From: Geliang Tang <geliangtang@gmail.com>
To: Yonglong Li <liyonglong@chinatelecom.cn>
Cc: mptcp@lists.linux.dev
Subject: Re: [MPTCP][PATCH mptcp-next] Squash to "mptcp: remove MPTCP_ADD_ADDR_IPV6 and MPTCP_ADD_ADDR_PORT"
Date: Thu, 15 Jul 2021 11:45:49 +0800 [thread overview]
Message-ID: <CA+WQbwsJJyzpCH8-uiNj9mkd=xHHtk5kq-8ooxJCdN4sRqqkYA@mail.gmail.com> (raw)
In-Reply-To: <f7ef8d86-d644-89c5-6153-f9fb4f31b897@chinatelecom.cn>
Yonglong Li <liyonglong@chinatelecom.cn> 于2021年7月14日周三 下午5:49写道:
>
> Hi Geliang,
>
> On 2021/7/14 11:10, Geliang Tang wrote:
> > Hi Yonglong,
> >
> > Yonglong Li <liyonglong@chinatelecom.cn> 于2021年7月13日周二 下午6:30写道:
> >>
> >>
> >>
> >> On 2021/7/13 14:44, Geliang Tang wrote:
> >>> Keep the debug info for "send ack".
> >>>
> >>> Don't drop mptcp_pm_should_add_signal_ipv6() and
> >>> mptcp_pm_should_add_signal_port().
> >>>
> >>> Signed-off-by: Geliang Tang <geliangtang@gmail.com>
> >>> ---
> >>> net/mptcp/pm_netlink.c | 7 +++++--
> >>> 1 file changed, 5 insertions(+), 2 deletions(-)
> >>>
> >>> diff --git a/net/mptcp/pm_netlink.c b/net/mptcp/pm_netlink.c
> >>> index 3e35720317ae..2cd6caaedb08 100644
> >>> --- a/net/mptcp/pm_netlink.c
> >>> +++ b/net/mptcp/pm_netlink.c
> >>> @@ -543,8 +543,11 @@ void mptcp_pm_nl_addr_send_ack(struct mptcp_sock *msk)
> >>> bool slow;
> >>>
> >>> spin_unlock_bh(&msk->pm.lock);
> >>> - pr_debug("send ack for %s",
> >>> - mptcp_pm_should_add_signal(msk) ? "add_addr" : "rm_addr");
> >>> + pr_debug("send ack for %s%s%s",
> >>> + mptcp_pm_should_add_signal_addr(msk) ? "add_addr" :
> >>> + (mptcp_pm_should_add_signal_echo(msk) ? "add_echo" : "rm_addr"),
> >>> + mptcp_pm_should_add_signal_ipv6(msk) ? " [ipv6]" : "",
> >>> + mptcp_pm_should_add_signal_port(msk) ? " [port]" : "");
> >>>
> >>> slow = lock_sock_fast(ssk);
> >>> tcp_send_ack(ssk);
> >>>
> >> Hi Geliang,
> >>
> >> I think the debug log will be incorrect if add_addr and add_echo events trigger at the same time.
> >> WDYT?
> >
> > Yes, how about moving this pr_debug line under the pm lock, just swap this
> > pr_debug line with the spin_unlock_bh line?
> >
> > If so, no need to use READ_ONCE() in mptcp_pm_should_add_signal_ipv6/port
> > too.
> I prefer to remove ipv6 check just like v8. if you want to debug we can get more detail info
> from debug log in mptcp_established_options_add_addr.
Sure, I had removed mptcp_pm_should_add_signal_ipv6/port in v9.
>
> >
> >
> > I had tested this squash-to patches all night yesterday. And I got this
> > error in the debug log:
> >
> > add_signal error, add_addr=2, echo=1
> >
> > This means the race occurs, right?
> >
> It seams like anther race case or bug? if more than one 'echo add addr' event be trigger at the same
> time this log will be show.
And I had used your testcase to test v9 all night yesterday, and everything
is fine, no this error anymore. It has been tested over 18000 times. So I
think v9 works well. I'll do more test and send out v9 recently.
Do you think 18000 times is enough? How many times do you usually test
before?
>
> > Does it mean that this version cannot deal with the race either? I'm not
> > sure.
> >
> > Could you please share your test scripts to me?
> >
> I add a test case in mptcp_join.sh, and run 'mptcp_join.sh -X -c' in a loop, if test case failed the race maybe occurs.
> and then analsys the pcap file to check the race.
>
> the test case add in mptcp_join.sh:
>
> diff --git a/mptcp_join.sh b/mptcp_join.sh
> index 523c779..162a451 100755
> --- a/tools/testing/selftests/net/mptcp/mptcp_join.sh
> +++ b/tools/testing/selftests/net/mptcp/mptcp_join.sh
> @@ -1462,6 +1462,24 @@ checksum_tests()
> chk_csum_nr "checksum test 1 0"
> }
>
> +xxoo()
> +{
> + reset
> + ip netns exec $ns1 ./pm_nl_ctl limits 4 4
> + ip netns exec $ns2 ./pm_nl_ctl limits 4 4
> + ip netns exec $ns1 ./pm_nl_ctl add 10.0.1.1 flags signal
> + ip netns exec $ns1 ./pm_nl_ctl add 10.0.2.1 flags signal
> + ip netns exec $ns1 ./pm_nl_ctl add 10.0.3.1 flags signal
> + ip netns exec $ns1 ./pm_nl_ctl add 10.0.4.1 flags signal
> + ip netns exec $ns2 ./pm_nl_ctl add 10.0.1.2 flags signal
> + ip netns exec $ns2 ./pm_nl_ctl add 10.0.2.2 flags signal
> + ip netns exec $ns2 ./pm_nl_ctl add 10.0.3.2 flags signal
> + ip netns exec $ns2 ./pm_nl_ctl add 10.0.4.2 flags signal
> + run_tests $ns1 $ns2 10.0.1.1
> + chk_add_nr 4 4
> +}
I think this testcase is worth to be added into the selftests script at
the end of the function signal_address_tests() as an extra signal address
test. Do you agree?
If so, I'll add this testcase into v9 and sign your name as the author.
Thanks,
-Geliang
> +
> +
> all_tests()
> {
> subflows_tests
> @@ -1566,6 +1584,9 @@ while getopts 'fsltra64bpkchCS' opt; do
> S)
> checksum_tests
> ;;
> + X)
> + xxoo
> + ;;
> c)
> ;;
> C)
>
>
> > Thanks,
> > -Geliang
> >
> >
> >
> >>
> >> --
> >> Li YongLong
> >
> >
>
> --
> Li YongLong
>
next prev parent reply other threads:[~2021-07-15 3:46 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-07-13 6:44 [MPTCP][PATCH v2 mptcp-next] Squash to "mptcp: fix ADD_ADDR and RM_ADDR maybe flush addr_signal each other" Geliang Tang
2021-07-13 6:44 ` [MPTCP][PATCH v2 mptcp-next] Squash to "mptcp: build ADD_ADDR/echo-ADD_ADDR option according pm.add_signal" Geliang Tang
2021-07-13 7:32 ` Geliang Tang
2021-07-13 20:39 ` Mat Martineau
2021-07-15 3:17 ` Geliang Tang
2021-07-13 10:30 ` Yonglong Li
2021-07-13 6:44 ` [MPTCP][PATCH mptcp-next] Squash to "mptcp: remove MPTCP_ADD_ADDR_IPV6 and MPTCP_ADD_ADDR_PORT" Geliang Tang
2021-07-13 10:30 ` Yonglong Li
2021-07-14 3:10 ` Geliang Tang
2021-07-14 9:49 ` Yonglong Li
2021-07-15 3:45 ` Geliang Tang [this message]
2021-07-15 6:13 ` Yonglong Li
2021-07-13 7:30 ` [MPTCP][PATCH v2 mptcp-next] Squash to "mptcp: fix ADD_ADDR and RM_ADDR maybe flush addr_signal each other" Geliang Tang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CA+WQbwsJJyzpCH8-uiNj9mkd=xHHtk5kq-8ooxJCdN4sRqqkYA@mail.gmail.com' \
--to=geliangtang@gmail.com \
--cc=liyonglong@chinatelecom.cn \
--cc=mptcp@lists.linux.dev \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).