Re: [PATCH net-next 2/3] bpf: add TCP_BPF_RTO_MAX for bpf_setsockopt

[Martin KaFai Lau <martin.lau@linux.dev>]

On 2/13/25 10:12 PM, Jason Xing wrote:
> On Fri, Feb 14, 2025 at 1:41 PM Martin KaFai Lau <martin.lau@linux.dev> wrote:
>>
>> On 2/13/25 7:09 PM, Jason Xing wrote:
>>> On Fri, Feb 14, 2025 at 10:14 AM Martin KaFai Lau <martin.lau@linux.dev> wrote:
>>>>
>>>> On 2/13/25 3:57 PM, Jason Xing wrote:
>>>>> On Fri, Feb 14, 2025 at 7:41 AM Stanislav Fomichev<stfomichev@gmail.com> wrote:
>>>>>> On 02/13, Jason Xing wrote:
>>>>>>> Support bpf_setsockopt() to set the maximum value of RTO for
>>>>>>> BPF program.
>>>>>>>
>>>>>>> Signed-off-by: Jason Xing<kerneljasonxing@gmail.com>
>>>>>>> ---
>>>>>>>     Documentation/networking/ip-sysctl.rst | 3 ++-
>>>>>>>     include/uapi/linux/bpf.h               | 2 ++
>>>>>>>     net/core/filter.c                      | 6 ++++++
>>>>>>>     tools/include/uapi/linux/bpf.h         | 2 ++
>>>>>>>     4 files changed, 12 insertions(+), 1 deletion(-)
>>>>>>>
>>>>>>> diff --git a/Documentation/networking/ip-sysctl.rst b/Documentation/networking/ip-sysctl.rst
>>>>>>> index 054561f8dcae..78eb0959438a 100644
>>>>>>> --- a/Documentation/networking/ip-sysctl.rst
>>>>>>> +++ b/Documentation/networking/ip-sysctl.rst
>>>>>>> @@ -1241,7 +1241,8 @@ tcp_rto_min_us - INTEGER
>>>>>>>
>>>>>>>     tcp_rto_max_ms - INTEGER
>>>>>>>          Maximal TCP retransmission timeout (in ms).
>>>>>>> -     Note that TCP_RTO_MAX_MS socket option has higher precedence.
>>>>>>> +     Note that TCP_BPF_RTO_MAX and TCP_RTO_MAX_MS socket option have the
>>>>>>> +     higher precedence for configuring this setting.
>>>>>> The cover letter needs more explanation about the motivation.
>>>>
>>>> +1
>>>>
>>>> I haven't looked at the patches. The cover letter has no word on the use case.
>>
>> The question was your _use case_ in bpf. Not what the TCP_RTO_MAX_MS does. Your
>> current use case is to have bpf setting it after reading the tcp header option,
>> like the selftest in patch 3?
> 
> Oops, I misunderstood the real situation of the tcp header option
> test. My intention is to bpf_setsockopt() just like setget_sockopt
> does.
> 
> Thanks for reminding me. I will totally remove the header test in the
> next version.

If your use case was in the header, it is ok although it won't be the first 
useful place I have in my mind. Regardless, it is useful to say a few words 
where you are planning to set it in the bpf. During a cb in sockops or during 
socket create ...etc. Without it, we can only guess from the selftest :(

> 
>>
>>>
>>> I will add and copy some words from Eric's patch series :)
>>
>>
>>>>> I am targeting the net-next tree because of recent changes[1] made by
>>>>> Eric. It probably hasn't merged into the bpf-next tree.
>>>>
>>>> There is the bpf-next/net tree. It should have the needed changes.
>>>
>>> [1] was recently merged in the net-next tree, so the only one branch I
>>> can target is net-next.
>>>
>>> [1]: https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net-next.git/commit/?id=ae9b3c0e79bc
>>>
>>> Am I missing something?
>>
>> There is a net branch:
               ^^^

>> https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next.git
> 
> But this branch hasn't included the rto max feature. I was trying to

Which branch? I was talking about the **net** branch. Not the master branch. Try 
to pull again if your local copy does not have it. The net branch should have 
the TCP_RTO_MAX_MS patches.

> say that what I wrote is based on the rto max feature which only
> exists in the net-next tree for now.
>