Dom0 - fatal page fault in supervisor mode when root FS is on a slice of disklabel lying on a raidframe device, built up from two GPT wedges

[Matthias Petermann <matthias%petermann-it.de@localhost>]

Hello everybody,

After I described to netbsd-users yesterday[1] my problem with a certain 
configuration of NetBSD with Xen, the thought came to me that the 
port-xen list may be the more suitable place for it.

In the meantime, I have been able to narrow down the problem a bit more 
precisely. My original post is that Xen with NetBSD as Dom0 brings a 
kernel panic (page fault) to one and the same system when the root 
filesystem is inside a disklabel lying on a raidframe mirror, which in 
turn is made up of two components is present as GPT partitions. A 
comparable setup without the raidframe mirror doesn't have this issue.

Various assumptions have not been confirmed, in further attempts I could 
narrow the problem as follows:

1) Works perfectly:
* NetBSD / Xen 8.0-RELEASE Dom0
* Xen 4.11
* a single disk, GPT partitioned
* Root file system within the first GPT partition (dk0)
* BIOSboot (gpt biosboot .....) because the system has no UEFI BIOS, but 
the hard disk requires GPT due to its size (3 TB)

2) Bring the kernel panic:
* NetBSD / Xen 8.0-RELEASE Dom0
* Xen 4.11
* two disks, GPT partitioned
* raidframe mirror raid0 consisting of the first GPT partition of each 
hard disk
* A disklabel within the raid0
* Root file system within the slice raid0a
* BIOSboot (gpt biosboot .....) because the system has no UEFI BIOS, but 
the hard disk requires GPT due to its size (3 TB)

Interesting:
If variant 2) without Xen but with the standard kernel of NetBSD 8.0 / 
amd64 booted, there is no problem.

Does anyone have any idea what could be the cause here? I would like to 
understand how Dom0 accesses the root filesystem or the underlying devices.

Unfortunately I can not serve with a comprehensive log at the moment. My 
serial interface card will be delivered tomorrow, so hopefully I can 
collect some more information and followup accordingly.

Best regards,
Matthias

[1] https://mail-index.netbsd.org/netbsd-users/2018/08/22/msg021304.html

--
Matthias Petermann <matthias%petermann-it.de@localhost> | www.petermann-it.de
GnuPG: 0x5C3E6D75 | 5930 86EF 7965 2BBA 6572  C3D7 7B1D A3C3 5C3E 6D75