Re: Dom0 ballooning: crash on guest start

[Jean-Yves Migeon <jeanyves.migeon%free.fr@localhost>]

On Fri, 29 Apr 2011 11:49:33 +0200, Christoph Egger wrote:
> On 04/29/11 11:49, Jean-Yves Migeon wrote:
> > On Fri, 29 Apr 2011 11:14:54 +0200, Christoph Egger wrote:
> > > Then when I start a HVM guest the balloon driver crashes:
> > >
> > > balloon0: current reservation: 2039476 pages => target: 2039476 
> > > pages
> > > balloon0: current reservation: 2039476 pages => target: 981172 
> > > pages
> > > uvm_fault(0xffffffff80c6b220, 0xffffffff81400000, 1) -> e
> > > fatal page fault in supervisor mode
> > > trap type 6 code 0 rip ffffffff8054584d cs e030 rflags 10216 cr2
> > > ffffffff81400000 cpl 0 rsp ffffa00057ca8a50
> > > kernel: page fault trap, code=0
> > > Stopped in pid 0.58 (system) at netbsd:pmap_zero_page+0x41: movq 
> > > 0(%rax,%
> > > rbx,8),%rbx
> > > pmap_zero_page() at netbsd:pmap_zero_page+0x41
> > > uvm_pagealloc_strat() at netbsd:uvm_pagealloc_strat+0x22d
> > > bpge_ctor() at netbsd:bpge_ctor+0x22
> >
> > Hmm, a fault while zeroing the page. Race, TLB out of sync, or maybe
> > developer's (my) mistake :o
> >
> > Is this kernel compiled with DIAGNOSTIC?
>
> It is compiled with DEBUG + DIAGNOSTIC + LOCKDEBUG

Okay, I'll try reproducing the issue. In the meanwhile, could you 
trigger the issue again with BALLOONDEBUG set to 1 (in xen/balloon.c), 
and post the last dmesg output before the fault? I'd like to see if it 
happens under high memory pressure or right from the start.

Thanks for testing!

--
Jean-Yves Migeon
jeanyves.migeon%free.fr@localhost