XSAVE consistency problem for RHEL 8.x VM on ESXi 7.0 with a Skylake CPU

  Symptoms

Booting RHEL 8.x VM on ESXi 7.0 with a Skylake CPU, XSAVE consistency problem error is reported in the dmesg log of the VM.

[ 0.000000] ------------[ cut here ]------------
[ 0.000000] XSAVE consistency problem, dumping leaves
[ 0.000000] WARNING: CPU: 0 PID: 0 at arch/x86/kernel/fpu/xstate.c:614 do_extra_xstate_size_checks+0x2f3/0x3de
[ 0.000000] Modules linked in:
[ 0.000000] CPU: 0 PID: 0 Comm: swapper Not tainted 4.18.0-193.el8.x86_64 #1
[ 0.000000] RIP: 0010:do_extra_xstate_size_checks+0x2f3/0x3de
[ 0.000000] Code: ff 44 3b 25 e5 93 b8 01 0f 84 d9 00 00 00 80 3d d9 96 39 01 00 75 27 48 c7 c7 f0 d1 47 bb c6 05 c9 96 39 01 01 e8 65 3e 08 00 <0f> 0b eb 10 83 3d a0 50 1f 01 00 74 23 e8 4a fc ff ff eb 1c 83 3d
[ 0.000000] RSP: 0000:ffffffffbb603da8 EFLAGS: 00010086 ORIG_RAX: 0000000000000000
[ 0.000000] RAX: 0000000000000000 RBX: 000000000000000a RCX: 0000000000000029
[ 0.000000] RDX: 0000000000000009 RSI: ffffffffbc01af40 RDI: 0000000000000046
[ 0.000000] RBP: 0000000000000008 R08: 657661656c20676e R09: 73657661656c2067
[ 0.000000] R10: 656c626f72702079 R11: 69706d7564202c6d R12: 0000000000000988
[ 0.000000] R13: ffffffffbb603db0 R14: ffffffffbb603dac R15: ffffffffbb603da8
[ 0.000000] FS: 0000000000000000(0000) GS:ffffffffbbd4b000(0000) knlGS:0000000000000000
[ 0.000000] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[ 0.000000] CR2: ffff888000094810 CR3: 0000000091dee000 CR4: 00000000000406a0
[ 0.000000] Call Trace:
[ 0.000000] ? do_clear_cpu_cap+0xf4/0x110
[ 0.000000] ? fpu__init_system_xstate+0x179/0x458
[ 0.000000] ? 0xffffffffba400000
[ 0.000000] ? fpu__init_system+0x208/0x234
[ 0.000000] ? early_init_intel+0x150/0x370
[ 0.000000] ? 0xffffffffba400000
[ 0.000000] ? x86_match_cpu+0x5/0xa0
[ 0.000000] ? early_cpu_init+0x2ae/0x2cb
[ 0.000000] ? setup_arch+0xe0/0x101a
[ 0.000000] ? printk+0x58/0x6f
[ 0.000000] ? start_kernel+0x63/0x55b
[ 0.000000] ? x86_family+0x5/0x20
[ 0.000000] ? load_ucode_bsp+0x42/0x12e
[ 0.000000] ? secondary_startup_64+0xb7/0xc0
[ 0.000000] random: get_random_bytes called from print_oops_end_marker+0x26/0x40 with crng_init=0
[ 0.000000] ---[ end trace 009e620edbd0a5f1 ]---

Cause

You can see this message when the set of xsave features exposed to the virtual machine is a subset of the xsave features support by the host. In such cases, the size of the xsave region enumerated in CPUID may be larger than expected.

Resolution

This is a known issue affecting VMware ESXi 7.0. VMware is working to resolve this issue in a future release.
Note: This message is an advisory from the Linux kernel.