[Xen-users] Instant panic+reboot when starting HVM on PowerEdge R815s

[xenusers <m+xenusers@xxxxxxxxxx>]

Hello list,

I'm hoping someone else has seen and/or solved this.  We have two older R815s that run HVM domUs no problem.  They're running AMD Opteron 6174 (48 core) processors.  We have five newer R815s, they run AMD Opteron 6274 processors.  The new servers have handled some testing with PV guests wonderfully, spinning up dozens and dozens and making them as busy as possible.  Today I've tried to start a first test Windows VM on the new machines, and it's an instant panic+reboot.  The dom0s are Debian 6.0.5. with Xen 4.0.1 (just straight up Debian Squeeze that was freshly installed and updated two days ago).

 (XEN) ----[ Xen-4.0.1  x86_64  debug=n  Not tainted ]----
(XEN) CPU:    2
(XEN) RIP:    e008:[<ffff82c4801af3f4>] svm_fpu_enter+0x74/0xb0
(XEN) RFLAGS: 0000000000010256   CONTEXT: hypervisor
(XEN) rax: 0000000000000003   rbx: ffff83041ac9d000   rcx: 0000000000000000
(XEN) rdx: 0000000000000000   rsi: 0000000000000003   rdi: ffff8300df2d8000
(XEN) rbp: ffff83041ac9d000   rsp: ffff83041ff27758   r8:  ffff83041ff27c58
(XEN) r9:  0000000000000048   r10: 0000000000000022   r11: 0000000000000000
(XEN) r12: 0000000000000011   r13: 0000000000000002   r14: ffff82c48022a7e0
(XEN) r15: 0000000000000000   cr0: 0000000080050033   cr4: 00000000000006f0
(XEN) cr3: 0000000419f19000   cr2: 0000000000000000
(XEN) ds: 0000   es: 0000   fs: 0010   gs: 0010   ss: 0000   cs: e008
(XEN) Xen stack trace from rsp=ffff83041ff27758:
(XEN)    ffff8300df2d8000 ffff82c4801af7bd ffff82c48022a7e0 ffff8300df2d8000
(XEN)    0000000000000010 ffff82c48019e682 ffffffff0c930020 ffff83041ff27c58
(XEN)    0000000000000002 0000000000000000 0000000000000000 0000000000000000
(XEN)    0000000000000002 ffff82c480199c4c 0000000000000022 ffff82c480199d3c
(XEN)    ffff83041ff27c58 ffff82c4801787bf ffffffff08930010 ffff83041ff27c58
(XEN)    0000000000000000 ffff82c480186d03 0000000000000002 ffff83041ff278a0
(XEN)    0000000000000000 0000000000000022 ffff830400000048 0000000000000004
(XEN)    000000000000006e ffff83041ff27b70 0000000000000000 0000000000000002
(XEN)    ffff83041ff278e8 0000000000000002 0000000000000048 0000000000000001
(XEN)    0000000000000028 00000000801e3d07 ffff83041ff279f8 0100830400000000
(XEN)    0000000000000000 0000000000000000 00000000000000c0 0000000000000000
(XEN)    0000000000000000 0000000000000008 0000000000000000 0000000000000000
(XEN)    0000000000000000 0000000000000000 0000000000000000 0000000000000000
(XEN)    0000000000000000 0000000000000000 0000000000000000 0000000000000000
(XEN)    0000000000000000 0000000000000000 0000000000000000 0000000000000000
(XEN)    0000000000000000 0000000000000000 0000000000000000 0000000000000000
(XEN)    0000000000000000 0000000000000004 0000000000000003 0000000000000000
(XEN)    0000000000000046 0000000000000000 0000000000000000 0000000000000000
(XEN)    0000000000000000 0000000000000000 0000000000000000 0000000200000001
(XEN)    0000000000000000 0000000000000000 0000000000000000 0000000000000000
(XEN) Xen call trace:
(XEN)    [<ffff82c4801af3f4>] svm_fpu_enter+0x74/0xb0
(XEN)    [<ffff82c4801af7bd>] svm_update_guest_cr+0x14d/0x160
(XEN)    [<ffff82c48019e682>] hvm_set_cr0+0x172/0x3e0
(XEN)    [<ffff82c480199c4c>] hvmemul_write_cr+0x5c/0xf0
(XEN)    [<ffff82c480199d3c>] hvmemul_read_segment+0xc/0x20
(XEN)    [<ffff82c4801787bf>] get_cpl+0x3f/0x50
(XEN)    [<ffff82c480186d03>] x86_emulate+0xdd63/0xff10
(XEN)    [<ffff82c4801a00d0>] __hvm_copy+0x350/0x3c0
(XEN)    [<ffff82c4801a00d0>] __hvm_copy+0x350/0x3c0
(XEN)    [<ffff82c4801999d3>] hvm_emulate_one+0xd3/0x1d0
(XEN)    [<ffff82c4801996b8>] hvmemul_get_seg_reg+0x58/0x70
(XEN)    [<ffff82c4801999d3>] hvm_emulate_one+0xd3/0x1d0
(XEN)    [<ffff82c4801996b8>] hvmemul_get_seg_reg+0x58/0x70
(XEN)    [<ffff82c4801a3c04>] handle_mmio+0x64/0x1b0
(XEN)    [<ffff82c480170001>] time_calibration_std_rendezvous+0x91/0xc0
(XEN)    [<ffff82c4801b13f5>] svm_vmexit_handler+0x3c5/0x10f0
(XEN)    [<ffff82c4801b0d74>] svm_do_resume+0x74/0xc0
(XEN)    [<ffff82c48014e3a4>] context_switch+0x174/0xf90
(XEN)    [<ffff82c48011c153>] schedule+0x1e3/0x500
(XEN)    [<ffff82c4801ae22c>] pt_update_irq+0x2c/0x210
(XEN)    [<ffff82c4801a4f6c>] hvm_vcpu_has_pending_irq+0x4c/0xb0
(XEN)    [<ffff82c4801aebc1>] svm_intr_assist+0x41/0x150
(XEN)    [<ffff82c48011d77e>] __do_softirq+0x6e/0x90
(XEN)    [<ffff82c4801aea84>] svm_stgi_label+0x8/0x24
(XEN)    
(XEN) 
(XEN) ****************************************
(XEN) Panic on CPU 2:
(XEN) FATAL TRAP: vector = 6 (invalid opcode)
(XEN) ****************************************
(XEN) 
(XEN) Reboot in five seconds...


I tried with a config based on a Windows guest that is running perfectly on the older R815s AMD 6174 processors.  The only things I had to change were name, MAC, disks.  It crashes immediately on the newer R815s.

kernel = "hvmloader"
builder='hvm'
memory = 1024
name = "winhost"
vcpus=1
# Since cpus are 0 and 1 are dedicated to dom0 by the Xen command line
# we added to grub, make domUs run on the other cpus.
cpus="2-63"
vif = [ 'bridge=xenbr0, mac=0A:33:50:10:7B:FE' ]
disk = [ 'phy:/dev/mapper/winhost,hda,w' ]
device_model = 'qemu-dm'
boot="cda"
sdl=0
opengl=1
vnc=1
vncpasswd=''
stdvga=0
serial='pty'
tsc_mode=0
usbdevice='tablet'
localtime=1

Is there anybody out there successfully running Xen HVM domUs on PowerEdge R815s with the AMD 6274 CPUs?  If so, can you share a (sanitized) config that I might try?  I have no idea what these crashes are a result of, but I'd love to have a known good working HVM config to start from.

Regards,
Mark


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users