Re: domU won't start with recent netbsd-10

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

On Sat, Jul 29, 2023 at 07:20:50PM -0400, Greg Troxel wrote:
> I have a box which has been netbsd-10 amd64, xen 4.15, and it has been
> ok for a long time.  I have some local patches to zfs arc memory
> limits -- but zfs is fine.  I have recently update from late June to:
> 
>   NetBSD bar.example.com 10.0_BETA NetBSD 10.0_BETA (XEN3_DOM0) #11: Thu Jul 27 18:57:20 EDT 2023  gdt%foo.example.com@localhost:/usr/obj/gdt-10/amd64/sys/arch/amd64/compile/XEN3_DOM0 amd64
> 
> I made a netbsd-9 amd64 domU in the last month, and it was ok.  I then
> tried nvmm, but it needs new CPUs (says I :-), so I'm back to xen.
> 
> I am setting memory exactly equal to what is in "xl list"; the dom0 has
> 6144M.
> 
> Starting the domU mostly works and then crashes with:
> 
> console:
> 
>   [   1.0000000] NetBSD 9.3_STABLE (XEN3_DOMU) #42: Wed Jun 28 15:38:35 EDT 2023
>   [   1.0000000]  gdt%foo.example.com@localhost:/usr/obj/gdt-9/amd64/sys/arch/amd64/compile/XEN3_DOMU
>   [   1.0000000] total memory = 1916 MB
>   [   1.0000000] avail memory = 1845 MB
>   [   1.0000000] running cgd selftest aes-xts-256 aes-xts-512 done
>   [   1.0000000] mainbus0 (root)
>   [   1.0000000] hypervisor0 at mainbus0: Xen version 4.15.4nb0
>   [   1.0000000] vcpu0 at hypervisor0
>   [   1.0000000] vcpu0: Pentium(R) Dual-Core  CPU      E5700  @ 3.00GHz, id 0x1067a
>   [   1.0000000] vcpu0: package 0, core 1, smt 0
>   [   1.0000000] xenbus0 at hypervisor0: Xen Virtual Bus Interface
>   [   1.0000000] xencons0 at hypervisor0: Xen Virtual Console Driver
> 
> end of xl dmesg:
> 
>   (XEN) Dom0 has maximum 2 VCPUs
>   (XEN) Initial low memory virq threshold set at 0x4000 pages.
>   (XEN) Scrubbing Free RAM in background
>   (XEN) Std. Loglevel: Errors and warnings
>   (XEN) Guest Loglevel: Nothing (Rate-limited: Errors and warnings)
>   (XEN) Xen is relinquishing VGA console.
>   (XEN) *** Serial input to DOM0 (type 'CTRL-a' three times to switch input)
>   (XEN) Freed 648kB init memory
>   (XEN) d1 L1TF-vulnerable L4e 00000000cb1b9000 - Shadowing
>   (XEN) d1 Failed to enable PG_SH_forced: -12
>   (XEN) domain_crash called from common.c:3533
>   (XEN) Domain 1 reported crashed by domain 32767 on cpu#0:
>   (XEN) d2 L1TF-vulnerable L4e 00000000cb192000 - Shadowing
>   (XEN) d2 Failed to enable PG_SH_forced: -12
>   (XEN) domain_crash called from common.c:3533
>   (XEN) Domain 2 reported crashed by domain 32767 on cpu#0:
>   (XEN) d3 L1TF-vulnerable L4e 00000000cb192000 - Shadowing
>   (XEN) d3 Failed to enable PG_SH_forced: -12
>   (XEN) domain_crash called from common.c:3533
>   (XEN) Domain 3 reported crashed by domain 32767 on cpu#0:
> 
> Is there some new over-eager bug mitigation?

I don't know whar PG_SH_forced is.
The last tested build (2023-07-26 05:00 UTC) is running fine:
https://www-soc.lip6.fr/~bouyer/NetBSD-tests/xen/netbsd-10/

maybe some BIOS setting ?


-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--