[Differential] [Request, 6 lines] D4825: tcp/lro: Add network driver configurable LRO entry depth

[Sepherosa Ziehau]

On Sat, Jan 9, 2016 at 1:03 AM, Adrian Chadd <adrian.chadd at gmail.com> wrote:
> On 8 January 2016 at 07:59, Ravi Pokala <rpokala at mac.com> wrote:
>> -----Original Message-----
>>
>>
>>>Date: Fri, 8 Jan 2016 02:29:22 +0000
>>>From: "sepherosa_gmail.com (Sepherosa Ziehau)"
>>>       <phabric-noreply at FreeBSD.org>
>>>To: freebsd-net at freebsd.org
>>>Subject: [Differential] [Request, 6 lines] D4825: tcp/lro: Add network
>>>       driver configurable LRO entry depth
>>>Message-ID:
>>>       <differential-rev-PHID-DREV-ou2jiti5cx3pzqhm5pb2-req at FreeBSD.org>
>>>Content-Type: text/plain; charset="utf-8"
>>>
>>>sepherosa_gmail.com created this revision.
>>>sepherosa_gmail.com added reviewers: network, adrian, delphij, decui_microsoft.com, honzhan_microsoft.com, howard0su_gmail.com, glebius.
>>>sepherosa_gmail.com added a subscriber: freebsd-net-list.
>>>Herald added a reviewer: transport.
>>>
>>>REVISION SUMMARY
>>>  When there is only tiny amount of TCP connections and the host is slow, e.g. in VM, holding too much TCP segments in an LRO entry will cause RX performance degradation.  We now allow network drivers to configure how deep one LRO entry should be.
>>
>> Forgive me if I'm missing something obvious, but this patch doesn't actually change anything - rather than hard-coding 65535, you're using lc->lro_hiwat... which is hard-coded to 65535.
>>
>> Right?
>>
>
> Right; I'm assuming their driver will twiddle this appropriately over time.

Yeah, by drivers, and through a driver sysctl