| Message ID | 20240423130604.7013-1-fw@strlen.de |
|---|---|
| Headers | show |
| Series | selftest: netfilter: additional cleanups | expand |
On Tue, 23 Apr 2024 15:05:43 +0200 Florian Westphal wrote: > This is the last planned series of the netfilter-selftest-move. > It contains cleanups (and speedups) and a few small updates to > scripts to improve error/skip reporting. > > I intend to route future changes, if any, via nf(-next) trees > now that the 'massive code churn' phase is over. Got it. The main thing that seems to be popping up in the netdev runner is: # TEST: performance # net,port [SKIP] # perf not supported What is "perf" in this case? Some NFT module? the perf tool is installed, AFAICT..
Jakub Kicinski <kuba@kernel.org> wrote: > On Tue, 23 Apr 2024 15:05:43 +0200 Florian Westphal wrote: > > This is the last planned series of the netfilter-selftest-move. > > It contains cleanups (and speedups) and a few small updates to > > scripts to improve error/skip reporting. > > > > I intend to route future changes, if any, via nf(-next) trees > > now that the 'massive code churn' phase is over. > > Got it. > > The main thing that seems to be popping up in the netdev runner is: > > # TEST: performance > # net,port [SKIP] > # perf not supported > > What is "perf" in this case? Some NFT module? the perf tool is > installed, AFAICT.. Its looking for the pktgen wrapper script (pktgen_bench_xmit_mode_netif_receive.sh). I don't think it makes too much sense to have that run as part of the CI. I can either remove this or move it under some special commandline option, or I can look into this and see if I can get it to run. Let me know, thanks.
On Tue, 23 Apr 2024 21:42:21 +0200 Florian Westphal wrote: > > The main thing that seems to be popping up in the netdev runner is: > > > > # TEST: performance > > # net,port [SKIP] > > # perf not supported > > > > What is "perf" in this case? Some NFT module? the perf tool is > > installed, AFAICT.. > > Its looking for the pktgen wrapper script > (pktgen_bench_xmit_mode_netif_receive.sh). > > I don't think it makes too much sense to have that run as part of the CI. > > I can either remove this or move it under some special commandline > option, or I can look into this and see if I can get it to run. Hm, never used it myself but it makes me think of the extended ksft vars: | TEST_PROGS_EXTENDED, TEST_GEN_PROGS_EXTENDED mean it is the | executable which is not tested by default. https://docs.kernel.org/dev-tools/kselftest.html?highlight=test_progs_extended#contributing-new-tests-details
Hello: This series was applied to netdev/net-next.git (main) by Jakub Kicinski <kuba@kernel.org>: On Tue, 23 Apr 2024 15:05:43 +0200 you wrote: > This is the last planned series of the netfilter-selftest-move. > It contains cleanups (and speedups) and a few small updates to > scripts to improve error/skip reporting. > > I intend to route future changes, if any, via nf(-next) trees > now that the 'massive code churn' phase is over. > > [...] Here is the summary with links: - [net-next,1/7] selftests: netfilter: nft_concat_range.sh: move to lib.sh infra https://git.kernel.org/netdev/net-next/c/546fb63fe85e - [net-next,2/7] selftests: netfilter: nft_concat_range.sh: drop netcat support https://git.kernel.org/netdev/net-next/c/ba6fbd383c12 - [net-next,3/7] selftests: netfilter: nft_concat_range.sh: shellcheck cleanups https://git.kernel.org/netdev/net-next/c/c54fa6ae35b9 - [net-next,4/7] selftests: netfilter: nft_flowtable.sh: re-run with random mtu sizes https://git.kernel.org/netdev/net-next/c/f84ab634904c - [net-next,5/7] selftests: netfilter: nft_flowtable.sh: shellcheck cleanups https://git.kernel.org/netdev/net-next/c/a18f284574ad - [net-next,6/7] selftests: netfilter: skip tests on early errors https://git.kernel.org/netdev/net-next/c/bb0ee78f9418 - [net-next,7/7] selftests: netfilter: conntrack_vrf.sh: prefer socat, not iperf3 https://git.kernel.org/netdev/net-next/c/99bc5950ebd4 You are awesome, thank you!
On Tue, 23 Apr 2024 15:05:43 +0200 Florian Westphal wrote: > I intend to route future changes, if any, via nf(-next) trees > now that the 'massive code churn' phase is over. I hooked the tests which reliably pass up to patchwork now. (The PW reporting is "combined" so basically the daemon will take them into account for the "contest" check). On the debug kernels we have a bunch of tests failing due to timeout. We multiply the timeout set in ksft settings by 2 for the debug kernel, so in netfilter we give them 1000 sec, but looks like that's not enough. Would you be willing to bump it up? I can override it in the config of the runner, as last resort.
Jakub Kicinski <kuba@kernel.org> wrote: > On Tue, 23 Apr 2024 21:42:21 +0200 Florian Westphal wrote: > > > The main thing that seems to be popping up in the netdev runner is: > > > > > > # TEST: performance > > > # net,port [SKIP] > > > # perf not supported > > > > > > What is "perf" in this case? Some NFT module? the perf tool is > > > installed, AFAICT.. > > > > Its looking for the pktgen wrapper script > > (pktgen_bench_xmit_mode_netif_receive.sh). > > > > I don't think it makes too much sense to have that run as part of the CI. > > > > I can either remove this or move it under some special commandline > > option, or I can look into this and see if I can get it to run. It runs for ~25m, so I'd say lets NOT have it run by default. > Hm, never used it myself but it makes me think of the extended ksft > vars: > > | TEST_PROGS_EXTENDED, TEST_GEN_PROGS_EXTENDED mean it is the > | executable which is not tested by default. > > https://docs.kernel.org/dev-tools/kselftest.html?highlight=test_progs_extended#contributing-new-tests-details Thanks, I'll add a small wrapper via TEST_PROGS_EXTENDED so folks can run the performance (pps match rate) manually.