of ctx in BPF program matches to
kernel expected ctx type. For example, the XDP program type is:
BPF_PROG_TYPE(BPF_PROG_TYPE_XDP, xdp, struct xdp_md, struct xdp_buff)
That means that XDP program should be written as:
int xdp_prog(struct xdp_md *ctx) { ... }

Signed-off-by: Alexei Starovoitov <ast@kernel.org>

[...]

quoted

+	/* only compare that prog's ctx type name is the same as
+	 * kernel expects. No need to compare field by field.
+	 * It's ok for bpf prog to do:
+	 * struct __sk_buff {};
+	 * int socket_filter_bpf_prog(struct __sk_buff *skb)
+	 * { // no fields of skb are ever used }
+	 */
+	if (strcmp(ctx_tname, tname))
+		return NULL;

Do we need to check size of the two struct? I guess we should not 
allow something like

	struct __sk_buff {
		char data[REALLY_BIG_NUM]; 
	};
	int socket_filter_bpf_prog(struct __sk_buff *skb)
	{ /* access end of skb */ }

I don't think we should check sizes either. Same comment above applies. The
prog's __sk_buff can be different from kernel's view into __sk_buff. Either
bigger or larger doesn't matter. If it's accessed by the prog the verifier will
check that all accessed fields are correct. Extra unused fields (like char
data[REALLY_BIG_NUM];) don't affect safety.
When bpf-tracing is attaching to bpf-skb it doesn't use bpf-skb's
__sk_buff with giant fake data[BIG_NUM];. It's using kernel's __sk_buff.
That is what btf_translate_to_vmlinux() in patch 17 is doing.

I see. Thanks for the pointer. 

Song