On Tue, 2007-01-09 at 23:40 +1100, David Timms wrote:
> {Reposting to fedora-list rather than fedora-test-list, oops}
>
> Hi, I have some machines:
> pc1: dell sc430 , 1GB, raid 1 software, intel x86_64 kernel
> pc2: shuttle xcube, 1GB, raid 1 software, intel i686 kernel
> os1+2: fc6 +updated 2007-01-04, vmware-server-1.0.1
>
> Both machines run two vmware-guests:
> 1: fc6 +updated 2007-01-04, samba server, winbind
> 2: fc6 +updated 2007-01-04, dns, dhcp, openldap.
> All real and guest vms are running kernel-2.6.18-1.2868.fc6 #1 SMP
> After about 48 hours, pc1-vm1 crashed, restarted and still OK. 3 days.
> Today I have found that pc2-vm1 and pc2-vm2 have crashed.
>
> Message left on pc2:vm2 screen summary:
> worker_thread
> kthread
> kthread
> kernel_thread
> =====
> BUG: spinlock lockup on CPU#0, modprobe/13636 (Not tainted)
> dump_trace
> show_trace_log_lvl
> show_trace
> dump_stack
> _raw_spin_lock
> cache_alloc_refill
> kmem_cache_alloc
> get_empty_filp
> __path_lookup_open
> open_name
> do_filp_open
> do_sys_open
> sys_open
> syscall_call
> DWARF2 unwinder stuck at syscall_call
> Leftover inexact backtrace:
> =====
>
> the visible part of the trace with the other vm is different, but
> mentions the
> same spinlock and DWARF messages.
>
> The server and vm guests have quite a few services either disabled or
> removed,
> to reduce overall machine load.
>
> Is this an issue that needs to be fixed/reported to bugzilla.redhat ?
> Or is there another bug report location that would be best.
>
> DaveT.
>
In general, if you didn't enable the vmware services (and as a result,
you guest's kernel is not tainted), yes.
The lock itself seems to be vmware-free.
- Gilboa
