Jakub Kicinski wrote:
> On Tue, 14 May 2024 14:13:21 +0200 Geert Uytterhoeven wrote:
>> On Thu, May 9, 2024 at 9:09 PM Richard Gobert <richardbgobert@xxxxxxxxx> wrote:
>>> {inet,ipv6}_gro_receive functions perform flush checks (ttl, flags,
>>> iph->id, ...) against all packets in a loop. These flush checks are used in
>>> all merging UDP and TCP flows.
>>>
>>> These checks need to be done only once and only against the found p skb,
>>> since they only affect flush and not same_flow.
>>>
>>> This patch leverages correct network header offsets from the cb for both
>>> outer and inner network headers - allowing these checks to be done only
>>> once, in tcp_gro_receive and udp_gro_receive_segment. As a result,
>>> NAPI_GRO_CB(p)->flush is not used at all. In addition, flush_id checks are
>>> more declarative and contained in inet_gro_flush, thus removing the need
>>> for flush_id in napi_gro_cb.
>>>
>>> This results in less parsing code for non-loop flush tests for TCP and UDP
>>> flows.
>>>
>>> To make sure results are not within noise range - I've made netfilter drop
>>> all TCP packets, and measured CPU performance in GRO (in this case GRO is
>>> responsible for about 50% of the CPU utilization).
>>>
>>> perf top while replaying 64 parallel IP/TCP streams merging in GRO:
>>> (gro_receive_network_flush is compiled inline to tcp_gro_receive)
>>> net-next:
>>> 6.94% [kernel] [k] inet_gro_receive
>>> 3.02% [kernel] [k] tcp_gro_receive
>>>
>>> patch applied:
>>> 4.27% [kernel] [k] tcp_gro_receive
>>> 4.22% [kernel] [k] inet_gro_receive
>>>
>>> perf top while replaying 64 parallel IP/IP/TCP streams merging in GRO (same
>>> results for any encapsulation, in this case inet_gro_receive is top
>>> offender in net-next)
>>> net-next:
>>> 10.09% [kernel] [k] inet_gro_receive
>>> 2.08% [kernel] [k] tcp_gro_receive
>>>
>>> patch applied:
>>> 6.97% [kernel] [k] inet_gro_receive
>>> 3.68% [kernel] [k] tcp_gro_receive
>>>
>>> Signed-off-by: Richard Gobert <richardbgobert@xxxxxxxxx>
>>
>> Thanks for your patch, which is now commit 4b0ebbca3e167976 ("net: gro:
>> move L3 flush checks to tcp_gro_receive and udp_gro_receive_segment")
>> in net-next/main (next-20240514).
>>
>> noreply@xxxxxxxxxxxxxx reports build failures on m68k, e.g.
>> http://kisskb.ellerman.id.au/kisskb/buildresult/15168903/
>>
>> net/core/gro.c: In function ‘dev_gro_receive’:
>> ././include/linux/compiler_types.h:460:38: error: call to
>> ‘__compiletime_assert_654’ declared with attribute error: BUILD_BUG_ON
>> failed: !IS_ALIGNED(offsetof(struct napi_gro_cb, zeroed), sizeof(u32))
>
> Hi Richard, any chance of getting this fixed within the next 2 hours?
> I can't send the net-next PR if it doesn't build on one of the arches..
Hi Jakub and Geert,
I'm only seeing this mail now, sorry for the late response.
I can fix this within the next two hours, would you prefer a standalone
patch or should I add it to this patch series?
