Am 19.10.22 um 13:50 schrieb Michael S. Tsirkin:
On Wed, Oct 19, 2022 at 12:59:58PM +0200, Christian Borntraeger wrote:Michael, as a heads-up. I have not looked into any details yet but we do get the following during reboot of a system on s390. It seems to be new with 6.1-rc1 (over 6.0) [ 8.532461] ------------[ cut here ]------------ [ 8.532497] WARNING: CPU: 8 PID: 377 at include/linux/cpumask.h:110 __netif_set_xps_queue+0x3d8/0xca8 [ 8.532507] Modules linked in: sha1_s390(+) sha_common virtio_net(+) net_failover failover pkey zcrypt rng_core autofs4 [ 8.532528] CPU: 8 PID: 377 Comm: systemd-udevd Not tainted 6.1.0-20221018.rc1.git15.0fd5f2557625.300.fc36.s390x+debug #1 [ 8.532533] Hardware name: IBM 8561 T01 701 (KVM/Linux) [ 8.532537] Krnl PSW : 0704e00180000000 00000000b05ec33c (__netif_set_xps_queue+0x3dc/0xca8) [ 8.532546] R:0 T:1 IO:1 EX:1 Key:0 M:1 W:0 P:0 AS:3 CC:2 PM:0 RI:0 EA:3 [ 8.532552] Krnl GPRS: 00000000e7fb8b3f 0000000080000001 00000000b1870700 00000000b0ca1d3c [ 8.532557] 0000000000000100 0000000000000300 000000008b362500 00000000b133ba48 [ 8.532561] 000000000000000c 0000038000000100 000000000000000c 0000000000000070 [ 8.532566] 0000000084cd3200 0000000000000000 00000000b05ec0c2 00000380010b77c8 [ 8.532575] Krnl Code: 00000000b05ec32e: c0e500187331 brasl %r14,00000000b08fa990 00000000b05ec334: a7f4ff0c brc 15,00000000b05ec14c #00000000b05ec338: af000000 mc 0,0 >00000000b05ec33c: ec76fed8007c cgij %r7,0,6,00000000b05ec0ec 00000000b05ec342: e310f0b00004 lg %r1,176(%r15) 00000000b05ec348: ec16ffac007c cgij %r1,0,6,00000000b05ec2a0 00000000b05ec34e: ec680388007c cgij %r6,0,8,00000000b05eca5e 00000000b05ec354: e310f0b80004 lg %r1,184(%r15) [ 8.532600] Call Trace: [ 8.532604] [<00000000b05ec33c>] __netif_set_xps_queue+0x3dc/0xca8 [ 8.532609] ([<00000000b05ec0c2>] __netif_set_xps_queue+0x162/0xca8) [ 8.532614] [<000003ff7fbb81ce>] virtnet_set_affinity+0x1de/0x2a0 [virtio_net] [ 8.532622] [<000003ff7fbbb674>] virtnet_probe+0x4d4/0xc08 [virtio_net] [ 8.532630] [<00000000b04ec4e8>] virtio_dev_probe+0x1e8/0x418 [ 8.532638] [<00000000b05350ea>] really_probe+0xd2/0x480 [ 8.532644] [<00000000b0535648>] driver_probe_device+0x40/0xf0 [ 8.532649] [<00000000b0535fac>] __driver_attach+0x10c/0x208 [ 8.532655] [<00000000b0532542>] bus_for_each_dev+0x82/0xb8 [ 8.532662] [<00000000b053422e>] bus_add_driver+0x1d6/0x260 [ 8.532667] [<00000000b0536a70>] driver_register+0xa8/0x170 [ 8.532672] [<000003ff7fbc8088>] virtio_net_driver_init+0x88/0x1000 [virtio_net] [ 8.532680] [<00000000afb50ab0>] do_one_initcall+0x78/0x388 [ 8.532685] [<00000000afc7b5b8>] do_init_module+0x60/0x248 [ 8.532692] [<00000000afc7ce96>] __do_sys_init_module+0xbe/0xd8 [ 8.532698] [<00000000b09123b2>] __do_syscall+0x1da/0x208 [ 8.532704] [<00000000b0925b12>] system_call+0x82/0xb0 [ 8.532710] 3 locks held by systemd-udevd/377: [ 8.532715] #0: 0000000089af5188 (&dev->mutex){....}-{3:3}, at: __driver_attach+0xfe/0x208 [ 8.532728] #1: 00000000b14668f0 (cpu_hotplug_lock){++++}-{0:0}, at: virtnet_probe+0x4ca/0xc08 [virtio_net] [ 8.532744] #2: 00000000b1509d40 (xps_map_mutex){+.+.}-{3:3}, at: __netif_set_xps_queue+0x88/0xca8 [ 8.532757] Last Breaking-Event-Address: [ 8.532760] [<00000000b05ec0e0>] __netif_set_xps_queue+0x180/0xca8Does this fix it for you? https://lore.kernel.org/r/20221017030947.1295426-1-yury.norov%40gmail.com
Yes, it does. Thanks a lot. Tested-by: Christian Borntraeger <borntraeger@xxxxxxxxxxxxx>
