[PATCH net v2] vrf: Drop existing dst reference in vrf_ip6_input_dst

[PATCH net v2] vrf: Drop existing dst reference in vrf_ip6_input_dst

[Stanislav Fomichev]

Commit 8fba0fc7b3de ("selftests: tc: Add generic erspan_opts matching support
for tc-flower") started triggering the following kmemleak warning:

unreferenced object 0xffff888015fb0e00 (size 512):
  comm "softirq", pid 0, jiffies 4294679065
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 40 d2 85 9e ff ff ff ff  ........@.......
    41 69 59 9d ff ff ff ff 00 00 00 00 00 00 00 00  AiY.............
  backtrace (crc 30b71e8b):
    __kmalloc_noprof+0x359/0x460
    metadata_dst_alloc+0x28/0x490
    erspan_rcv+0x4f1/0x1160 [ip_gre]
    gre_rcv+0x217/0x240 [ip_gre]
    gre_rcv+0x1b8/0x400 [gre]
    ip_protocol_deliver_rcu+0x31d/0x3a0
    ip_local_deliver_finish+0x37d/0x620
    ip_local_deliver+0x174/0x460
    ip_rcv+0x52b/0x6b0
    __netif_receive_skb_one_core+0x149/0x1a0
    process_backlog+0x3c8/0x1390
    __napi_poll.constprop.0+0xa1/0x390
    net_rx_action+0x59b/0xe00
    handle_softirqs+0x22b/0x630
    do_softirq+0xb1/0xf0
    __local_bh_enable_ip+0x115/0x150

vrf_ip6_input_dst unconditionally sets skb dst entry, add a call to
skb_dst_drop to drop any existing entry.

Cc: David Ahern <dsahern@kernel.org>
Fixes: 9ff74384600a ("net: vrf: Handle ipv6 multicast and link-local addresses")
Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
---
 drivers/net/vrf.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/net/vrf.c b/drivers/net/vrf.c
index 9a4beea6ee0c..3ccd649913b5 100644
--- a/drivers/net/vrf.c
+++ b/drivers/net/vrf.c
@@ -1302,6 +1302,8 @@ static void vrf_ip6_input_dst(struct sk_buff *skb, struct net_device *vrf_dev,
 	struct net *net = dev_net(vrf_dev);
 	struct rt6_info *rt6;
 
+	skb_dst_drop(skb);
+
 	rt6 = vrf_ip6_route_lookup(net, vrf_dev, &fl6, ifindex, skb,
 				   RT6_LOOKUP_F_HAS_SADDR | RT6_LOOKUP_F_IFACE);
 	if (unlikely(!rt6))
-- 
2.50.1

Re: [PATCH net v2] vrf: Drop existing dst reference in vrf_ip6_input_dst

[Ido Schimmel]

On Wed, Jul 23, 2025 at 03:46:25PM -0700, Stanislav Fomichev wrote:
> Commit 8fba0fc7b3de ("selftests: tc: Add generic erspan_opts matching support
> for tc-flower") started triggering the following kmemleak warning:

Did you mean

ff3fbcdd4724 ("selftests: tc: Add generic erspan_opts matching support for tc-flower")

?

[...]

> vrf_ip6_input_dst unconditionally sets skb dst entry, add a call to
> skb_dst_drop to drop any existing entry.
> 
> Cc: David Ahern <dsahern@kernel.org>
> Fixes: 9ff74384600a ("net: vrf: Handle ipv6 multicast and link-local addresses")
> Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>

Reviewed-by: Ido Schimmel <idosch@nvidia.com>

Re: [PATCH net v2] vrf: Drop existing dst reference in vrf_ip6_input_dst

[Stanislav Fomichev]

On 07/24, Ido Schimmel wrote:
> On Wed, Jul 23, 2025 at 03:46:25PM -0700, Stanislav Fomichev wrote:
> > Commit 8fba0fc7b3de ("selftests: tc: Add generic erspan_opts matching support
> > for tc-flower") started triggering the following kmemleak warning:
> 
> Did you mean
> 
> ff3fbcdd4724 ("selftests: tc: Add generic erspan_opts matching support for tc-flower")
> 
> ?

Both seem to be reachable in my local tree:

$ git log --oneline 8fba0fc7b3dec1abc4dd57449d578d8c2682d60e | head
8fba0fc7b3de selftests: tc: Add generic erspan_opts matching support for tc-flower
a6906c86ff9c net: phy: qcom: qca807x: Enable WoL support using shared library
dea37f39a2ce ptp: prevent possible ABBA deadlock in ptp_clock_freerun()
545a0b84c678 netfilter: xt_nfacct: don't assume acct name is null-terminated
f472ea3cce90 netvsc: transfer lower device max tso size
a9a96c581e88 net: usb: smsc95xx: add support for ethtool pause parameters
5be49dfee611 selftests: drv-net: rss_api: context create and delete tests
4ee6aab279d6 ethtool: rss: support removing contexts via Netlink
8bd3be3a6690 ethtool: rss: support creating contexts via Netlink
f6a1fa5e4327 ethtool: move ethtool_rxfh_ctx_alloc() to common code

$ git log --oneline ff3fbcdd472453190d4f37fe1d1e69e50cce7612 | head
ff3fbcdd4724 selftests: tc: Add generic erspan_opts matching support for tc-flower
dd500e4aecf2 net: usb: Remove duplicate assignments for net->pcpu_stat_type
4701ee5044fb be2net: Use correct byte order and format string for TCP seq and ack_seq
190ccb817637 net: bcmasp: Add support for re-starting auto-negotiation
e07ba344a465 Merge branch 'net-maintain-netif-vs-dev-prefix-semantics'
88d3cec28274 net: s/dev_close_many/netif_close_many/
5d4d84618e1a net: s/dev_set_threaded/netif_set_threaded/
93893a57efd4 net: s/dev_get_flags/netif_get_flags/
303a8487a657 net: s/__dev_set_mtu/__netif_set_mtu/
0413a34ef678 net: s/dev_pre_changeaddr_notify/netif_pre_changeaddr_notify/

But yours is indeed the one that is in net-next. Not sure how that
happened.

> [...]
> 
> > vrf_ip6_input_dst unconditionally sets skb dst entry, add a call to
> > skb_dst_drop to drop any existing entry.
> > 
> > Cc: David Ahern <dsahern@kernel.org>
> > Fixes: 9ff74384600a ("net: vrf: Handle ipv6 multicast and link-local addresses")
> > Signed-off-by: Stanislav Fomichev <sdf@fomichev.me>
> 
> Reviewed-by: Ido Schimmel <idosch@nvidia.com>

Thanks!

Re: [PATCH net v2] vrf: Drop existing dst reference in vrf_ip6_input_dst

[Jakub Kicinski]

On Thu, 24 Jul 2025 11:06:48 -0700 Stanislav Fomichev wrote:
> Not sure how that happened.

FWIW looks like your hash is coming form the NIPA testing tree:

commit 8fba0fc7b3dec1abc4dd57449d578d8c2682d60e
Author: Li Shuang <shuali@redhat.com>
Date:   Fri Jul 18 22:16:12 2025 +0800

    selftests: tc: Add generic erspan_opts matching support for tc-flower
    
    Add test cases to tc_flower.sh to validate generic matching on ERSPAN
    options. Both ERSPAN Type II and Type III are covered.
    
    Also add check_tc_erspan_support() to verify whether tc supports
    erspan_opts.
    
    Signed-off-by: Li Shuang <shuali@redhat.com>
    Signed-off-by: NipaLocal <nipa@local>
-- 
pw-bot: cr