index | | raw e-mail
I don=E2=80=99t really understand why, but this commit causes bricoler te=
st =
runs to fail for me.
Or more precisely, it looks like parallel test runs see a number of =
failures in the pf tests.
Reverting only part of this change, specifically this:
diff --git a/sys/net/if_clone.c b/sys/net/if_clone.c
index db3db78c1b76..d0fe54b6146b 100644
--- a/sys/net/if_clone.c
+++ b/sys/net/if_clone.c
@@ -683,11 +683,12 @@ if_clone_detach(struct if_clone *ifc)
V_if_cloners_count--;
IF_CLONERS_UNLOCK();
- sx_xlock(&ifnet_detach_sxlock);
/* destroy all interfaces for this cloner */
- while (!LIST_EMPTY(&ifc->ifc_iflist))
+ while (!LIST_EMPTY(&ifc->ifc_iflist)) {
+ sx_xlock(&ifnet_detach_sxlock);
if_clone_destroyif_flags(ifc, =
LIST_FIRST(&ifc->ifc_iflist), IFC_F_FORCE);
- sx_xunlock(&ifnet_detach_sxlock);
+ sx_xunlock(&ifnet_detach_sxlock);
+ }
IF_CLONE_REMREF(ifc);
}
Seems to be enough to avoid the problem. So it looks like somehow =
deleting all interfaces for a given cloner under that lock (rather than =
releasing it and re-acquiring it) causes the test failures.
The failing tests I see are:
sys/netpfil/pf/divert-to:in_div -> failed: Test case body returned a =
non-ok exit code, but this is not allowed [36.591s]
sys/netpfil/pf/divert-to:in_div_in -> failed: atf-check failed; see =
the output of the test for details [30.867s]
sys/netpfil/pf/divert-to:in_div_in_fwd_out_div_out -> failed: =
atf-check failed; see the output of the test for details [39.705s]
sys/netpfil/pf/divert-to:in_dn_in_div_in_out_div_out_dn_out -> =
failed: atf-check failed; see the output of the test for details =
[45.059s]
sys/netpfil/pf/divert-to:out_div -> failed: Test case body returned a =
non-ok exit code, but this is not allowed [23.629s]
sys/netpfil/pf/killstate:gateway -> failed: Killing with a different =
gateway removed the state. [32.803s]
sys/netpfil/pf/killstate:id -> failed: Killing a different ID removed =
the state. [32.789s]
sys/netpfil/pf/killstate:label -> failed: Killing a non-existing =
label removed the state. [47.749s]
sys/netpfil/pf/killstate:multilabel -> failed: Setting new rules =
removed the state. [43.570s]
sys/netpfil/pf/killstate:src_dst -> failed: Killing with the wrong =
destination IP removed our state. [57.819s]
sys/netpfil/pf/killstate:v6 -> failed: Killing with the wrong IP =
removed our state. [54.957s]
sys/netpfil/pf/nat64:pool -> failed: atf-check failed; see the output =
of the test for details [51.118s]
sys/netpfil/pf/nat64:sctp_in -> failed: Failed to connect to SCTP =
server [55.495s]
sys/netpfil/pf/nat64.py:TestNAT64::test_udp_checksum -> failed: =
/usr/tests/sys/netpfil/pf/nat64.py:205: AttributeError [80.001s]
sys/netpfil/pf/pflog:matches -> failed: atf-check failed; see the =
output of the test for details [61.807s]
sys/netpfil/pf/pflog:matches_logif -> failed: atf-check failed; see =
the output of the test for details [51.261s]
sys/netpfil/pf/pflog:unspecified_v6 -> failed: atf-check failed; see =
the output of the test for details [36.281s]
sys/netpfil/pf/rdr:srcport_pass -> failed: atf-check failed; see the =
output of the test for details [76.815s]
sys/netpfil/pf/sctp:pfsync -> failed: Initial SCTP connection failed =
[134.866s]
sys/netpfil/pf/sctp:related_icmp -> failed: SCTP connection failed =
[59.620s]
sys/netpfil/pf/table:zero_one -> failed: atf-check failed; see the =
output of the test for details [64.125s]
sys/netpfil/pf/tcp:rst -> failed: atf-check failed; see the output of =
the test for details [80.261s]
Looking at it now, those runtimes seems suspiciously high as well.
Best regards,
Kristof
--=_MailMate_4D2EEB50-D9A0-4561-A201-3D0996B9026B_=
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: quoted-printable
<!DOCTYPE html>
<html>
<head>
<meta http-equiv=3D"Content-Type" content=3D"text/xhtml; charset=3Dutf-8"=
>
</head>
<body><div style=3D"font-family: sans-serif;"><div class=3D"markdown" sty=
le=3D"white-space: normal;">
<p dir=3D"auto">On 13 Apr 2026, at 6:39, Zhenlei Huang wrote:</p>
<blockquote style=3D"margin: 0 0 5px; padding-left: 5px; border-left: 2px=
solid #136BCE; color: #136BCE;">
<p dir=3D"auto">The branch main has been updated by zlei:</p>
<p dir=3D"auto">URL: <a href=3D"https://cgit.FreeBSD.org/src/commit/?id=3D=
e9fc0c538264355bd3fd9120c650078281c2a290">https://cgit.FreeBSD.org/src/co=
mmit/?id=3De9fc0c538264355bd3fd9120c650078281c2a290</a></p>
<p dir=3D"auto">commit e9fc0c538264355bd3fd9120c650078281c2a290<br>
Author: Zhenlei Huang <a href=3D"mailto:zlei@FreeBSD.org">zlei@FreeBS=
D.org</a><br>
AuthorDate: 2026-04-13 04:38:44 +0000<br>
Commit: Zhenlei Huang <a href=3D"mailto:zlei@FreeBSD.org">zlei@FreeBS=
D.org</a><br>
CommitDate: 2026-04-13 04:38:44 +0000</p>
<pre style=3D"margin-left: 15px; margin-right: 15px; padding: 5px; border=
: thin solid gray; overflow-x: auto; max-width: 90vw; background-color: #=
E4E4E4;"><code style=3D"padding: 0 0.25em; background-color: #E4E4E4;">if=
_clone: Make ifnet_detach_sxlock opaque to consumers
The change e133271fc1b5e introduced ifnet_detach_sxlock, and change
6d2a10d96fb5 widened its coverage, but there are still consumers,
net80211 and tuntap e.g., want it. Instead of sprinkling it everywhere,
make it opaque to consumers.
Out of tree drivers shall also benefit from this change.
Reviewed by: kp
MFC after: 2 weeks
Differential Revision: https://reviews.freebsd.org/D56298
</code></pre>
</blockquote>
<p dir=3D"auto">I don=E2=80=99t really understand why, but this commit ca=
uses bricoler test runs to fail for me.<br>
Or more precisely, it looks like parallel test runs see a number of failu=
res in the pf tests.</p>
<p dir=3D"auto">Reverting only part of this change, specifically this:</p=
>
<pre style=3D"margin-left: 15px; margin-right: 15px; padding: 5px; border=
: thin solid gray; overflow-x: auto; max-width: 90vw; background-color: #=
E4E4E4;"><code style=3D"padding: 0 0.25em; background-color: #E4E4E4;">di=
ff --git a/sys/net/if_clone.c b/sys/net/if_clone.c
index db3db78c1b76..d0fe54b6146b 100644
--- a/sys/net/if_clone.c
+++ b/sys/net/if_clone.c
@@ -683,11 +683,12 @@ if_clone_detach(struct if_clone *ifc)
V_if_cloners_count--;
IF_CLONERS_UNLOCK();
- sx_xlock(&ifnet_detach_sxlock);
/* destroy all interfaces for this cloner */
- while (!LIST_EMPTY(&ifc->ifc_iflist))
+ while (!LIST_EMPTY(&ifc->ifc_iflist)) {
+ sx_xlock(&ifnet_detach_sxlock);
if_clone_destroyif_flags(ifc, LIST_FIRST(&ifc->ifc=
_iflist), IFC_F_FORCE);
- sx_xunlock(&ifnet_detach_sxlock);
+ sx_xunlock(&ifnet_detach_sxlock);
+ }
IF_CLONE_REMREF(ifc);
}
</code></pre>
<p dir=3D"auto">Seems to be enough to avoid the problem. So it looks like=
somehow deleting all interfaces for a given cloner under that lock (rath=
er than releasing it and re-acquiring it) causes the test failures.</p>
<p dir=3D"auto">The failing tests I see are:</p>
<pre style=3D"margin-left: 15px; margin-right: 15px; padding: 5px; border=
: thin solid gray; overflow-x: auto; max-width: 90vw; background-color: #=
E4E4E4;"><code style=3D"padding: 0 0.25em; background-color: #E4E4E4;">sy=
s/netpfil/pf/divert-to:in_div -> failed: Test case body returned a n=
on-ok exit code, but this is not allowed [36.591s]
sys/netpfil/pf/divert-to:in_div_in -> failed: atf-check failed; see =
the output of the test for details [30.867s]
sys/netpfil/pf/divert-to:in_div_in_fwd_out_div_out -> failed: atf-ch=
eck failed; see the output of the test for details [39.705s]
sys/netpfil/pf/divert-to:in_dn_in_div_in_out_div_out_dn_out -> faile=
d: atf-check failed; see the output of the test for details [45.059s]
sys/netpfil/pf/divert-to:out_div -> failed: Test case body returned =
a non-ok exit code, but this is not allowed [23.629s]
sys/netpfil/pf/killstate:gateway -> failed: Killing with a different=
gateway removed the state. [32.803s]
sys/netpfil/pf/killstate:id -> failed: Killing a different ID remove=
d the state. [32.789s]
sys/netpfil/pf/killstate:label -> failed: Killing a non-existing lab=
el removed the state. [47.749s]
sys/netpfil/pf/killstate:multilabel -> failed: Setting new rules rem=
oved the state. [43.570s]
sys/netpfil/pf/killstate:src_dst -> failed: Killing with the wrong d=
estination IP removed our state. [57.819s]
sys/netpfil/pf/killstate:v6 -> failed: Killing with the wrong IP rem=
oved our state. [54.957s]
sys/netpfil/pf/nat64:pool -> failed: atf-check failed; see the outpu=
t of the test for details [51.118s]
sys/netpfil/pf/nat64:sctp_in -> failed: Failed to connect to SCTP se=
rver [55.495s]
sys/netpfil/pf/nat64.py:TestNAT64::test_udp_checksum -> failed: /usr=
/tests/sys/netpfil/pf/nat64.py:205: AttributeError [80.001s]
sys/netpfil/pf/pflog:matches -> failed: atf-check failed; see the ou=
tput of the test for details [61.807s]
sys/netpfil/pf/pflog:matches_logif -> failed: atf-check failed; see =
the output of the test for details [51.261s]
sys/netpfil/pf/pflog:unspecified_v6 -> failed: atf-check failed; see=
the output of the test for details [36.281s]
sys/netpfil/pf/rdr:srcport_pass -> failed: atf-check failed; see the=
output of the test for details [76.815s]
sys/netpfil/pf/sctp:pfsync -> failed: Initial SCTP connection failed=
[134.866s]
sys/netpfil/pf/sctp:related_icmp -> failed: SCTP connection failed =
[59.620s]
sys/netpfil/pf/table:zero_one -> failed: atf-check failed; see the o=
utput of the test for details [64.125s]
sys/netpfil/pf/tcp:rst -> failed: atf-check failed; see the output o=
f the test for details [80.261s]
</code></pre>
<p dir=3D"auto">Looking at it now, those runtimes seems suspiciously high=
as well.</p>
<p dir=3D"auto">Best regards,<br>
Kristof</p>
</div>
</div>
</body>
</html>
--=_MailMate_4D2EEB50-D9A0-4561-A201-3D0996B9026B_=--
home |
help