Re: When upgrading to kernel 6.12.4, I got the following error with rdma_rxe module.

[Zhu Yanjun <yanjun.zhu@linux.dev>]

在 2025/3/16 2:02, Yiyi Hu 写道:
> Hi, (Sorry to repost, Clicked "Reply" instead of "Reply to all")

Thanks for reporting this problem. I have already replied this problem 
in another mail thread.

Best Regards,

Zhu Yanjun

>
> I don't use 'git bisect' for now, Will report back when I have
> enough time to setup a separate host for 'git bisect',
> According to new observations, It seems it's triggered when user tries
> to use the rdma link over the veth pair,
> I see the same issue with different NIC, The network is something like
>   veth-port-a  <--> br-with-vlan-filtering[veth-port-ap, other-phy-ports]
> The error are as almost same follows:
>
> [657667.100894] ------------[ cut here ]------------
> [657667.100909] WARNING: CPU: 0 PID: 738197 at
> drivers/infiniband/sw/rxe/rxe_net.c:357 rxe_skb_tx_dtor+0xbc/0x110
> [rdma_rxe]
> [657667.100952] Modules linked in: rdma_rxe xt_addrtype xt_mark
> xt_comment macvtap macvlan overlay iscsi_target_mod target_core_user
> uio target_core_pscsi target_core_file target_core_iblock
> target_core_mod nvmet_tcp brd xt_tcpudp xt_nat xt_multiport
> xt_conntrack xt_set ip_set bcache pppoe pppox ppp_generic slhc nbd
> zram virtio_vdpa vduse vdpa nvme_rdma nvmet_rdma nvmet nvme_tcp
> nvme_fabrics nvme_keyring msr thermal rt2800usb rt2x00usb rt2800lib
> rt2x00lib mac80211 nf_nat_pptp nf_conntrack_pptp nf_conntrack_tftp
> nf_conntrack_netbios_ns nf_conntrack_broadcast nf_nat_ftp
> nf_conntrack_ftp nct6775 nct6775_core hwmon_vid vhost_net tun vhost
> vhost_iotlb tap k10temp iptable_nat xt_MASQUERADE rpcsec_gss_krb5
> iptable_filter ip_tables x_tables tcp_diag inet_diag wireguard
> curve25519_x86_64 libcurve25519_generic libchacha20poly1305
> chacha_x86_64 poly1305_x86_64 ip6_udp_tunnel udp_tunnel libchacha
> nfsv4 nfs netfs veth bridge bonding tls nft_nat nft_masq nft_chain_nat
> nf_nat nft_ct nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 cfg80211
> [657667.101117]  nf_tables rfkill 8021q garp mrp stp llc rpcrdma
> rdma_ucm ib_iser ib_umad rdma_cm ib_ipoib iw_cm ib_cm coretemp
> intel_uncore_frequency intel_uncore_frequency_common ee1004 evdev
> mxm_wmi x86_pkg_temp_thermal kvm_intel kvm rapl intel_cstate i2c_i801
> intel_uncore i2c_mux pcspkr i2c_smbus i915 rtc_cmos intel_gtt
> i2c_algo_bit drm_buddy video drm_display_helper intel_pmc_core wmi ttm
> intel_vsec cec pmt_telemetry drm_kms_helper pmt_class button acpi_pad
> mlx4_ib ib_uverbs ib_core nfsd sch_fq_codel auth_rpcgss nfs_acl lockd
> grace sunrpc drm fuse loop dm_snapshot raid0 dm_cache_smq dm_cache
> dm_persistent_data dm_bio_prison dm_bufio mlx4_en sd_mod
> crct10dif_pclmul crc32_pclmul crc32c_intel ghash_clmulni_intel
> sha512_ssse3 sha256_ssse3 sha1_ssse3 nvme ahci ixgbe libahci mdio
> mdio_devres nvme_core libphy e1000e hwmon aesni_intel crypto_simd
> xhci_pci ptp i2c_core nvme_auth libata mlx4_core xhci_hcd dm_mirror
> dm_region_hash dm_log be2iscsi iscsi_tcp libiscsi_tcp libiscsi
> scsi_transport_iscsi dm_multipath dm_mod efivarfs
> [657667.101341]  dmi_sysfs
> [657667.101392] CPU: 0 UID: 0 PID: 738197 Comm: kworker/u9:2 Tainted:
> G        W I        6.12.18-gentoo #2
> [657667.101409] Tainted: [W]=WARN, [I]=FIRMWARE_WORKAROUND
> [657667.101415] Hardware name: To Be Filled By O.E.M. To Be Filled By
> O.E.M./B150A-X1/Hyper, BIOS P7.20 11/18/2016
> [657667.101425] RIP: 0010:rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
> [657667.101449] Code: 00 f0 0f c1 87 80 00 00 00 83 f8 01 74 3a 85 c0
> 7f 96 5b be 03 00 00 00 5d 48 89 d7 41 5c e9 cb 4a 95 df 41 f6 04 24
> 01 75 27 <0f> 0b 5b 5d 41 5c c3 83 e8 01 83 f8 0f 77 a3 49 8d bc 24 d8
> 03 00
> [657667.101464] RSP: 0018:ffffc90000003d60 EFLAGS: 00010246
> [657667.101474] RAX: 0000000000000000 RBX: ffff8881033dde00 RCX:
> 0000000000000000
> [657667.101482] RDX: ffff888233873d00 RSI: 000000000000000e RDI:
> ffff888233873d00
> [657667.101490] RBP: 0000000000000000 R08: 004041d5d8b05506 R09:
> f96183f30388e37b
> [657667.101498] R10: 0000000000000000 R11: ffff888101b70980 R12:
> ffff888101b70000
> [657667.101506] R13: ffff88810704ca40 R14: 00000000fffffea4 R15:
> ffff88810034ca40
> [657667.101514] FS:  0000000000000000(0000) GS:ffff888463c00000(0000)
> knlGS:0000000000000000
> [657667.101523] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [657667.101530] CR2: 00007fb41fadd7e0 CR3: 00000001c52fa005 CR4:
> 00000000003726f0
> [657667.101539] Call Trace:
> [657667.101546]  <IRQ>
> [657667.101551]  ? rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
> [657667.101572]  ? __warn.cold+0x93/0xed
> [657667.101590]  ? rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
> [657667.101609]  ? report_bug+0xe2/0x170
> [657667.101622]  ? handle_bug+0x4f/0x90
> [657667.101632]  ? exc_invalid_op+0x13/0x60
> [657667.101641]  ? asm_exc_invalid_op+0x16/0x20
> [657667.101656]  ? rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
> [657667.101675]  skb_release_head_state+0x20/0x90
> [657667.101685]  napi_consume_skb+0x3e/0x100
> [657667.101696]  ixgbe_poll+0x21c/0x1210 [ixgbe]
> [657667.101740]  __napi_poll+0x25/0x150
> [657667.101753]  net_rx_action+0x301/0x370
> [657667.101767]  ? wakeup_preempt+0x2f/0x60
> [657667.101775]  ? __raise_softirq_irqoff+0x16/0x70
> [657667.101787]  ? ixgbe_msix_clean_rings+0x39/0x40 [ixgbe]
> [657667.101812]  handle_softirqs+0xf1/0x2b0
> [657667.101825]  irq_exit_rcu+0x74/0xd0
> [657667.101835]  common_interrupt+0xb8/0xd0
> [657667.101846]  </IRQ>
> [657667.101850]  <TASK>
> [657667.101855]  asm_common_interrupt+0x22/0x40
> [657667.101867] RIP: 0010:finish_task_switch.isra.0+0xdc/0x2e0
> [657667.101881] Code: 01 00 00 f0 41 ff 4d 00 0f 84 60 01 00 00 81 7d
> d4 80 00 00 00 0f 84 60 01 00 00 48 83 c4 08 5b 41 5c 41 5d 41 5e 41
> 5f 5d c3 <49> 8b 9e 00 03 00 00 48 85 db 74 bb 48 8b 43 10 48 89 df 65
> 8b 35
> [657667.101895] RSP: 0018:ffffc9000607fda8 EFLAGS: 00000282
> [657667.101903] RAX: 0000000080000001 RBX: ffff888463c33940 RCX:
> 0000000000000000
> [657667.101910] RDX: 0000000000000002 RSI: ffffffff8214864a RDI:
> 00000000ffffffff
> [657667.101918] RBP: ffffc9000607fdd8 R08: 00000000000162ad R09:
> ffff8881bbef2dc0
> [657667.101925] R10: ffff888103b9b510 R11: 0000000000000000 R12:
> ffff8881bbef2dc0
> [657667.101932] R13: 0000000000000000 R14: ffff888233873d00 R15:
> 0000000000000000
> [657667.101944]  ? finish_task_switch.isra.0+0xa0/0x2e0
> [657667.101957]  __schedule+0x3dd/0x14d0
> [657667.101966]  ? flush_delayed_work+0x40/0x40
> [657667.101982]  schedule+0x26/0xf0
> [657667.101989]  worker_thread+0x209/0x430
> [657667.102002]  ? _raw_spin_lock_irqsave+0x17/0x40
> [657667.102014]  ? flush_delayed_work+0x40/0x40
> [657667.102026]  kthread+0xda/0x110
> [657667.102036]  ? kthread_park+0x80/0x80
> [657667.102046]  ret_from_fork+0x2d/0x50
> [657667.102054]  ? kthread_park+0x80/0x80
> [657667.102064]  ret_from_fork_asm+0x11/0x20
> [657667.102079]  </TASK>
> [657667.102083] ---[ end trace 0000000000000000 ]---
>
> On Thu, Dec 26, 2024 at 7:50 AM Zhu Yanjun <yanjun.zhu@linux.dev> wrote:
>> 在 2024/12/25 21:04, Yiyi Hu 写道:
>>> Hi, I'm using 6.6.x kernel, after I upgraded to kernel 6.12.x for my
>>> home server, I got the following error. Which causes soft rdma doesn't
>>> work for 6.12.x with my setup.
>> I read this bug descriptions for several times. To be honest, I am also
>> confused with this problem.
>>
>> It seems that with 6.6.x kernel, it can work well. And with 6.12.x, it
>> can not work.
>>
>> So maybe "git bisect" is a good tool to find the commit that causes this
>> problems.
>>
>> Zhu Yanjun
>>
>>> The kernel warning is:
>>>
>>> # [  134.041117] ------------[ cut here ]------------
>>> # [  134.041120] WARNING: CPU: 9 PID: 0 at
>>> drivers/infiniband/sw/rxe/rxe_net.c:357 rxe_skb_tx_dtor+0xbc/0x110
>>> [rdma_rxe]
>>> # [  134.041128] Modules linked in: dm_cache_smq dm_cache
>>> dm_persistent_data dm_bio_prison 8021q garp mrp macvtap macvlan veth
>>> dummy bridge stp llc nft_ct nf_tables sch_fq_codel virtio_vdpa vduse
>>> vdpa rdma_rxe ip6_udp_tunnel udp_tunnel nvme_rdma nvmet_rdma raid456
>>> async_raid6_recov async_memcpy async_pq async_xor async_tx xor
>>> raid6_pq raid1 raid0 bcache nvmet nvme_auth dm_writecache msr thermal
>>> rpcrdma rdma_ucm ib_iser ib_umad rdma_cm ib_ipoib iw_cm ib_cm
>>> rt2800usb rt2x00usb rt2800lib rt2x00lib ee1004 wmi_bmof mxm_wmi evdev
>>> mac80211 snd_hda_codec_realtek snd_hda_codec_generic
>>> snd_hda_scodec_component cfg80211 rfkill snd_hda_intel
>>> snd_intel_dspcfg nf_conntrack_tftp snd_intel_sdw_acpi
>>> nf_conntrack_netbios_ns nf_conntrack_broadcast nf_nat_ftp
>>> nf_conntrack_ftp nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4
>>> snd_hda_codec nct6775 snd_hda_core nct6775_core snd_hwdep hwmon_vid
>>> vhost_net snd_pcm mlx4_ib tun kvm_amd snd_timer vhost gpio_amdpt
>>> ib_uverbs vhost_iotlb i2c_piix4 snd tap rapl pcspkr soundcore
>>> i2c_smbus rtc_cmos wmi
>>> # [  134.041242]  button gpio_generic ib_core kvm k10temp drm fuse
>>> loop dmi_sysfs dm_snapshot dm_bufio mlx4_en sd_mod nvme_tcp
>>> nvme_fabrics nvme ixgbe crct10dif_pclmul mdio crc32_pclmul mdio_devres
>>> crc32c_intel igb ghash_clmulni_intel i2c_algo_bit sha512_ssse3 libphy
>>> nvme_core sha256_ssse3 ptp sha1_ssse3 hwmon xhci_pci aesni_intel ahci
>>> mlx4_core crypto_simd libahci xhci_hcd libata i2c_core sunrpc
>>> dm_mirror dm_region_hash dm_log be2iscsi iscsi_tcp libiscsi_tcp
>>> libiscsi scsi_transport_iscsi dm_multipath dm_mod efivarfs
>>> # [  134.041324] CPU: 9 UID: 0 PID: 0 Comm: swapper/9 Tainted: G
>>>    W          6.12.4-gentoo #2
>>> # [  134.041329] Tainted: [W]=WARN
>>> # [  134.041331] Hardware name: To Be Filled By O.E.M. X370 Killer
>>> SLI/X370 Killer SLI, BIOS P10.31 08/23/2024
>>> # [  134.041335] RIP: 0010:rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
>>> # [  134.041342] Code: 00 f0 0f c1 87 80 00 00 00 83 f8 01 74 39 85 c0
>>> 7f 96 5b 5d be 03 00 00 00 48 89 d7 41 5c e9 7b 43 39 e0 41 f6 04 24
>>> 01 75 26 <0f> 0b 5b 5d 41 5c c3 ff c8 83 f8 0f 77 a4 49 8d bc 24 d8 03
>>> 00 00
>>> # [  134.041347] RSP: 0018:ffffc90000338d20 EFLAGS: 00010246
>>> # [  134.041352] RAX: 0000000000000000 RBX: ffff888101b38200 RCX:
>>> 0000000000000000
>>> # [  134.041355] RDX: ffff88810092de80 RSI: 000000000000000e RDI:
>>> ffff88810092de80
>>> # [  134.041358] RBP: 0000000000000000 R08: ffff8881253b0000 R09:
>>> 0000000000000001
>>> # [  134.041360] R10: ffff8881251fac00 R11: 0000000000000080 R12:
>>> ffff888125220000
>>> # [  134.041364] R13: 0000000000000056 R14: 0000000000010000 R15:
>>> ffff888101b38200
>>> # [  134.041367] FS:  0000000000000000(0000) GS:ffff889fbee40000(0000)
>>> knlGS:0000000000000000
>>> # [  134.041370] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>>> # [  134.041373] CR2: 000055ec3e726750 CR3: 000000000341b000 CR4:
>>> 0000000000350ef0
>>> # [  134.041376] Call Trace:
>>> # [  134.041379]  <IRQ>
>>> # [  134.041381]  ? rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
>>> # [  134.041388]  ? __warn.cold+0x93/0xed
>>> # [  134.041392]  ? rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
>>> # [  134.041398]  ? report_bug+0xe2/0x170
>>> # [  134.041402]  ? handle_bug+0x4f/0x90
>>> # [  134.041405]  ? exc_invalid_op+0x13/0x60
>>> # [  134.041409]  ? asm_exc_invalid_op+0x16/0x20
>>> # [  134.041413]  ? rxe_skb_tx_dtor+0xbc/0x110 [rdma_rxe]
>>> # [  134.041420]  skb_release_head_state+0x20/0x90
>>> # [  134.041424]  napi_consume_skb+0x3e/0x100
>>> # [  134.041428]  mlx4_en_free_tx_desc+0x11d/0x200 [mlx4_en]
>>> # [  134.041433]  mlx4_en_process_tx_cq+0x182/0x560 [mlx4_en]
>>> # [  134.041439]  mlx4_en_poll_tx_cq+0x28/0x70 [mlx4_en]
>>> # [  134.041443]  __napi_poll+0x25/0x150
>>> # [  134.041447]  net_rx_action+0x300/0x370
>>> # [  134.041452]  ? mlx4_msi_x_interrupt+0xd/0x20 [mlx4_core]
>>> # [  134.041465]  handle_softirqs+0xf0/0x2a0
>>> # [  134.041469]  irq_exit_rcu+0x74/0xd0
>>> # [  134.041473]  common_interrupt+0xb8/0xd0
>>> # [  134.041476]  </IRQ>
>>> # [  134.041478]  <TASK>
>>> # [  134.041480]  asm_common_interrupt+0x22/0x40
>>> # [  134.041484] RIP: 0010:cpuidle_enter_state+0xaf/0x410
>>> # [  134.041488] Code: de 01 00 00 e8 d2 9c 5f ff e8 dd f2 ff ff 49 89
>>> c5 0f 1f 44 00 00 31 ff e8 9e bd 5e ff 45 84 ff 0f 85 b0 01 00 00 fb
>>> 45 85 f6 <0f> 88 88 01 00 00 48 8b 04 24 49 63 ce 4c 89 ea 48 6b f1 68
>>> 48 29
>>> # [  134.041492] RSP: 0018:ffffc9000013be78 EFLAGS: 00000202
>>> # [  134.041496] RAX: ffff889fbee40000 RBX: ffff888104c68400 RCX:
>>> 0000000000000000
>>> # [  134.041500] RDX: 0000001f357196a8 RSI: ffffffff82142fc3 RDI:
>>> ffffffff82146c8f
>>> # [  134.041503] RBP: 0000000000000002 R08: 0000000000000002 R09:
>>> 0000000000000000
>>> # [  134.041506] R10: 0000000000000018 R11: ffff889fbee72144 R12:
>>> ffffffff824dfd40
>>> # [  134.041509] R13: 0000001f357196a8 R14: 0000000000000002 R15:
>>> 0000000000000000
>>> # [  134.041513]  ? cpuidle_enter_state+0xa2/0x410
>>> # [  134.041517]  cpuidle_enter+0x29/0x40
>>> # [  134.041520]  do_idle+0x19a/0x200
>>> # [  134.041524]  cpu_startup_entry+0x25/0x30
>>> # [  134.041527]  start_secondary+0xfe/0x120
>>> # [  134.041530]  common_startup_64+0x13e/0x141
>>> # [  134.041535]  </TASK>
>>> # [  134.041537] ---[ end trace 0000000000000000 ]---
>>>
>>> the system has a "Hewlett-Packard Company InfiniBand FDR/Ethernet
>>> 10Gb/40Gb 2-port 544+FLR-QSFP Adapter"
>>> I created a software bridge named br-fp and bridge 2 ports on this adapter,
>>> I also created a veth pair (venfp <-> venfpbr), venfpbr is connected
>>> to bridge br-fp.
>>> then I do 'rdma link add rxe_venfp type rxe netdev venfp', If I try to
>>> do rdma connect to remote storge, the error occurs.
>>>
>>> if you need other info, Please tell me.
>>>
>>> Thanks.

-- 
Best Regards,
Yanjun.Zhu