Re: xenmp testing

["Cherry G. Mathew" <cherry.g.mathew%gmail.com@localhost>]

On 19 January 2012 01:35, Manuel Bouyer <bouyer%antioche.eu.org@localhost> 
wrote:
> On Mon, Jan 09, 2012 at 01:18:35PM +0530, Cherry G. Mathew wrote:
>> Hi,
>>
>> I'd like to call for participation and user feedback of xen MP on -current
>>
>> There are a few known bugs (clock drift, diagnostic message spam on
>> console, etc.), but please feel free to file bugs and/or report here.
>
> On a amd64 domU with 4 CPUs I see under high load:
> xpq_flush_queue: 1 entries (0 successful)
> 0x0000000220a58010: 0x000000014d121067
> panic: HYPERVISOR_mmu_update failed, ret: -22
>
> fatal breakpoint trap in supervisor mode
> trap type 1 code 0 rip ffffffff80134585 cs e030 rflags 246 cr2  603000 cpl 6 
> rsp
>  ffffa0004fc4af60
> Stopped in pid 22483.1 (env) at netbsd:breakpoint+0x5:  leave
> breakpoint() at netbsd:breakpoint+0x5
> vpanic() at netbsd:vpanic+0x1f2
> printf_nolog() at netbsd:printf_nolog
> xpq_queue_machphys_update() at netbsd:xpq_queue_machphys_update
> pmap_zero_page() at netbsd:pmap_zero_page+0xd6
> uvm_pagealloc_strat() at netbsd:uvm_pagealloc_strat+0x3d8
> uvmfault_promote() at netbsd:uvmfault_promote+0x138
> uvm_fault_internal() at netbsd:uvm_fault_internal+0xaff
> trap() at netbsd:trap+0x5f5
>
> And on xen console:
> (XEN) mm.c:2424:d1 Bad type (saw 3400000000000001 != exp 7000000000000000) 
> for mfn 20e43d (pfn 13727)

I've been having trouble deciphering what PGT_xxx flags are set via mm.h

> (XEN) mm.c:907:d1 Error getting mfn 20e43d (pfn 13727) from L1 entry 
> 000000020e43d067 for l1e_owner=1, pg_owner=1
>


> I don't know yet if it's related to kern/45830
>

There's a race where a pmap pm_pdir[] can have a lingering entry in
per-cpu L4 pdirs if they have alternate mappings to that pmap. I've
tried to solve it with this:

http://85.88.27.205/tmp/apte.diff

But yet to hear if that fixes the problem.

I'm not sure if these are related. Can it be tested ?

Cheers,
-- 
~Cherry