selftests/bpf: Test BPF_F_REPLACE in cgroup_attach_multi
authorAndrey Ignatov <rdna@fb.com>
Thu, 19 Dec 2019 07:44:38 +0000 (23:44 -0800)
committerAlexei Starovoitov <ast@kernel.org>
Fri, 20 Dec 2019 05:22:26 +0000 (21:22 -0800)
Test replacing a cgroup-bpf program attached with BPF_F_ALLOW_MULTI and
possible failure modes: invalid combination of flags, invalid
replace_bpf_fd, replacing a non-attachd to specified cgroup program.

Example of program replacing:

  # gdb -q --args ./test_progs --name=cgroup_attach_multi
  ...
  Breakpoint 1, test_cgroup_attach_multi () at cgroup_attach_multi.c:227
  (gdb)
  [1]+  Stopped                 gdb -q --args ./test_progs --name=cgroup_attach_multi
  # bpftool c s /mnt/cgroup2/cgroup-test-work-dir/cg1
  ID       AttachType      AttachFlags     Name
  2133     egress          multi
  2134     egress          multi
  # fg
  gdb -q --args ./test_progs --name=cgroup_attach_multi
  (gdb) c
  Continuing.

  Breakpoint 2, test_cgroup_attach_multi () at cgroup_attach_multi.c:233
  (gdb)
  [1]+  Stopped                 gdb -q --args ./test_progs --name=cgroup_attach_multi
  # bpftool c s /mnt/cgroup2/cgroup-test-work-dir/cg1
  ID       AttachType      AttachFlags     Name
  2139     egress          multi
  2134     egress          multi

Signed-off-by: Andrey Ignatov <rdna@fb.com>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Link: https://lore.kernel.org/bpf/7b9b83e8d5fb82e15b034341bd40b6fb2431eeba.1576741281.git.rdna@fb.com
tools/testing/selftests/bpf/prog_tests/cgroup_attach_multi.c

index 4eaab743504406e0a8980c4f12129176f711c94d..2ff21dbce179c0b0468d55677522551364016726 100644 (file)
@@ -78,7 +78,8 @@ void test_cgroup_attach_multi(void)
 {
        __u32 prog_ids[4], prog_cnt = 0, attach_flags, saved_prog_id;
        int cg1 = 0, cg2 = 0, cg3 = 0, cg4 = 0, cg5 = 0, key = 0;
-       int allow_prog[6] = {-1};
+       DECLARE_LIBBPF_OPTS(bpf_prog_attach_opts, attach_opts);
+       int allow_prog[7] = {-1};
        unsigned long long value;
        __u32 duration = 0;
        int i = 0;
@@ -189,6 +190,52 @@ void test_cgroup_attach_multi(void)
        CHECK_FAIL(bpf_map_lookup_elem(map_fd, &key, &value));
        CHECK_FAIL(value != 1 + 2 + 8 + 16);
 
+       /* test replace */
+
+       attach_opts.flags = BPF_F_ALLOW_OVERRIDE | BPF_F_REPLACE;
+       attach_opts.replace_prog_fd = allow_prog[0];
+       if (CHECK(!bpf_prog_attach_xattr(allow_prog[6], cg1,
+                                        BPF_CGROUP_INET_EGRESS, &attach_opts),
+                 "fail_prog_replace_override", "unexpected success\n"))
+               goto err;
+       CHECK_FAIL(errno != EINVAL);
+
+       attach_opts.flags = BPF_F_REPLACE;
+       if (CHECK(!bpf_prog_attach_xattr(allow_prog[6], cg1,
+                                        BPF_CGROUP_INET_EGRESS, &attach_opts),
+                 "fail_prog_replace_no_multi", "unexpected success\n"))
+               goto err;
+       CHECK_FAIL(errno != EINVAL);
+
+       attach_opts.flags = BPF_F_ALLOW_MULTI | BPF_F_REPLACE;
+       attach_opts.replace_prog_fd = -1;
+       if (CHECK(!bpf_prog_attach_xattr(allow_prog[6], cg1,
+                                        BPF_CGROUP_INET_EGRESS, &attach_opts),
+                 "fail_prog_replace_bad_fd", "unexpected success\n"))
+               goto err;
+       CHECK_FAIL(errno != EBADF);
+
+       /* replacing a program that is not attached to cgroup should fail  */
+       attach_opts.replace_prog_fd = allow_prog[3];
+       if (CHECK(!bpf_prog_attach_xattr(allow_prog[6], cg1,
+                                        BPF_CGROUP_INET_EGRESS, &attach_opts),
+                 "fail_prog_replace_no_ent", "unexpected success\n"))
+               goto err;
+       CHECK_FAIL(errno != ENOENT);
+
+       /* replace 1st from the top program */
+       attach_opts.replace_prog_fd = allow_prog[0];
+       if (CHECK(bpf_prog_attach_xattr(allow_prog[6], cg1,
+                                       BPF_CGROUP_INET_EGRESS, &attach_opts),
+                 "prog_replace", "errno=%d\n", errno))
+               goto err;
+
+       value = 0;
+       CHECK_FAIL(bpf_map_update_elem(map_fd, &key, &value, 0));
+       CHECK_FAIL(system(PING_CMD));
+       CHECK_FAIL(bpf_map_lookup_elem(map_fd, &key, &value));
+       CHECK_FAIL(value != 64 + 2 + 8 + 16);
+
        /* detach 3rd from bottom program and ping again */
        if (CHECK(!bpf_prog_detach2(0, cg3, BPF_CGROUP_INET_EGRESS),
                  "fail_prog_detach_from_cg3", "unexpected success\n"))
@@ -202,7 +249,7 @@ void test_cgroup_attach_multi(void)
        CHECK_FAIL(bpf_map_update_elem(map_fd, &key, &value, 0));
        CHECK_FAIL(system(PING_CMD));
        CHECK_FAIL(bpf_map_lookup_elem(map_fd, &key, &value));
-       CHECK_FAIL(value != 1 + 2 + 16);
+       CHECK_FAIL(value != 64 + 2 + 16);
 
        /* detach 2nd from bottom program and ping again */
        if (CHECK(bpf_prog_detach2(-1, cg4, BPF_CGROUP_INET_EGRESS),
@@ -213,7 +260,7 @@ void test_cgroup_attach_multi(void)
        CHECK_FAIL(bpf_map_update_elem(map_fd, &key, &value, 0));
        CHECK_FAIL(system(PING_CMD));
        CHECK_FAIL(bpf_map_lookup_elem(map_fd, &key, &value));
-       CHECK_FAIL(value != 1 + 2 + 4);
+       CHECK_FAIL(value != 64 + 2 + 4);
 
        prog_cnt = 4;
        CHECK_FAIL(bpf_prog_query(cg5, BPF_CGROUP_INET_EGRESS,