APPLIED: [PATCH 0/2][impish/jammy linux-azure] Azure: CVM instances warn of a DMA mapping
Tim Gardner
tim.gardner at canonical.com
Thu May 26 14:25:06 UTC 2022
Applied to impish/jammy linux-azure:master-next. Thanks.
-rtg
On 5/25/22 08:37, Tim Gardner wrote:
> BugLink: https://bugs.launchpad.net/bugs/1975717
>
> SRU Justification
>
> [Impact]
>
> When I hot-add a NIC to a Ubuntu 20.04 VM (“5.13.0-1023-azure #27~20.04.1-Ubuntu”), I get the the below call-trace.
> Please include the March-2022 fix: “Drivers: hv: vmbus: Fix initialization of device object in vmbus_device_register()” (https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=3a5469582c241abca22500f36a9cb8e9331969cf)
>
> BTW, this call-trace should be harmless to a non-CVM VM.
>
> [ 364.275105] hv_utils: Heartbeat IC version 3.0
> [ 364.275137] hv_utils: KVP IC version 4.0
> [ 364.275146] hv_utils: Shutdown IC version 3.2
> [ 364.275153] hv_utils: TimeSync IC version 4.0
> [ 365.281376] ------------[ cut here ]------------
> [ 365.281380] WARNING: CPU: 0 PID: 499 at include/linux/dma-mapping.h:555 netvsc_probe+0x3c9/0x3e0 [hv_netvsc]
> [ 365.281392] Modules linked in: udf crc_itu_t iptable_mangle iptable_filter iptable_raw xt_LOG nf_log_syslog bpfilter nls_iso8859_1 dm_multipath scsi_dh_rdac scsi_dh_emc scsi_dh_alua kvm_intel kvm joydev hid_generic crct10dif_pclmul crc32_pclmul ghash_clmulni_intel aesni_intel crypto_simd cryptd serio_raw hv_netvsc pata_acpi hyperv_keyboard hid_hyperv hv_utils hv_balloon hyperv_fb hid sch_fq_codel ipmi_devintf drm ipmi_msghandler msr i2c_core ip_tables x_tables autofs4
> [ 365.281422] CPU: 0 PID: 499 Comm: kworker/0:3 Not tainted 5.13.0-1023-azure #27~20.04.1-Ubuntu
> [ 365.281424] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS 090008 12/07/2018
> [ 365.281427] Workqueue: hv_pri_chan vmbus_add_channel_work
> [ 365.281434] RIP: 0010:netvsc_probe+0x3c9/0x3e0 [hv_netvsc]
> [ 365.281440] Code: 37 c0 4c 89 e6 48 c7 c7 98 f2 37 c0 e8 70 b1 c4 da e9 e9 fc ff ff 49 c7 84 24 88 0b 00 00 00 00 00 00 41 bf f4 ff ff ff eb b0 <0f> 0b e9 19 fe ff ff 41 bf f4 ff ff ff e9 51 ff ff ff 0f 1f 44 00
> [ 365.281442] RSP: 0018:ffff99270075bbd0 EFLAGS: 00010246
> [ 365.281444] RAX: 0000000000000000 RBX: ffff890840b21000 RCX: 0000000000000002
> [ 365.281446] RDX: 0000000000000002 RSI: 0000000000000000 RDI: 0000000000000002
> [ 365.281447] RBP: ffff99270075bbf8 R08: 0000000000000000 R09: 0000000000000000
> [ 365.281448] R10: ffffffff9c74e820 R11: 0000000000000394 R12: ffff890844dc6000
> [ 365.281449] R13: 0000000000000002 R14: 0000301d0800ae28 R15: 000000000000000f
> [ 365.281451] FS: 0000000000000000(0000) GS:ffff8909f7c00000(0000) knlGS:0000000000000000
> [ 365.281453] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [ 365.281454] CR2: 00005619b3504d38 CR3: 00000001091f8005 CR4: 00000000003706f0
> [ 365.281459] Call Trace:
> [ 365.281461] <TASK>
> [ 365.281465] vmbus_probe+0x67/0x80
> [ 365.281468] really_probe+0x1dc/0x440
> [ 365.281472] driver_probe_device+0xf0/0x160
> [ 365.281476] __device_attach_driver+0x79/0xe0
> [ 365.281479] ? driver_allows_async_probing+0x50/0x50
> [ 365.281482] bus_for_each_drv+0x84/0xd0
> [ 365.281485] __device_attach+0xed/0x170
> [ 365.281488] device_initial_probe+0x13/0x20
> [ 365.281491] bus_probe_device+0x8f/0xa0
> [ 365.281494] device_add+0x3f4/0x8e0
> [ 365.281496] ? hrtimer_init+0x2b/0x70
> [ 365.281501] device_register+0x1b/0x20
> [ 365.281503] vmbus_device_register+0x5e/0xe0
> [ 365.281505] vmbus_add_channel_work+0x12d/0x190
> [ 365.281507] process_one_work+0x21a/0x3b0
> [ 365.281511] worker_thread+0x4d/0x3e0
> [ 365.281515] ? process_one_work+0x3b0/0x3b0
> [ 365.281517] kthread+0x12b/0x150
> [ 365.281521] ? set_kthread_struct+0x40/0x40
> [ 365.281523] ret_from_fork+0x22/0x30
> [ 365.281528] </TASK>
> [ 365.281529] ---[ end trace 80a393e06f0ee58d ]---
> [ 365.514764] hv_balloon: Max. dynamic memory size: 8192 MB
> [ 367.937000] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
>
> [Test Case]
>
> Microsoft tested
>
> [Where things could go wrong]
>
> CVM VNIC instances may not initialize correctly.
>
> [Other Info]
>
> SF: #00337569
>
>
>
--
-----------
Tim Gardner
Canonical, Inc
More information about the kernel-team
mailing list