Re: [PATCH] selftests: seccomp: fix compile error seccomp_bpf

[Kees Cook]

On Fri, Jan 5, 2018 at 6:29 PM, Naresh Kamboju
<naresh.kamboju@xxxxxxxxxx> wrote:
> On 5 January 2018 at 22:01, Anders Roxell <anders.roxell@xxxxxxxxxx> wrote:
>> aarch64-linux-gnu-gcc -Wl,-no-as-needed -Wall
>>     -lpthread seccomp_bpf.c -o seccomp_bpf
>> seccomp_bpf.c: In function 'tracer_ptrace':
>> seccomp_bpf.c:1720:12: error: '__NR_open' undeclared
>>     (first use in this function)
>>   if (nr == __NR_open)
>>             ^~~~~~~~~
>> seccomp_bpf.c:1720:12: note: each undeclared identifier is reported
>>     only once for each function it appears in
>> In file included from seccomp_bpf.c:48:0:
>> seccomp_bpf.c: In function 'TRACE_syscall_ptrace_syscall_dropped':
>> seccomp_bpf.c:1795:39: error: '__NR_open' undeclared
>>     (first use in this function)
>>   EXPECT_SYSCALL_RETURN(EPERM, syscall(__NR_open));
>>                                        ^
>> open(2) is a legacy syscall, replaced with openat(2) since 2.6.16.
>> Thus new architectures in the kernel, such as arm64, don't implement
>> these legacy syscalls.
>>
>> Signed-off-by: Anders Roxell <anders.roxell@xxxxxxxxxx>
>
> Thanks for the patch Anders.
> Tested-by: Naresh Kamboju <naresh.kamboju@xxxxxxxxxx>

Did something change recently? This has built fine on arm64 for a
while -- at least since commit 256d0afb11d6 ("selftests/seccomp: build
and pass on arm64").

-Kees

>
>> ---
>>  tools/testing/selftests/seccomp/seccomp_bpf.c | 4 ++--
>>  1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/tools/testing/selftests/seccomp/seccomp_bpf.c b/tools/testing/selftests/seccomp/seccomp_bpf.c
>> index 24dbf634e2dd..0b457e8e0f0c 100644
>> --- a/tools/testing/selftests/seccomp/seccomp_bpf.c
>> +++ b/tools/testing/selftests/seccomp/seccomp_bpf.c
>> @@ -1717,7 +1717,7 @@ void tracer_ptrace(struct __test_metadata *_metadata, pid_t tracee,
>>
>>         if (nr == __NR_getpid)
>>                 change_syscall(_metadata, tracee, __NR_getppid);
>> -       if (nr == __NR_open)
>> +       if (nr == __NR_openat)
>>                 change_syscall(_metadata, tracee, -1);
>>  }
>>
>> @@ -1792,7 +1792,7 @@ TEST_F(TRACE_syscall, ptrace_syscall_dropped)
>>                                            true);
>>
>>         /* Tracer should skip the open syscall, resulting in EPERM. */
>> -       EXPECT_SYSCALL_RETURN(EPERM, syscall(__NR_open));
>> +       EXPECT_SYSCALL_RETURN(EPERM, syscall(__NR_openat));
>>  }
>>
>>  TEST_F(TRACE_syscall, syscall_allowed)
>> --
>> 2.11.0
>>



-- 
Kees Cook
Pixel Security