Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 04 Sep 2023 10:15:46 +0000
From:      bugzilla-noreply@freebsd.org
To:        bugs@FreeBSD.org
Subject:   [Bug 273559] [iflib] numa allocation
Message-ID:  <bug-273559-227@https.bugs.freebsd.org/bugzilla/>

next in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D273559

            Bug ID: 273559
           Summary: [iflib] numa allocation
           Product: Base System
           Version: 13.2-STABLE
          Hardware: amd64
                OS: Any
            Status: New
          Severity: Affects Only Me
          Priority: ---
         Component: kern
          Assignee: bugs@FreeBSD.org
          Reporter: konrad.kreciwilk@korbank.pl

Hello,

Based on https://reviews.freebsd.org/D19930 I did a patch for iflib:

diff --git a/sys/net/iflib.c b/sys/net/iflib.c
index d6dcb0931..4437f6e57 100644
--- a/sys/net/iflib.c
+++ b/sys/net/iflib.c
@@ -5953,7 +5953,7 @@ iflib_register(if_ctx_t ctx)
                        type =3D IFT_PPP;
        } else
                type =3D IFT_ETHER;
-       ifp =3D ctx->ifc_ifp =3D if_alloc(type);
+       ifp =3D ctx->ifc_ifp =3D if_alloc_dev(type, dev);
        if (ifp =3D=3D NULL) {
                device_printf(dev, "can not allocate ifnet structure\n");
                return (ENOMEM);

for allocate NUMA local memory. I used it for lagg lacp egress port selecti=
on
based on NUMA domain. So far everything worked fine (stable/13 build 12 Apr
2022). Yesterday I did upgrade to latest stable/13 and patch has stopped
working. The same flow comes to ix1 (numa-domain 0) and goes out ix3
(numa-domain 1). Lagg0 has enabled use_numa (as before). System has a few v=
lans
on lagg0 and does the routing

#dev numa domain
dev.ix.3.%domain: 1
dev.ix.2.%domain: 1
dev.ix.1.%domain: 0
dev.ix.0.%domain: 0

#interfaces
#10G - ix0
ifconfig_ix0=3D"up"
#10G - ix1
ifconfig_ix1=3D"up"
#10G - ix2
ifconfig_ix2=3D"up"
#10G - ix3
ifconfig_ix3=3D"up"

#LAGG LACP
ifconfig_lagg0=3D"laggproto lacp laggport ix0 laggport ix1 laggport ix2 lag=
gport
ix3 up -wol -vlanhwtso -tso -lro"

lagg0: flags=3D8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 15=
00
=20=20=20=20=20=20=20
options=3D4e100bb<RXCSUM,TXCSUM,VLAN_MTU,VLAN_HWTAGGING,JUMBO_MTU,VLAN_HWCS=
UM,VLAN_HWFILTER,RXCSUM_IPV6,TXCSUM_IPV6,NOMAP>
        ether 90:e2:ba:78:9d:2c
        laggproto lacp lagghash l2,l3,l4
        lagg options:
                flags=3D15<USE_FLOWID,USE_NUMA,LACP_STRICT>
                flowid_shift: 16
        lagg statistics:
                active ports: 2
                flapping: 0
        lag id: [(8000,90-E2-BA-78-9D-2C,0152,0000,0000),
                 (0000,00-11-22-33-44-55,000C,0000,0000)]
        laggport: ix0 flags=3D0<> state=3D41<ACTIVITY,DEFAULTED>
                [(8000,90-E2-BA-78-9D-2C,8001,8000,0001),
                 (FFFF,00-00-00-00-00-00,0000,FFFF,0000)]
        laggport: ix1 flags=3D1c<ACTIVE,COLLECTING,DISTRIBUTING>
state=3D3d<ACTIVITY,AGGREGATION,SYNC,COLLECTING,DISTRIBUTING>
                [(8000,90-E2-BA-78-9D-2C,0152,8000,0002),
                 (0000,00-11-22-33-44-55,000C,0000,03EF)]
        laggport: ix2 flags=3D0<> state=3D41<ACTIVITY,DEFAULTED>
                [(8000,90-E2-BA-78-9D-2C,8007,8000,0007),
                 (FFFF,00-00-00-00-00-00,0000,FFFF,0000)]
        laggport: ix3 flags=3D1c<ACTIVE,COLLECTING,DISTRIBUTING>
state=3D3d<ACTIVITY,AGGREGATION,SYNC,COLLECTING,DISTRIBUTING>
                [(8000,90-E2-BA-78-9D-2C,0152,8000,0008),
                 (0000,00-11-22-33-44-55,000C,0000,03EF)]
        groups: lagg
        media: Ethernet autoselect
        status: active
        nd6 options=3D29<PERFORMNUD,IFDISABLED,AUTO_LINKLOCAL>



Where is the problem ?

--=20
You are receiving this mail because:
You are the assignee for the bug.=



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-273559-227>