Hi Lorenzo,
On 09/03/2023 13:25, Lorenzo Bianconi wrote:
> Take into account tx/rx queues reconfiguration setting device
> xdp_features flag. Moreover consider NETIF_F_GRO flag in order to enable
> ndo_xdp_xmit callback.
>
> Fixes: 66c0e13ad236 ("drivers: net: turn on XDP features")
> Signed-off-by: Lorenzo Bianconi <lorenzo@xxxxxxxxxx>
Thank you for the modification.
Unfortunately, 'git bisect' just told me this modification is the origin
of a new WARN when using veth in a netns:
###################### 8< ######################
=============================
WARNING: suspicious RCU usage
6.3.0-rc1-00144-g064d70527aaa #149 Not tainted
-----------------------------
drivers/net/veth.c:1265 suspicious rcu_dereference_check() usage!
other info that might help us debug this:
rcu_scheduler_active = 2, debug_locks = 1
1 lock held by ip/135:
#0: ffffffff8dc4b108 (rtnl_mutex){+.+.}-{3:3}, at: rtnetlink_rcv_msg
(net/core/rtnetlink.c:6172)
stack backtrace:
CPU: 1 PID: 135 Comm: ip Not tainted 6.3.0-rc1-00144-g064d70527aaa #149
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1
04/01/2014
Call Trace:
<TASK>
dump_stack_lvl (lib/dump_stack.c:107)
lockdep_rcu_suspicious (include/linux/context_tracking.h:152)
veth_set_xdp_features (drivers/net/veth.c:1265 (discriminator 9))
veth_newlink (drivers/net/veth.c:1892)
? veth_set_features (drivers/net/veth.c:1774)
? kasan_save_stack (mm/kasan/common.c:47)
? kasan_save_stack (mm/kasan/common.c:46)
? kasan_set_track (mm/kasan/common.c:52)
? alloc_netdev_mqs (include/linux/slab.h:737)
? rcu_read_lock_sched_held (kernel/rcu/update.c:125)
? trace_kmalloc (include/trace/events/kmem.h:54)
? __xdp_rxq_info_reg (net/core/xdp.c:188)
? alloc_netdev_mqs (net/core/dev.c:10657)
? rtnl_create_link (net/core/rtnetlink.c:3312)
rtnl_newlink_create (net/core/rtnetlink.c:3440)
? rtnl_link_get_net_capable.constprop.0 (net/core/rtnetlink.c:3391)
__rtnl_newlink (net/core/rtnetlink.c:3657)
? lock_downgrade (kernel/locking/lockdep.c:5321)
? rtnl_link_unregister (net/core/rtnetlink.c:3487)
rtnl_newlink (net/core/rtnetlink.c:3671)
rtnetlink_rcv_msg (net/core/rtnetlink.c:6174)
? rtnl_link_fill (net/core/rtnetlink.c:6070)
? mark_usage (kernel/locking/lockdep.c:4914)
? mark_usage (kernel/locking/lockdep.c:4914)
netlink_rcv_skb (net/netlink/af_netlink.c:2574)
? rtnl_link_fill (net/core/rtnetlink.c:6070)
? netlink_ack (net/netlink/af_netlink.c:2551)
? lock_acquire (kernel/locking/lockdep.c:467)
? net_generic (include/linux/rcupdate.h:805)
? netlink_deliver_tap (include/linux/rcupdate.h:805)
netlink_unicast (net/netlink/af_netlink.c:1340)
? netlink_attachskb (net/netlink/af_netlink.c:1350)
netlink_sendmsg (net/netlink/af_netlink.c:1942)
? netlink_unicast (net/netlink/af_netlink.c:1861)
? netlink_unicast (net/netlink/af_netlink.c:1861)
sock_sendmsg (net/socket.c:727)
____sys_sendmsg (net/socket.c:2501)
? kernel_sendmsg (net/socket.c:2448)
? __copy_msghdr (net/socket.c:2428)
___sys_sendmsg (net/socket.c:2557)
? mark_usage (kernel/locking/lockdep.c:4914)
? do_recvmmsg (net/socket.c:2544)
? lock_acquire (kernel/locking/lockdep.c:467)
? find_held_lock (kernel/locking/lockdep.c:5159)
? __lock_release (kernel/locking/lockdep.c:5345)
? __might_fault (mm/memory.c:5625)
? lock_downgrade (kernel/locking/lockdep.c:5321)
? __fget_light (include/linux/atomic/atomic-arch-fallback.h:227)
__sys_sendmsg (include/linux/file.h:31)
? __sys_sendmsg_sock (net/socket.c:2572)
? rseq_get_rseq_cs (kernel/rseq.c:275)
? lockdep_hardirqs_on_prepare.part.0 (kernel/locking/lockdep.c:4263)
do_syscall_64 (arch/x86/entry/common.c:50)
entry_SYSCALL_64_after_hwframe (arch/x86/entry/entry_64.S:120)
RIP: 0033:0x7f0d1aadeb17
Code: 0f 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b9 0f 1f 00 f3 0f 1e
fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 2e 00 00 00 0f 05 <48> 3d 00
f0 ff ff 77 51 c3 48 83 ec 28 89 54 24 1c 48 89 74 24 10
All code
========
0: 0f 00 (bad)
2: f7 d8 neg %eax
4: 64 89 02 mov %eax,%fs:(%rdx)
7: 48 c7 c0 ff ff ff ff mov $0xffffffffffffffff,%rax
e: eb b9 jmp 0xffffffffffffffc9
10: 0f 1f 00 nopl (%rax)
13: f3 0f 1e fa endbr64
17: 64 8b 04 25 18 00 00 mov %fs:0x18,%eax
1e: 00
1f: 85 c0 test %eax,%eax
21: 75 10 jne 0x33
23: b8 2e 00 00 00 mov $0x2e,%eax
28: 0f 05 syscall
2a:* 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax
<-- trapping instruction
30: 77 51 ja 0x83
32: c3 ret
33: 48 83 ec 28 sub $0x28,%rsp
37: 89 54 24 1c mov %edx,0x1c(%rsp)
3b: 48 89 74 24 10 mov %rsi,0x10(%rsp)
Code starting with the faulting instruction
===========================================
0: 48 3d 00 f0 ff ff cmp $0xfffffffffffff000,%rax
6: 77 51 ja 0x59
8: c3 ret
9: 48 83 ec 28 sub $0x28,%rsp
d: 89 54 24 1c mov %edx,0x1c(%rsp)
11: 48 89 74 24 10 mov %rsi,0x10(%rsp)
RSP: 002b:00007ffca3305d48 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
RAX: ffffffffffffffda RBX: 00000000640f2bb2 RCX: 00007f0d1aadeb17
RDX: 0000000000000000 RSI: 00007ffca3305db0 RDI: 0000000000000003
RBP: 0000000000000000 R08: 0000000000000001 R09: 00007ffca3304ae0
R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000001
R13: 00007ffca3305eb4 R14: 00007ffca3305e80 R15: 0000561e28bf5020
</TASK>
ip (135) used greatest stack depth: 24544 bytes left
###################### 8< ######################
I can reproduce the issue on the "net" tree with just these 3 commands:
# ip netns add ns1
# ip netns add ns2
# ip link add ns1eth1 netns ns1 type veth peer name ns2eth1 netns ns2
Without this commit fccca038f300 ("veth: take into account device
reconfiguration for xdp_features flag"), I don't have the issue.
For more details about the issue detected by CIs validating our MPTCP
tree, including kconfig and vmlinux if needed:
https://github.com/multipath-tcp/mptcp_net-next/issues/372
Do you mind looking at this regression please? :)
On our side, we will revert this patch in our tree for the moment to
unblock our CI jobs.
Cheers,
Matt
--
Tessares | Belgium | Hybrid Access Solutions
www.tessares.net
