On 11/20/2018 07:42 AM, Paolo Abeni wrote:
> Hi,
>
> On Tue, 2018-11-20 at 05:49 -0800, Eric Dumazet wrote:
>>
>> On 11/20/2018 02:17 AM, Paolo Abeni wrote:
>>> Eric noted that with UDP GRO and napi timeout, we could keep a single
>>> UDP packet inside the GRO hash forever, if the related NAPI instance
>>> calls napi_gro_complete() at an higher frequency than the napi timeout.
>>> Willem noted that even TCP packets could be trapped there, till the
>>> next retransmission.
>>> This patch tries to address the issue, flushing the oldest packets before
>>> scheduling the NAPI timeout. The rationale is that such a timeout should be
>>> well below a jiffy and we are not flushing packets eligible for sane GRO.
>>>
>>> Reported-by: Eric Dumazet <eric.duma...@gmail.com>
>>> Signed-off-by: Paolo Abeni <pab...@redhat.com>
>>> ---
>>> Sending as RFC, as I fear I'm missing some relevant pieces.
>>> Also I'm unsure if this should considered a fixes for "udp: implement
>>> GRO for plain UDP sockets." or for "net: gro: add a per device gro flush
>>> timer"
>
> Thank you for your feedback!
>
>> Truth be told, relying on jiffies change is a bit fragile for HZ=100 or
>> HZ=250 kernels.
>
> Yes, we have higher bound there.
>
>> See recent TCP commit that got rid of tcp_tso_should_defer() dependency on
>> HZ/jiffies
>>
>> https://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git/commit/?id=a682850a114aef947da5d603f7fd2cfe7eabbd72
>
> I'm unsure I follow correctly. Are you suggesting to use ns precision
> for skb aging in GRO? If so, could that be a separate change? (looks
> more invasive)
I am not suggesting adding ns in your patch.
That can be done later if we care.
I simply warn that some distros have low HZ value and thus the fix wont prevent
packet sitting 4 or 10 ms in the queue.
