[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <f3dd8b70-f44b-128a-42a5-98135d457ffd@linux.dev>
Date: Tue, 18 Oct 2022 18:30:07 -0700
From: Martin KaFai Lau <martin.lau@...ux.dev>
To: wangyufen <wangyufen@...wei.com>
Cc: Lina Wang <lina.wang@...iatek.com>, bpf@...r.kernel.org,
davem@...emloft.net, edumazet@...gle.com, pabeni@...hat.com,
deso@...teo.net, netdev@...r.kernel.org,
Jakub Kicinski <kuba@...nel.org>
Subject: Re: [net 1/2] selftests/net: fix opening object file failed
On 10/18/22 11:00 AM, Jakub Kicinski wrote:
> On Tue, 18 Oct 2022 17:50:19 +0800 wangyufen wrote:
>> So, there are two possible approaches: the first moving nat6to4.c and
>> the actual test programs to selftests/bpf;
>>
>> second add make dependency on libbpf for the nat6to4.c.
>>
>> Which one is better?
>
> Can we move the programs and create a dependency from them back
> to networking? Perhaps shared components like udpgso_* need to live
> under tools/net so they can be easily "depended on"?
>
> Either that or they need to switch to a different traffic generator for
> the BPF test, cause there's more networking selftests using the UDP
> generators :(
All (at least most) of the selftests/bpf/test_prog's tests generate its own
traffic for unit test purpose such that each test is self contained. The
udpgro_frglist test should do the same in selftests/bpf/test_prog (meaning the
test itself should generate its own testing traffic). Also, it does not look
like it is actually using udpgso_bench_* to do benchmarking.
Powered by blists - more mailing lists