Re: [PATCH] selftests/bpf: Move test_dev_cgroup to prog_tests

[Muhammad Usama Anjum]

Thank you for review. I'll fix.

On 2/16/24 10:25 PM, Daniel Borkmann wrote:
> Hi Muhammad,
> 
> Small nit, pls use $subj: [PATCH bpf-next]
Sure, I'll update.

> 
> On 2/15/24 1:01 PM, Muhammad Usama Anjum wrote:
>> Move test_dev_cgroup to prog_tests to be able to run it with test_progs.
>> Replace dev_cgroup.bpf.o with skel header file, dev_cgroup.skel.h and
>> load program from it accourdingly.
>>
>>    ./test_progs -t test_dev_cgroup
>>    mknod: /tmp/test_dev_cgroup_null: Operation not permitted
>>    64+0 records in
>>    64+0 records out
>>    32768 bytes (33 kB, 32 KiB) copied, 0.000856684 s, 38.2 MB/s
>>    dd: failed to open '/dev/full': Operation not permitted
>>    dd: failed to open '/dev/random': Operation not permitted
>>    #365     test_dev_cgroup:OK
>>    Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED
>>
>> Signed-off-by: Muhammad Usama Anjum <usama.anjum@xxxxxxxxxxxxx>
> 
> BPF CI currently fails this with :
> 
> https://github.com/kernel-patches/bpf/actions/runs/7924507406/job/21636353124
> 
>   [...]
>   All error logs:
>   test_test_dev_cgroup:PASS:skel_open_and_load 0 nsec
>   test_test_dev_cgroup:PASS:cgroup_setup_and_join 0 nsec
>   test_test_dev_cgroup:PASS:bpf_attach 0 nsec
>   test_test_dev_cgroup:PASS:bpf_query 0 nsec
>   test_test_dev_cgroup:PASS:rm 0 nsec
>   test_test_dev_cgroup:PASS:mknod 0 nsec
>   test_test_dev_cgroup:PASS:rm 0 nsec
>   test_test_dev_cgroup:PASS:rm 0 nsec
>   test_test_dev_cgroup:FAIL:mknod unexpected mknod on _zero
>   test_test_dev_cgroup:PASS:rm 0 nsec
>   test_test_dev_cgroup:PASS:dd 0 nsec
>   test_test_dev_cgroup:PASS:dd 0 nsec
>   test_test_dev_cgroup:PASS:dd 0 nsec
>   (cgroup_helpers.c:353: errno: Device or resource busy) umount cgroup2
>   #366     test_dev_cgroup:FAIL
>   Summary: 517/3837 PASSED, 53 SKIPPED, 1 FAILED
> 
> You can also use vmtest.sh tool to check locally :
> 
>   # ./vmtest.sh -- ./test_progs -t test_dev_cgroup
It is passing on my side on next-20240207 and next-20240218. I'll test
again by running all the tests with test_progs. Maybe something else is
making it fail.

> 
>> diff --git a/tools/testing/selftests/bpf/prog_tests/test_dev_cgroup.c
>> b/tools/testing/selftests/bpf/prog_tests/test_dev_cgroup.c
>> new file mode 100644
>> index 0000000000000..ee37ce52dec9f
>> --- /dev/null
>> +++ b/tools/testing/selftests/bpf/prog_tests/test_dev_cgroup.c
>> @@ -0,0 +1,67 @@
>> +// SPDX-License-Identifier: GPL-2.0-only
>> +/* Copyright (c) 2017 Facebook
>> + */
>> +
>> +#include <test_progs.h>
>> +#include <time.h>
>> +#include "cgroup_helpers.h"
>> +#include "dev_cgroup.skel.h"
>> +
>> +#define TEST_CGROUP "/test-bpf-based-device-cgroup/"
>> +
>> +void test_test_dev_cgroup(void)
> 
> nit: test_dev_cgroup ?
I was trying not to rename the file. I'll rename the file.

> 
>> +{
>> +    int cgroup_fd, err, duration = 0;
>> +    struct dev_cgroup *skel;
>> +    __u32 prog_cnt;
>> +
>> +    skel = dev_cgroup__open_and_load();
>> +    if (CHECK(!skel, "skel_open_and_load", "failed\n"))
>> +        goto cleanup;

> 
> Nit: please use ASSERT_* macros everywhere, the CHECK() is deprecated.
I'd not any deprecated notice/comment. I'll use ASSERT_* from now on.

> 
>> +    cgroup_fd = cgroup_setup_and_join(TEST_CGROUP);
>> +    if (CHECK(cgroup_fd < 0, "cgroup_setup_and_join", "failed: %d\n",
>> cgroup_fd))
>> +        goto cleanup;
>> +
>> +    err = bpf_prog_attach(bpf_program__fd(skel->progs.bpf_prog1),
>> cgroup_fd,
>> +                  BPF_CGROUP_DEVICE, 0);
>> +    if (CHECK(err, "bpf_attach", "failed: %d\n", err))
>> +        goto cleanup;
>> +
>> +    err = bpf_prog_query(cgroup_fd, BPF_CGROUP_DEVICE, 0, NULL, NULL,
>> &prog_cnt);
>> +    if (CHECK(err || prog_cnt != 1, "bpf_query", "failed: %d %d\n", err,
>> prog_cnt))
>> +        goto cleanup;
>> +
>> +    /* All operations with /dev/zero and /dev/urandom are allowed,
>> +     * everything else is forbidden.
>> +     */
>> +    CHECK(system("rm -f /tmp/test_dev_cgroup_null"), "rm",
>> +          "unexpected rm on _null\n");
>> +    CHECK(!system("mknod /tmp/test_dev_cgroup_null c 1 3"),
>> +          "mknod", "unexpected mknod on _null\n");
>> +    CHECK(system("rm -f /tmp/test_dev_cgroup_null"), "rm",
>> +          "unexpected rm on _null\n");
>> +
>> +    /* /dev/zero is whitelisted */
>> +    CHECK(system("rm -f /tmp/test_dev_cgroup_zero"), "rm",
>> +          "unexpected rm on _zero\n");
>> +    CHECK(system("mknod /tmp/test_dev_cgroup_zero c 1 5"),
>> +          "mknod", "unexpected mknod on _zero\n");
>> +    CHECK(system("rm -f /tmp/test_dev_cgroup_zero"), "rm",
>> +          "unexpected rm on _zero\n");
>> +
>> +    CHECK(system("dd if=/dev/urandom of=/dev/zero count=64"), "dd",
>> +          "unexpected dd on /dev/zero\n");
>> +
>> +    /* src is allowed, target is forbidden */
>> +    CHECK(!system("dd if=/dev/urandom of=/dev/full count=64"), "dd",
>> +          "unexpected dd on /dev/full\n");
>> +
>> +    /* src is forbidden, target is allowed */
>> +    CHECK(!system("dd if=/dev/random of=/dev/zero count=64"), "dd",
>> +          "unexpected dd on /dev/zero\n");
>> +
>> +cleanup:
>> +    cleanup_cgroup_environment();
>> +    dev_cgroup__destroy(skel);
>> +}
> 
> Thanks,
> Daniel
> 

-- 
BR,
Muhammad Usama Anjum