Received-SPF: pass (google.com: domain of linux-kernel+bounces-85875-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:40f1:3f00::1 as permitted sender) client-ip=2604:1380:40f1:3f00::1;
Message-ID: <7b4c84ce-6cb1-4791-8821-fd0022238e48@quicinc.com>
Date: Wed, 28 Feb 2024 16:13:39 -0800
Precedence: bulk
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: [PATCH net-next v2] net: Modify mono_delivery_time with
 clockid_delivery_time
Content-Language: en-US
To: Willem de Bruijn <willemdebruijn.kernel@gmail.com>,
        "David S. Miller"
	<davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>, Jakub Kicinski
	<kuba@kernel.org>,
        Paolo Abeni <pabeni@redhat.com>, <netdev@vger.kernel.org>,
        <linux-kernel@vger.kernel.org>, Andrew Halaney <ahalaney@redhat.com>,
        "Martin
 KaFai Lau" <martin.lau@kernel.org>,
        Martin KaFai Lau <martin.lau@linux.dev>
CC: <kernel@quicinc.com>
References: <20240228011219.1119105-1-quic_abchauha@quicinc.com>
 <65df56f6ba002_7162829435@willemb.c.googlers.com.notmuch>
 <f38efc6d-20af-4cc1-9b8a-5fcb676b2845@quicinc.com>
 <65df94185a2c1_b2ad829442@willemb.c.googlers.com.notmuch>
 <8a5d15aa-96e2-4d9c-9479-bf2ba8fb2a79@quicinc.com>
 <65dfcb37844e0_ba1f8294b7@willemb.c.googlers.com.notmuch>
From: "Abhishek Chauhan (ABC)" <quic_abchauha@quicinc.com>
In-Reply-To: <65dfcb37844e0_ba1f8294b7@willemb.c.googlers.com.notmuch>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: 7bit


On 2/28/2024 4:09 PM, Willem de Bruijn wrote:
> Abhishek Chauhan (ABC) wrote:
>>
>>
>> On 2/28/2024 12:14 PM, Willem de Bruijn wrote:
>>> Abhishek Chauhan (ABC) wrote:
>>>>
>>>>
>>>> On 2/28/2024 7:53 AM, Willem de Bruijn wrote:
>>>>> Abhishek Chauhan wrote:
>>>>>> Bridge driver today has no support to forward the userspace timestamp
>>>>>> packets and ends up resetting the timestamp. ETF qdisc checks the
>>>>>> packet coming from userspace and encounters to be 0 thereby dropping
>>>>>> time sensitive packets. These changes will allow userspace timestamps
>>>>>> packets to be forwarded from the bridge to NIC drivers.
>>>>>>
>>>>>> Existing functionality of mono_delivery_time is not altered here
>>>>>> instead just extended with userspace tstamp support for bridge
>>>>>> forwarding path.
>>>>>>
>>>>>> Signed-off-by: Abhishek Chauhan <quic_abchauha@quicinc.com>
>>>>>> ---
>>>>>> Changes since v1 
>>>>>> - Changed the commit subject as i am modifying the mono_delivery_time 
>>>>>>   bit with clockid_delivery_time.
>>>>>> - Took care of suggestion mentioned by Willem to use the same bit for 
>>>>>>   userspace delivery time as there are no conflicts between TCP and 
>>>>>>   SCM_TXTIME, because explicit cmsg makes no sense for TCP and only
>>>>>>   RAW and DGRAM sockets interprets it.
>>>>>
>>>>> The variable rename churn makes it hard to spot the functional
>>>>> changes. Perhaps it makes sense just keep the variable name as is,
>>>>> even though the "mono" is not always technically correct anymore.
>>>>>
>>>>   
>>>>
>>>> I think a better approach would be to keep the variable as ease and add
>>>> comments and documentation in the header file of skbuff.h like 
>>>> how i have done in this patch. The reason why i say this is
>>>> a. We can avoid alot of code churn just to solve this basic problem of 
>>>> propagating timestamp through forwarding bridge path 
>>>> b. Re-use the same variable name and have better documentation 
>>>> c. Complexity will be as minimal as possible.
>>>>
>>>> Let me know what you think. 
>>>
>>> Agreed
>>>
>>
>> Okay i will make the changes accordingly. 
>>>>> Or else to split into two patches. One that renames the field.
>>>>> And one that adds the new behavior of setting the bit for SO_TXTIME.
>>>>>
>>>>
>>>> Regarding the sidenote. I dont see how they are using clock_id to determine 
>>>> if the skb->tstamp is set in monotonic. Please correct me or point me to 
>>>> the piece of code which is doing so.
>>>
>>> That's really out of scope of this series anyway
>>>
>>
>> Sounds good. Really appreciate your review and discussion on this topic. 
>>  
>>>>
>>>> I hope the check against sock_flag is a better implementation as 
>>>> it clearly stats and is inline with the implementation that the tstamp is 
>>>> coming from userspace. 
>>>> skb->mono_delivery_time = sock_flag(sk, SOCK_TXTIME);
>>>
>>> Enabling the socket flag is not sufficient to configure a delivery
>>> time on a packet. A transmit time must be communicated per packet
>>> with cork->transmit_time. And on top of that, it is cheaper to test.
>>
>>
>> So to re-use the same bit of mono_delivery_time. I want to set this bit 
>> when user-space sets the timestamps using SCM_TXTIME. 
>> Is it okay if i do the below when we make skb in ipv4/ipv6 and raw packets
>> to ensure that bridge doesn't reset the packet tstamp or do you have a better 
>> suggestion to set the bit so br_forward_finish does not reset the timestamp. 
>>
>> skb->mono_delivery_time = sock_flag(sk, SOCK_TXTIME);
> 
> I already gave my suggestion.
> 
> The timestamp is passed using sockcm_cookie field transmit_time, which
> is set from a control message of type SCM_TXTIME passed to sendmsg.
> 
> Right above where you wanted to add this check is where skb->tstamp is
> initialized from cork->transmit_time, which got it from this field. So
> clearly that is the indication that a transmit time is set. And the
> field is hot in the cache.

Oh i see what you are saying. I got it . I will make the changes accordingly. 
Let me post patch set v3 after testing it internally. 
Thanks again for all the review comments.