Re: panic: HYPERVISOR_mmu_update failed

[Christoph Egger <Christoph_Egger%gmx.de@localhost>]

Sarton O'Brien wrote:
> On Thu, 31 Jul 2008 09:40:49 pm Christoph Egger wrote:
> > On Thursday 31 July 2008 00:59:01 Sarton O'Brien wrote:
> > > # gdb -w netbsd.gdb /dev/mem
> > > GNU gdb 6.5
> > > Copyright (C) 2006 Free Software Foundation, Inc.
> > > GDB is free software, covered by the GNU General Public License, and you
> > > are welcome to change it and/or distribute copies of it under certain
> > > conditions.
> > > Type "show copying" to see the conditions.
> > > There is absolutely no warranty for GDB.  Type "show warranty" for 
> details.
> > > This GDB was configured as "x86_64--netbsd"...netbsd.gdb: No such file
> > > or directory.
> > >
> > > xpq_flush_queue: 1 entries
> > > 0x000000004324f020: 0x0000000000000225
> > > panic: HYPERVISOR_mmu_update failed
> > >
> > > fatal breakpoint trap in supervisor mode
> > > trap type 1 code 0 rip ffffffff803649f5 cs e030 rflags 246 cr2  54dfa0
> > > cpl 6 rsp ffffa00013afa7e0
> > > Stopped in pid 496.1 (gdb) at   netbsd:breakpoint+0x5:  leave
> > > breakpoint() at netbsd:breakpoint+0x5
> > > panic() at netbsd:panic+0x255
> > > xpq_flush_queue() at netbsd:xpq_flush_queue+0xaf
> > > pmap_enter_ma() at netbsd:pmap_enter_ma+0x4ec
> > > pmap_enter() at netbsd:pmap_enter+0x5a
> > > mmrw() at netbsd:mmrw+0x1c2
> > > spec_read() at netbsd:spec_read+0x1f6
> > > VOP_READ() at netbsd:VOP_READ+0x2d
> > > vn_read() at netbsd:vn_read+0x9f
> > > dofileread() at netbsd:dofileread+0x7e
> > > sys_read() at netbsd:sys_read+0x72
> > > syscall() at netbsd:syscall+0x98
> > > ds          0xa7f0
> > > es          0x57cc
> > > fs          0xa7f0
> > > gs          0x58a7
> > > rdi         0
> > > rsi         0xd
> > > rbp         0xffffa00013afa7e0
> > > rbx         0xffffa00013afa7f0
> > > rdx         0
> > > rcx         0
> > > rax         0x1
> > > r8          0xffffffff80578240  cpu_info_primary
> > > r9          0x1
> > > r10         0xffffa00013afa700
> > > r11         0xffffffff8037c420  xenconscn_putc
> > > r12         0x100
> > > r13         0xffffffff803f330b  copyright+0x12eab
> > > r14         0x225
> > > r15         0xffffffff80633600  kernel_pmap_store
> > > rip         0xffffffff803649f5  breakpoint+0x5
> > > cs          0xe030
> > > rflags      0x246
> > > rsp         0xffffa00013afa7e0
> > > ss          0xe02b
> > > netbsd:breakpoint+0x5:  leave
> > > db>
> > Some questions:
>
> Sorry, I meant to include uname -a:
>
> NetBSD babylon.internal 4.99.71 NetBSD 4.99.71 (XEN3_DOMU) #1: Fri Aug  1 
> 07:32:33 EST 2008  
> root%spike.internal@localhost:/usr/obj/sys/arch/amd64/compile/XEN3_DOMU amd64
>
> > Is this reproducable?
>
> Yes, just tested on the kernel above. That's +10 EST. The command was a cut 
> and paste from somebody assisting me with another issue I'm fighting with. 
> The end result was an attempt to tip the system over by 
> setting 'uvm_debug_check_rbtree = 1'.
>
> > Are you running a debug or non-debug Xen hypervisor?
>
> non-debug
>
> > Can you provide the dmesg output?
>
> The dmesg from the domu is probably of no use. Do you want either dmesg from 
> dom0 or hypervisor? Should I run a debug hypervisor for more info?

Run a debug hypervisor. You should get some more information from the
hypervisor (output shown on dom0 console and in 'xm dmesg') right before
the hypercall_mmu_update failure.

Christoph