On 6/22/20 11:45 PM, Andrii Nakryiko wrote:
On Mon, Jun 22, 2020 at 5:38 PM Yonghong Song <yhs@xxxxxx> wrote:
Refactor bpf_iter_ipv6_route.c and bpf_iter_netlink.c
so net macros, originally from various include/linux header
files, are moved to a new libbpf installable header file
bpf_tracing_net.h. The goal is to improve reuse so
networking tracing programs do not need to
copy these macros every time they use them.
Signed-off-by: Yonghong Song <yhs@xxxxxx>
---
tools/lib/bpf/Makefile | 1 +
tools/lib/bpf/bpf_tracing_net.h | 16 ++++++++++++++++
.../selftests/bpf/progs/bpf_iter_ipv6_route.c | 7 +------
.../selftests/bpf/progs/bpf_iter_netlink.c | 4 +---
4 files changed, 19 insertions(+), 9 deletions(-)
create mode 100644 tools/lib/bpf/bpf_tracing_net.h
diff --git a/tools/lib/bpf/Makefile b/tools/lib/bpf/Makefile
index bf8ed134cb8a..3d766c80eb78 100644
--- a/tools/lib/bpf/Makefile
+++ b/tools/lib/bpf/Makefile
@@ -257,6 +257,7 @@ install_headers: $(BPF_HELPER_DEFS)
$(call do_install,bpf_helpers.h,$(prefix)/include/bpf,644); \
$(call do_install,$(BPF_HELPER_DEFS),$(prefix)/include/bpf,644); \
$(call do_install,bpf_tracing.h,$(prefix)/include/bpf,644); \
+ $(call do_install,bpf_tracing_net.h,$(prefix)/include/bpf,644); \
$(call do_install,bpf_endian.h,$(prefix)/include/bpf,644); \
$(call do_install,bpf_core_read.h,$(prefix)/include/bpf,644);
diff --git a/tools/lib/bpf/bpf_tracing_net.h b/tools/lib/bpf/bpf_tracing_net.h
new file mode 100644
index 000000000000..1f38a1098727
--- /dev/null
+++ b/tools/lib/bpf/bpf_tracing_net.h
@@ -0,0 +1,16 @@
+/* SPDX-License-Identifier: (LGPL-2.1 OR BSD-2-Clause) */
+#ifndef __BPF_TRACING_NET_H__
+#define __BPF_TRACING_NET_H__
+
+#define IFNAMSIZ 16
+
+#define RTF_GATEWAY 0x0002
+
+#define fib_nh_dev nh_common.nhc_dev
+#define fib_nh_gw_family nh_common.nhc_gw_family
+#define fib_nh_gw6 nh_common.nhc_gw.ipv6
+
+#define sk_rmem_alloc sk_backlog.rmem_alloc
+#define sk_refcnt __sk_common.skc_refcnt
Question to networking guys. How probable it is for these and similar
definitions to ever be changed?
I'm a bit hesitant to make any stability guarantees (which is implied
by libbpf-provided headers). I don't want us to get into the game of
trying to maintain this across multiple kernel versions, if they are
going to be changed.
Let's for now keep bpf_tracing_net.h under selftests/bpf? It's still
good to have these definitions, because we can point people to it.
I am using bpf_tracing_net.h name to somehow signal that it may
*potentially* change. But it does have long-time stability issue.
Will move to selftests/bpf then.
+
+#endif
[...]
