[PATCH 0/2] [focal/linux, focal/linux-azure] Vm crash on Dv5/Ev5

[Tim Gardner]

BugLink: https://bugs.launchpad.net/bugs/1950462

SRU Justification

[Impact]

We are seeing a below crash for Nested VM scenario in Dv5/Ev5.

[ 284.769421] ------------[ cut here ]------------
[ 284.769422] KVM: accessing unsupported EVMCS field 2032
[ 284.769443] WARNING: CPU: 30 PID: 8426 at /build/linux-azure-5.4-YivnXz/linux-azure-5.4-5.4.0/arch/x86/kvm/vmx/evmcs.h:85 evmcs_write64+0x65/0x70 [kvm_intel]
[ 284.769443] Modules linked in: vhost_net vhost tap ipt_REJECT nf_reject_ipv4 xt_tcpudp iptable_filter xt_MASQUERADE iptable_nat nf_nat bridge stp llc xt_owner xt_conntrack nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 iptable_security bpfilter udf crc_itu_t nls_iso8859_1 kvm_intel kvm serio_raw hv_balloon joydev sch_fq_codel ib_iser rdma_cm iw_cm ib_cm ib_core iscsi_tcp libiscsi_tcp libiscsi scsi_transport_iscsi ip_tables x_tables autofs4 btrfs zstd_compress raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx xor raid6_pq libcrc32c raid1 raid0 multipath linear hid_generic crct10dif_pclmul hid_hyperv crc32_pclmul ghash_clmulni_intel aesni_intel crypto_simd cryptd hyperv_fb cfbfillrect glue_helper cfbimgblt hid hv_netvsc hv_utils hyperv_keyboard cfbcopyarea
[ 284.769463] CPU: 30 PID: 8426 Comm: qemu-system-x86 Not tainted 5.4.0-1062-azure #65~18.04.1-Ubuntu
[ 284.769464] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 07/22/2021
[ 284.769467] RIP: 0010:evmcs_write64+0x65/0x70 [kvm_intel]
[ 284.769469] Code: c2 f7 d0 21 81 38 03 00 00 5d c3 80 3d 1c 32 03 00 00 75 f5 48 89 fe 48 c7 c7 f8 63 57 c0 c6 05 09 32 03 00 01 e8 eb d1 53 cd <0f> 0b 5d c3 0f 1f 80 00 00 00 00 0f 1f 44 00 00 48 8b 07 80 b8 ea
[ 284.769469] RSP: 0018:ffffb75a03f0fb68 EFLAGS: 00010282
[ 284.769471] RAX: 0000000000000000 RBX: ffff8e126a9e8000 RCX: 0000000000000006
[ 284.769471] RDX: 0000000000000007 RSI: 0000000000000082 RDI: ffff8e12dfb96580
[ 284.769472] RBP: ffffb75a03f0fb68 R08: 000000000000022b R09: 0000000000000004
[ 284.769472] R10: ffffb75a03f0fcf8 R11: 0000000000000001 R12: 000000000000001e
[ 284.769473] R13: fffffe00005fd000 R14: 0000000000000000 R15: 0000000000000000
[ 284.769474] FS: 00007f4bc4c09700(0000) GS:ffff8e12dfb80000(0000) knlGS:0000000000000000
[ 284.769476] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 284.769477] CR2: 00007f3fddb8eba0 CR3: 0000003f69dbe002 CR4: 0000000000372ee0
[ 284.769478] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[ 284.769478] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[ 284.769479] Call Trace:
[ 284.769485] vmx_vcpu_load_vmcs+0x2f9/0x440 [kvm_intel]
[ 284.769488] vmx_vcpu_load+0x47/0x200 [kvm_intel]
[ 284.769493] ? __memcg_kmem_charge+0x87/0x150
[ 284.769495] ? __alloc_pages_nodemask+0x246/0x320
[ 284.769499] vmx_create_vcpu+0x362/0x720 [kvm_intel]
[ 284.769500] ? __get_free_pages+0x11/0x40
[ 284.769504] ? alloc_loaded_vmcs+0xa2/0x120 [kvm_intel]
[ 284.769507] ? vmx_create_vcpu+0x362/0x720 [kvm_intel]
[ 284.769528] kvm_arch_vcpu_create+0x4f/0x70 [kvm]
[ 284.769538] kvm_vm_ioctl+0x2e2/0x980 [kvm]
[ 284.769542] do_vfs_ioctl+0xa9/0x640
[ 284.769545] ? __switch_to_asm+0x40/0x70
[ 284.769546] ? __switch_to_asm+0x34/0x70
[ 284.769547] ? __switch_to_asm+0x40/0x70
[ 284.769548] ? __switch_to_asm+0x34/0x70
[ 284.769550] ? __switch_to_asm+0x40/0x70
[ 284.769551] ? __switch_to_asm+0x34/0x70
[ 284.769552] ? __switch_to_asm+0x40/0x70
[ 284.769553] ? __switch_to_asm+0x34/0x70
[ 284.769554] ? __switch_to_asm+0x40/0x70
[ 284.769555] ksys_ioctl+0x75/0x80
[ 284.769556] ? __switch_to_asm+0x34/0x70
[ 284.769557] __x64_sys_ioctl+0x1a/0x20
[ 284.769559] do_syscall_64+0x5e/0x200
[ 284.769561] entry_SYSCALL_64_after_hwframe+0x44/0xa9
[ 284.769562] RIP: 0033:0x7f4bcf01d317
[ 284.769563] Code: b3 66 90 48 8b 05 71 4b 2d 00 64 c7 00 26 00 00 00 48 c7 c0 ff ff ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 b8 10 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 41 4b 2d 00 f7 d8 64 89 01 48
[ 284.769564] RSP: 002b:00007f4bc4c08888 EFLAGS: 00000246 ORIG_RAX: 0000000000000010
[ 284.769565] RAX: ffffffffffffffda RBX: 000000000000ae41 RCX: 00007f4bcf01d317
[ 284.769566] RDX: 0000000000000000 RSI: 000000000000ae41 RDI: 000000000000000b
[ 284.769566] RBP: 0000000000000000 R08: 00005596f71e0ec0 R09: 00005596f896c170
[ 284.769567] R10: 00005596f77fb8e0 R11: 0000000000000246 R12: 00005596f892ae90
[ 284.769568] R13: 0000000000000000 R14: 00005596f896c170 R15: 00007fffa5dffce0
[ 284.769569] ---[ end trace 481983b25fa8f1f4 ]---
[ 284.795366] set kvm_intel.dump_invalid_vmcs=1 to dump internal KVM state.

[Fix]

55d2eba8e7cd ("jump_label: Fix usage in module __init")
064eedf2c50f ("KVM: VMX: eVMCS: make evmcs_sanitize_exec_ctrls() work again")

[Test Case]

Create a nested VM on an Azure Dv5/Ev5 instance.
Tested by Microsoft and verified to fix the issue.

[Where things could go wrong]

KVM instance creation could fail in other unusual ways.

[Other info]

SF: #00322790