Re: NetBSD DomU (PV) hangs on Xen 4.13.1 / NetBSD 9.1

[Brad Spencer <brad%anduin.eldar.org@localhost>]

Matthias Petermann <mp%petermann-it.de@localhost> writes:

> Hi Brad,
>
> Am 22.12.2020 um 13:36 schrieb Brad Spencer:
>> I have always had to use the block device when presenting devices to any
>> sort of PV type guest.  Try /dev/zvol/dsk/...  When I attempted to use
>> character devices I think I had the same problem that you have.
>
> Thanks for the hint! With this using the block device instead of the raw 
> device:
>
> name = "builder91"
> type="pv"
> kernel = "/netbsd-XEN3_DOMU.gz"
> memory = 16384
> vcpus = 8
> vif = [ 'mac=00:16:3E:74:32:42,bridge=bridge0' ]
> disk = [
>     'phy:/dev/zvol/dsk/tank0/volumes/builder91,hda,rw',
> ]
>
> The kernel seems to go further, but then panic into the debugger. 
> Although it is not deterministic. I had it panic at different 
> stages...one time it already arrived the rc.d scripts starting sshd and 
> postfix (and then panic). So this is only one example:

[snip]

I honestly do not know if this is related to your panic, but you may
need to recompile at least the zfs modules with something like this diff
in place:

--- src/sys/modules/zfs/Makefile.zfsmod.DIST    2019-02-05 04:54:36.000000000 -0500
+++ src/sys/modules/zfs/Makefile.zfsmod 2020-09-10 18:35:34.683493656 -0400
@@ -135,5 +135,6 @@
 CPPFLAGS+=     -I${ZFSDIR}/dist/uts/common
 
 CPPFLAGS+=     -D_PROPLIB_ZFS_CONFLICT
+CPPFLAGS+=     -DMAXPHYS=32768
 
 NEED_ISYSTEM=  # duplicate headers, depends on -isystem to build


This probably won't be needed in -current anywhere, but may very well be
needed on the DOM0 and DOMU in the 9.x generation.  I know that it will
be required on a 9.x DOMU right now if the DOMU uses ZFS (I don't
actually know if DOM0 is effected, as I do not currently have a DOM0
using ZFS).  There was some email conversation going on about this and
at least one PR.



-- 
Brad Spencer - brad%anduin.eldar.org@localhost - KC8VKS - http://anduin.eldar.org