Merge branch 'cgroup-bpf_link'

Andrii Nakryiko says:

====================
bpf_link abstraction itself was formalized in [0] with justifications for why
its semantics is a good fit for attaching BPF programs of various types. This
patch set adds bpf_link-based BPF program attachment mechanism for cgroup BPF
programs.

Cgroup BPF link is semantically compatible with current BPF_F_ALLOW_MULTI
semantics of attaching cgroup BPF programs directly. Thus cgroup bpf_link can
co-exist with legacy BPF program multi-attachment.

bpf_link is destroyed and automatically detached when the last open FD holding
the reference to bpf_link is closed. This means that by default, when the
process that created bpf_link exits, attached BPF program will be
automatically detached due to bpf_link's clean up code. Cgroup bpf_link, like
any other bpf_link, can be pinned in BPF FS and by those means survive the
exit of process that created the link. This is useful in many scenarios to
provide long-living BPF program attachments. Pinning also means that there
could be many owners of bpf_link through independent FDs.

Additionally, auto-detachmet of cgroup bpf_link is implemented. When cgroup is
dying it will automatically detach all active bpf_links. This ensures that
cgroup clean up is not delayed due to active bpf_link even despite no chance
for any BPF program to be run for a given cgroup. In that sense it's similar
to existing behavior of dropping refcnt of attached bpf_prog. But in the case
of bpf_link, bpf_link is not destroyed and is still available to user as long
as at least one active FD is still open (or if it's pinned in BPF FS).

There are two main cgroup-specific differences between bpf_link-based and
direct bpf_prog-based attachment.

First, as opposed to direct bpf_prog attachment, cgroup itself doesn't "own"
bpf_link, which makes it possible to auto-clean up attached bpf_link when user
process abruptly exits without explicitly detaching BPF program. This makes
for a safe default behavior proven in BPF tracing program types. But bpf_link
doesn't bump cgroup->bpf.refcnt as well and because of that doesn't prevent
cgroup from cleaning up its BPF state.

Second, only owners of bpf_link (those who created bpf_link in the first place
or obtained a new FD by opening bpf_link from BPF FS) can detach and/or update
it. This makes sure that no other process can accidentally remove/replace BPF
program.

This patch set also implements LINK_UPDATE sub-command, which allows to
replace bpf_link's underlying bpf_prog, similarly to BPF_F_REPLACE flag
behavior for direct bpf_prog cgroup attachment. Similarly to LINK_CREATE, it
is supposed to be generic command for different types of bpf_links.

  [0] https://lore.kernel.org/bpf/20200228223948.360936-1-andriin@fb.com/

v2->v3:
  - revert back to just MULTI mode (Alexei);
  - fix tinyconfig compilation warning (kbuild test robot);

v1->v2:
  - implement exclusive and overridable exclusive modes (Andrey Ignatov);
  - fix build for !CONFIG_CGROUP_BPF build;
  - add more selftests for non-multi mode and inter-operability;
====================

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

1 files changed, 244 insertions, 0 deletions

diff --git a/tools/testing/selftests/bpf/prog_tests/cgroup_link.c b/tools/testing/selftests/bpf/prog_tests/cgroup_link.c
new file mode 100644
index 000000000000..6e04f8d1d15b
--- /dev/null
+++ b/tools/testing/selftests/bpf/prog_tests/cgroup_link.c
@@ -0,0 +1,244 @@
+// SPDX-License-Identifier: GPL-2.0
+
+#include <test_progs.h>
+#include "cgroup_helpers.h"
+#include "test_cgroup_link.skel.h"
+
+static __u32 duration = 0;
+#define PING_CMD	"ping -q -c1 -w1 127.0.0.1 > /dev/null"
+
+static struct test_cgroup_link *skel = NULL;
+
+int ping_and_check(int exp_calls, int exp_alt_calls)
+{
+	skel->bss->calls = 0;
+	skel->bss->alt_calls = 0;
+	CHECK_FAIL(system(PING_CMD));
+	if (CHECK(skel->bss->calls != exp_calls, "call_cnt",
+		  "exp %d, got %d\n", exp_calls, skel->bss->calls))
+		return -EINVAL;
+	if (CHECK(skel->bss->alt_calls != exp_alt_calls, "alt_call_cnt",
+		  "exp %d, got %d\n", exp_alt_calls, skel->bss->alt_calls))
+		return -EINVAL;
+	return 0;
+}
+
+void test_cgroup_link(void)
+{
+	struct {
+		const char *path;
+		int fd;
+	} cgs[] = {
+		{ "/cg1" },
+		{ "/cg1/cg2" },
+		{ "/cg1/cg2/cg3" },
+		{ "/cg1/cg2/cg3/cg4" },
+	};
+	int last_cg = ARRAY_SIZE(cgs) - 1, cg_nr = ARRAY_SIZE(cgs);
+	DECLARE_LIBBPF_OPTS(bpf_link_update_opts, link_upd_opts);
+	struct bpf_link *links[ARRAY_SIZE(cgs)] = {}, *tmp_link;
+	__u32 prog_ids[ARRAY_SIZE(cgs)], prog_cnt = 0, attach_flags;
+	int i = 0, err, prog_fd;
+	bool detach_legacy = false;
+
+	skel = test_cgroup_link__open_and_load();
+	if (CHECK(!skel, "skel_open_load", "failed to open/load skeleton\n"))
+		return;
+	prog_fd = bpf_program__fd(skel->progs.egress);
+
+	err = setup_cgroup_environment();
+	if (CHECK(err, "cg_init", "failed: %d\n", err))
+		goto cleanup;
+
+	for (i = 0; i < cg_nr; i++) {
+		cgs[i].fd = create_and_get_cgroup(cgs[i].path);
+		if (CHECK(cgs[i].fd < 0, "cg_create", "fail: %d\n", cgs[i].fd))
+			goto cleanup;
+	}
+
+	err = join_cgroup(cgs[last_cg].path);
+	if (CHECK(err, "cg_join", "fail: %d\n", err))
+		goto cleanup;
+
+	for (i = 0; i < cg_nr; i++) {
+		links[i] = bpf_program__attach_cgroup(skel->progs.egress,
+						      cgs[i].fd);
+		if (CHECK(IS_ERR(links[i]), "cg_attach", "i: %d, err: %ld\n",
+				 i, PTR_ERR(links[i])))
+			goto cleanup;
+	}
+
+	ping_and_check(cg_nr, 0);
+
+	/* query the number of effective progs and attach flags in root cg */
+	err = bpf_prog_query(cgs[0].fd, BPF_CGROUP_INET_EGRESS,
+			     BPF_F_QUERY_EFFECTIVE, &attach_flags, NULL,
+			     &prog_cnt);
+	CHECK_FAIL(err);
+	CHECK_FAIL(attach_flags != BPF_F_ALLOW_MULTI);
+	if (CHECK(prog_cnt != 1, "effect_cnt", "exp %d, got %d\n", 1, prog_cnt))
+		goto cleanup;
+
+	/* query the number of effective progs in last cg */
+	err = bpf_prog_query(cgs[last_cg].fd, BPF_CGROUP_INET_EGRESS,
+			     BPF_F_QUERY_EFFECTIVE, NULL, NULL,
+			     &prog_cnt);
+	CHECK_FAIL(err);
+	CHECK_FAIL(attach_flags != BPF_F_ALLOW_MULTI);
+	if (CHECK(prog_cnt != cg_nr, "effect_cnt", "exp %d, got %d\n",
+		  cg_nr, prog_cnt))
+		goto cleanup;
+
+	/* query the effective prog IDs in last cg */
+	err = bpf_prog_query(cgs[last_cg].fd, BPF_CGROUP_INET_EGRESS,
+			     BPF_F_QUERY_EFFECTIVE, &attach_flags,
+			     prog_ids, &prog_cnt);
+	CHECK_FAIL(err);
+	CHECK_FAIL(attach_flags != BPF_F_ALLOW_MULTI);
+	if (CHECK(prog_cnt != cg_nr, "effect_cnt", "exp %d, got %d\n",
+		  cg_nr, prog_cnt))
+		goto cleanup;
+	for (i = 1; i < prog_cnt; i++) {
+		CHECK(prog_ids[i - 1] != prog_ids[i], "prog_id_check",
+		      "idx %d, prev id %d, cur id %d\n",
+		      i, prog_ids[i - 1], prog_ids[i]);
+	}
+
+	/* detach bottom program and ping again */
+	bpf_link__destroy(links[last_cg]);
+	links[last_cg] = NULL;
+
+	ping_and_check(cg_nr - 1, 0);
+
+	/* mix in with non link-based multi-attachments */
+	err = bpf_prog_attach(prog_fd, cgs[last_cg].fd,
+			      BPF_CGROUP_INET_EGRESS, BPF_F_ALLOW_MULTI);
+	if (CHECK(err, "cg_attach_legacy", "errno=%d\n", errno))
+		goto cleanup;
+	detach_legacy = true;
+
+	links[last_cg] = bpf_program__attach_cgroup(skel->progs.egress,
+						    cgs[last_cg].fd);
+	if (CHECK(IS_ERR(links[last_cg]), "cg_attach", "err: %ld\n",
+		  PTR_ERR(links[last_cg])))
+		goto cleanup;
+
+	ping_and_check(cg_nr + 1, 0);
+
+	/* detach link */
+	bpf_link__destroy(links[last_cg]);
+	links[last_cg] = NULL;
+
+	/* detach legacy */
+	err = bpf_prog_detach2(prog_fd, cgs[last_cg].fd, BPF_CGROUP_INET_EGRESS);
+	if (CHECK(err, "cg_detach_legacy", "errno=%d\n", errno))
+		goto cleanup;
+	detach_legacy = false;
+
+	/* attach legacy exclusive prog attachment */
+	err = bpf_prog_attach(prog_fd, cgs[last_cg].fd,
+			      BPF_CGROUP_INET_EGRESS, 0);
+	if (CHECK(err, "cg_attach_exclusive", "errno=%d\n", errno))
+		goto cleanup;
+	detach_legacy = true;
+
+	/* attempt to mix in with multi-attach bpf_link */
+	tmp_link = bpf_program__attach_cgroup(skel->progs.egress,
+					      cgs[last_cg].fd);
+	if (CHECK(!IS_ERR(tmp_link), "cg_attach_fail", "unexpected success!\n")) {
+		bpf_link__destroy(tmp_link);
+		goto cleanup;
+	}
+
+	ping_and_check(cg_nr, 0);
+
+	/* detach */
+	err = bpf_prog_detach2(prog_fd, cgs[last_cg].fd, BPF_CGROUP_INET_EGRESS);
+	if (CHECK(err, "cg_detach_legacy", "errno=%d\n", errno))
+		goto cleanup;
+	detach_legacy = false;
+
+	ping_and_check(cg_nr - 1, 0);
+
+	/* attach back link-based one */
+	links[last_cg] = bpf_program__attach_cgroup(skel->progs.egress,
+						    cgs[last_cg].fd);
+	if (CHECK(IS_ERR(links[last_cg]), "cg_attach", "err: %ld\n",
+		  PTR_ERR(links[last_cg])))
+		goto cleanup;
+
+	ping_and_check(cg_nr, 0);
+
+	/* check legacy exclusive prog can't be attached */
+	err = bpf_prog_attach(prog_fd, cgs[last_cg].fd,
+			      BPF_CGROUP_INET_EGRESS, 0);
+	if (CHECK(!err, "cg_attach_exclusive", "unexpected success")) {
+		bpf_prog_detach2(prog_fd, cgs[last_cg].fd, BPF_CGROUP_INET_EGRESS);
+		goto cleanup;
+	}
+
+	/* replace BPF programs inside their links for all but first link */
+	for (i = 1; i < cg_nr; i++) {
+		err = bpf_link__update_program(links[i], skel->progs.egress_alt);
+		if (CHECK(err, "prog_upd", "link #%d\n", i))
+			goto cleanup;
+	}
+
+	ping_and_check(1, cg_nr - 1);
+
+	/* Attempt program update with wrong expected BPF program */
+	link_upd_opts.old_prog_fd = bpf_program__fd(skel->progs.egress_alt);
+	link_upd_opts.flags = BPF_F_REPLACE;
+	err = bpf_link_update(bpf_link__fd(links[0]),
+			      bpf_program__fd(skel->progs.egress_alt),
+			      &link_upd_opts);
+	if (CHECK(err == 0 || errno != EPERM, "prog_cmpxchg1",
+		  "unexpectedly succeeded, err %d, errno %d\n", err, -errno))
+		goto cleanup;
+
+	/* Compare-exchange single link program from egress to egress_alt */
+	link_upd_opts.old_prog_fd = bpf_program__fd(skel->progs.egress);
+	link_upd_opts.flags = BPF_F_REPLACE;
+	err = bpf_link_update(bpf_link__fd(links[0]),
+			      bpf_program__fd(skel->progs.egress_alt),
+			      &link_upd_opts);
+	if (CHECK(err, "prog_cmpxchg2", "errno %d\n", -errno))
+		goto cleanup;
+
+	/* ping */
+	ping_and_check(0, cg_nr);
+
+	/* close cgroup FDs before detaching links */
+	for (i = 0; i < cg_nr; i++) {
+		if (cgs[i].fd > 0) {
+			close(cgs[i].fd);
+			cgs[i].fd = -1;
+		}
+	}
+
+	/* BPF programs should still get called */
+	ping_and_check(0, cg_nr);
+
+	/* leave cgroup and remove them, don't detach programs */
+	cleanup_cgroup_environment();
+
+	/* BPF programs should have been auto-detached */
+	ping_and_check(0, 0);
+
+cleanup:
+	if (detach_legacy)
+		bpf_prog_detach2(prog_fd, cgs[last_cg].fd,
+				 BPF_CGROUP_INET_EGRESS);
+
+	for (i = 0; i < cg_nr; i++) {
+		if (!IS_ERR(links[i]))
+			bpf_link__destroy(links[i]);
+	}
+	test_cgroup_link__destroy(skel);
+
+	for (i = 0; i < cg_nr; i++) {
+		if (cgs[i].fd > 0)
+			close(cgs[i].fd);
+	}
+	cleanup_cgroup_environment();
+}