Re: [PATCH bpf-next v3 13/15] tools/bpf: selftests: implement sample tcp/tcp6 bpf_iter programs

From: Yonghong Song <yhs@fb.com>
To: Andrii Nakryiko <andrii.nakryiko@gmail.com>
Cc: bpf <bpf@vger.kernel.org>, Networking <netdev@vger.kernel.org>,
	Alexei Starovoitov <ast@kernel.org>,
	Daniel Borkmann <daniel@iogearbox.net>,
	Kernel Team <kernel-team@fb.com>, Martin KaFai Lau <kafai@fb.com>
Subject: Re: [PATCH bpf-next v3 13/15] tools/bpf: selftests: implement sample tcp/tcp6 bpf_iter programs
Date: Tue, 23 Jun 2020 08:03:09 -0700	[thread overview]
Message-ID: <6ff28837-63b1-754d-17aa-fd5877409b64@fb.com> (raw)
In-Reply-To: <CAEf4BzatNEOJSuM2t-1eLQuT4E8gcRLB38B=rqZU3G=vVGkizQ@mail.gmail.com>

On 6/22/20 11:56 PM, Andrii Nakryiko wrote:
> On Mon, Jun 22, 2020 at 5:38 PM Yonghong Song <yhs@fb.com> wrote:
>>
>> In my VM, I got identical result compared to /proc/net/{tcp,tcp6}.
>> For tcp6:
>>    $ cat /proc/net/tcp6
>>      sl  local_address                         remote_address                        st tx_queue rx_queue tr tm->when retrnsmt   uid  timeout inode
>>       0: 00000000000000000000000000000000:0016 00000000000000000000000000000000:0000 0A 00000000:00000000 00:00000001 00000000     0        0 17955 1 000000003eb3102e 100 0 0 10 0
>>
>>    $ cat /sys/fs/bpf/p1
>>      sl  local_address                         remote_address                        st tx_queue rx_queue tr tm->when retrnsmt   uid  timeout inode
>>       0: 00000000000000000000000000000000:0016 00000000000000000000000000000000:0000 0A 00000000:00000000 00:00000000 00000000     0        0 17955 1 000000003eb3102e 100 0 0 10 0
>>
>> For tcp:
>>    $ cat /proc/net/tcp
>>    sl  local_address rem_address   st tx_queue rx_queue tr tm->when retrnsmt   uid  timeout inode
>>     0: 00000000:0016 00000000:0000 0A 00000000:00000000 00:00000000 00000000     0        0 2666 1 000000007152e43f 100 0 0 10 0
>>    $ cat /sys/fs/bpf/p2
>>    sl  local_address                         remote_address                        st tx_queue rx_queue tr tm->when retrnsmt   uid  timeout inode
>>     1: 00000000:0016 00000000:0000 0A 00000000:00000000 00:00000000 00000000     0        0 2666 1 000000007152e43f 100 0 0 10 0
>>
>> Signed-off-by: Yonghong Song <yhs@fb.com>
>> ---
> 
> Looks reasonable, to the extent possible ;)
> 
> Acked-by: Andrii Nakryiko <andriin@fb.com>
> 
>>   tools/testing/selftests/bpf/progs/bpf_iter.h  |  15 ++
>>   .../selftests/bpf/progs/bpf_iter_tcp4.c       | 235 ++++++++++++++++
>>   .../selftests/bpf/progs/bpf_iter_tcp6.c       | 250 ++++++++++++++++++
>>   3 files changed, 500 insertions(+)
>>   create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_tcp4.c
>>   create mode 100644 tools/testing/selftests/bpf/progs/bpf_iter_tcp6.c
>>
> 
> [...]
> 
>> +static int hlist_unhashed_lockless(const struct hlist_node *h)
>> +{
>> +        return !(h->pprev);
>> +}
>> +
>> +static int timer_pending(const struct timer_list * timer)
>> +{
>> +       return !hlist_unhashed_lockless(&timer->entry);
>> +}
>> +
>> +extern unsigned CONFIG_HZ __kconfig __weak;
> 
> Why the __weak? We expect to have /proc/kconfig.gz in other tests
> anyway? __weak will make CONFIG_HZ to be a zero and you'll get a bunch
> of divisions by zero.

Make sense. Will change.

> 
>> +
>> +#define USER_HZ                100
>> +#define NSEC_PER_SEC   1000000000ULL
>> +static clock_t jiffies_to_clock_t(unsigned long x)
>> +{
>> +       /* The implementation here tailored to a particular
>> +        * setting of USER_HZ.
>> +        */
>> +       u64 tick_nsec = (NSEC_PER_SEC + CONFIG_HZ/2) / CONFIG_HZ;
>> +       u64 user_hz_nsec = NSEC_PER_SEC / USER_HZ;
>> +
>> +       if ((tick_nsec % user_hz_nsec) == 0) {
>> +               if (CONFIG_HZ < USER_HZ)
>> +                       return x * (USER_HZ / CONFIG_HZ);
>> +               else
>> +                       return x / (CONFIG_HZ / USER_HZ);
>> +       }
>> +       return x * tick_nsec/user_hz_nsec;
>> +}
>> +
> 
> [...]
> 
>> +       if (sk_common->skc_family != AF_INET)
>> +               return 0;
>> +
>> +       tp = bpf_skc_to_tcp_sock(sk_common);
>> +       if (tp) {
>> +               return dump_tcp_sock(seq, tp, uid, seq_num);
>> +       }
> 
> nit: unnecessary {}
> 
>> +
>> +       tw = bpf_skc_to_tcp_timewait_sock(sk_common);
>> +       if (tw)
>> +               return dump_tw_sock(seq, tw, uid, seq_num);
>> +
>> +       req = bpf_skc_to_tcp_request_sock(sk_common);
>> +       if (req)
>> +               return dump_req_sock(seq, req, uid, seq_num);
>> +
>> +       return 0;
>> +}
> 
> [...]
> 
>> +static int timer_pending(const struct timer_list * timer)
>> +{
>> +       return !hlist_unhashed_lockless(&timer->entry);
>> +}
>> +
>> +extern unsigned CONFIG_HZ __kconfig __weak;
> 
> same about __weak here
> 
>> +
>> +#define USER_HZ                100
>> +#define NSEC_PER_SEC   1000000000ULL
>> +static clock_t jiffies_to_clock_t(unsigned long x)
>> +{
>> +       /* The implementation here tailored to a particular
>> +        * setting of USER_HZ.
>> +        */
>> +       u64 tick_nsec = (NSEC_PER_SEC + CONFIG_HZ/2) / CONFIG_HZ;
>> +       u64 user_hz_nsec = NSEC_PER_SEC / USER_HZ;
>> +
>> +       if ((tick_nsec % user_hz_nsec) == 0) {
>> +               if (CONFIG_HZ < USER_HZ)
>> +                       return x * (USER_HZ / CONFIG_HZ);
>> +               else
>> +                       return x / (CONFIG_HZ / USER_HZ);
>> +       }
>> +       return x * tick_nsec/user_hz_nsec;
>> +}
> 
> nit: jiffies_to_clock_t() implementation looks like an overkill for
> this use case... Would it be just `x / CONFIG_HZ * NSEC_PER_SEC` with
> some potential rounding error?

We really want to have the output the same as /proc/net/{tcp,tcp6}.
Otherwise, it may cause confusion when comparing bpf_iter_tcp[6] outputs
vs. /proc/net/tcp[6] outputs.

> 
>> +
>> +static clock_t jiffies_delta_to_clock_t(long delta)
>> +{
>> +       if (delta <= 0)
>> +               return 0;
>> +
>> +       return jiffies_to_clock_t(delta);
>> +}
>> +
> 
> [...]
> 