linux-azure 4.15 fails to boot on Standard_M416s_v2 in Azure

[Gauthier Jolly]

To reproduce:

 * start a bionic VM in azure:
az vm create --name bionic --resource-group test-bionic --image "Canonical:UbuntuServer:18_04-lts-gen2:latest" --size Standard_M416s_v2 --admin-username ubuntu --ssh-key-value SSH_KEY_PATH

 * "downgrade" the kernel to 4.15 and delete the 5.4 kernel

 * reboot (the machine should fail to boot)

The serial console logs can be found on the azure portal (boot diagnostic needs to be enabled for the VM first).

Logs: https://pastebin.ubuntu.com/p/mhKMdMJCtX/

[Bartlomiej Zolnierkiewicz (bzolnier)]

[SRU Justification]

[Impact]

Current linux-azure-4.15 kernel in bionic fails to boot Standard_M416s_v2 VM because of x2apic being disabled:

[ 0.000000] x2apic: IRQ remapping doesn't support X2APIC mode
[ 0.000000] unchecked MSR access error: WRMSR to 0x1b (tried to write 0x00000000fee00100) at rIP: 0xffffffff9325d1f8 (native_write_msr+0x8/0x30)
[ 0.000000] Call Trace:
[ 0.000000] __x2apic_disable.part.5+0x49/0x80
[ 0.000000] enable_IR_x2apic+0x123/0x18c
[ 0.000000] default_setup_apic_routing+0x16/0x73
[ 0.000000] apic_intr_mode_init+0x84/0x91
[ 0.000000] x86_late_time_init+0x24/0x2b
[ 0.000000] start_kernel+0x444/0x505
[ 0.000000] x86_64_start_reservations+0x24/0x26
[ 0.000000] x86_64_start_kernel+0x74/0x77
[ 0.000000] secondary_startup_64+0xa5/0xb0
[ 0.000000] x2apic disabled
[ 0.000000] Switched APIC routing to physical flat.

[Test Plan]

Run updated kernel (with x2apic support backported from linux-azure-5.4 kernel).

It should boot fine and display:

[ 0.000000] Setting APIC routing to physical x2apic.

[Where problems could occur]

Potentially x2apic may now be used on other VM instances.

[Other Info]

None.

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This bug is awaiting verification that the linux-azure-4.15/4.15.0-1130.143 kernel in -proposed solves the problem. Please test the kernel and update this bug with the results. If the problem is solved, change the tag 'verification-needed-bionic' to 'verification-done-bionic'. If the problem still exists, change the tag 'verification-needed-bionic' to 'verification-failed-bionic'.

If verification is not done by 5 working days from today, this fix will be dropped from the source code, and this bug will be closed.

See https://wiki.ubuntu.com/Testing/EnableProposed for documentation how to enable and use -proposed. Thank you!

[Marcelo Cerri (mhcerri)]

Kernel version 4.15.0-1130.143 is booting fine on M416s_v2 instances.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux-azure-4.15 - 4.15.0-1130.143

---------------
linux-azure-4.15 (4.15.0-1130.143) bionic; urgency=medium

  * bionic/linux-azure-4.15: 4.15.0-1130.143 -proposed tracker (LP: #1955257)

  * linux-azure 4.15 fails to boot on Standard_M416s_v2 in Azure (LP: #1951924)
    - PCI: hv: Replace hv_vp_set with hv_vpset
    - PCI: hv: Refactor hv_irq_unmask() to use cpumask_to_vpset()
    - x86/Hyper-V: Set x2apic destination mode to physical when x2apic is
      available
    - iommu/hyper-v: Add Hyper-V stub IOMMU driver
    - [Config] linux-azure: CONFIG_HYPERV_IOMMU=y

  [ Ubuntu: 4.15.0-167.175 ]

  * bionic/linux: 4.15.0-167.175 -proposed tracker (LP: #1955276)
  * hisi_sas driver may oops in prep_ssp_v3_hw() (LP: #1953386)
    - scsi: hisi_sas: Fix to only call scsi_get_prot_op() for non-NULL scsi_cmnd
  * Bionic update: upstream stable patchset 2021-12-13 (LP: #1954703)
    - xhci: Fix USB 3.1 enumeration issues by increasing roothub power-on-good
      delay
    - binder: use euid from cred instead of using task
    - Input: elantench - fix misreporting trackpoint coordinates
    - Input: i8042 - Add quirk for Fujitsu Lifebook T725
    - libata: fix read log timeout value
    - ocfs2: fix data corruption on truncate
    - mmc: dw_mmc: Dont wait for DRTO on Write RSP error
    - parisc: Fix ptrace check on syscall return
    - tpm: Check for integer overflow in tpm2_map_response_body()
    - media: ite-cir: IR receiver stop working after receive overflow
    - ALSA: ua101: fix division by zero at probe
    - ALSA: 6fire: fix control and bulk message timeouts
    - ALSA: line6: fix control and interrupt message timeouts
    - ALSA: synth: missing check for possible NULL after the call to kstrdup
    - ALSA: timer: Fix use-after-free problem
    - ALSA: timer: Unconditionally unlink slave instances, too
    - x86/irq: Ensure PI wakeup handler is unregistered before module unload
    - cavium: Return negative value when pci_alloc_irq_vectors() fails
    - scsi: qla2xxx: Fix unmap of already freed sgl
    - cavium: Fix return values of the probe function
    - sfc: Don't use netif_info before net_device setup
    - hyperv/vmbus: include linux/bitops.h
    - mmc: winbond: don't build on M68K
    - bpf: Prevent increasing bpf_jit_limit above max
    - xen/netfront: stop tx queues during live migration
    - spi: spl022: fix Microwire full duplex mode
    - watchdog: Fix OMAP watchdog early handling
    - vmxnet3: do not stop tx queues after netif_device_detach()
    - btrfs: fix lost error handling when replaying directory deletes
    - hwmon: (pmbus/lm25066) Add offset coefficients
    - regulator: s5m8767: do not use reset value as DVS voltage if GPIO DVS is
      disabled
    - regulator: dt-bindings: samsung,s5m8767: correct s5m8767,pmic-buck-default-
      dvs-idx property
    - EDAC/sb_edac: Fix top-of-high-memory value for Broadwell/Haswell
    - mwifiex: fix division by zero in fw download path
    - ath6kl: fix division by zero in send path
    - ath6kl: fix control-message timeout
    - ath10k: fix control-message timeout
    - ath10k: fix division by zero in send path
    - PCI: Mark Atheros QCA6174 t...