Re: [PATCH bpf-next v3 2/3] libbpf: add low level TC-BPF API

On 4/22/21 1:30 AM, Kumar Kartikeya Dwivedi wrote:

On Thu, Apr 22, 2021 at 04:51:55AM IST, Daniel Borkmann wrote:

quoted

On 4/22/21 1:08 AM, Kumar Kartikeya Dwivedi wrote:

quoted

On Thu, Apr 22, 2021 at 04:29:28AM IST, Daniel Borkmann wrote:

quoted

On 4/20/21 9:37 PM, Kumar Kartikeya Dwivedi wrote:
[...]

quoted

diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
index bec4e6a6e31d..b4ed6a41ea70 100644
--- a/tools/lib/bpf/libbpf.h
+++ b/tools/lib/bpf/libbpf.h

@@ -16,6 +16,8 @@
    #include <stdbool.h>
    #include <sys/types.h>  // for size_t
    #include <linux/bpf.h>
+#include <linux/pkt_sched.h>
+#include <linux/tc_act/tc_bpf.h>
    #include "libbpf_common.h"

@@ -775,6 +777,48 @@ LIBBPF_API int bpf_linker__add_file(struct bpf_linker *linker, const char *filen
    LIBBPF_API int bpf_linker__finalize(struct bpf_linker *linker);
    LIBBPF_API void bpf_linker__free(struct bpf_linker *linker);
+/* Convenience macros for the clsact attach hooks */
+#define BPF_TC_CLSACT_INGRESS TC_H_MAKE(TC_H_CLSACT, TC_H_MIN_INGRESS)
+#define BPF_TC_CLSACT_EGRESS TC_H_MAKE(TC_H_CLSACT, TC_H_MIN_EGRESS)

I would abstract those away into an enum, plus avoid having to pull in
linux/pkt_sched.h and linux/tc_act/tc_bpf.h from main libbpf.h header.

Just add a enum { BPF_TC_DIR_INGRESS, BPF_TC_DIR_EGRESS, } and then the
concrete tc bits (TC_H_MAKE()) can be translated internally.

Ok, will do.

quoted

quoted

+struct bpf_tc_opts {
+	size_t sz;

Is this set anywhere?

This is needed by the OPTS_* infrastructure.

quoted

quoted

+	__u32 handle;
+	__u32 class_id;

I'd remove class_id from here as well given in direct-action a BPF prog can
set it if needed.

Ok, makes sense.

quoted

quoted

+	__u16 priority;
+	bool replace;
+	size_t :0;

What's the rationale for this padding?

dde7b3f5f2f4 ("libbpf: Add explicit padding to bpf_xdp_set_link_opts")

Hm, fair enough.

quoted

quoted

quoted

+};
+
+#define bpf_tc_opts__last_field replace
+
+/* Acts as a handle for an attached filter */
+struct bpf_tc_attach_id {

nit: maybe bpf_tc_ctx

Noted.

quoted

quoted

+	__u32 handle;
+	__u16 priority;
+};
+
+struct bpf_tc_info {
+	struct bpf_tc_attach_id id;
+	__u16 protocol;
+	__u32 chain_index;
+	__u32 prog_id;
+	__u8 tag[BPF_TAG_SIZE];
+	__u32 class_id;
+	__u32 bpf_flags;
+	__u32 bpf_flags_gen;

Given we do not yet have any setters e.g. for offload, etc, the one thing
I'd see useful and crucial initially is prog_id.

The protocol, chain_index, and I would also include tag should be dropped.

A future user of this API needs to know the tag, so I would like to keep that.
The rest we can drop, and probably document the default values explicitly.

Couldn't this be added along with the future patch for the [future] user?

True.

quoted

The tag should be the tag of the prog itself, so if you have prog_id, you
could also retrieve the same tag from the prog. The tag was basically from
the early days where we didn't have bpf_prog_get_info_by_fd().

What does that future user need to do different here?

 From Shaun Crampton:
"My particular use case is to load a program, link it with its maps and then
check if its tag matches the existing program on the interface (and if so, abort
the update)"

Also CC'd, they would be able to elaborate better, and whether or not dropping
it is ok.

Nope, just get it from the prog itself.