Re: [Xen-users] xen 4.6 & dom0pvh

[Roger Pau Monnà <roger.pau@xxxxxxxxxx>]

Hello,

Please don't top post, it completely screws the conversation log.

El 14/12/15 a les 17.08, ÃliÃs TamÃs ha escrit:
> Hi.
> 
> Ok, I recompiled with xen-staging (4.7-unstable), and made several
> outputs for you.
> 
> I attach them all - standard text files in a tar.xz archive: The server
> is an ASUS P8P67 motherboard with an I5-3550 CPU. I also attach the
> cpuinfo and the lspci -v output, hwinfo and acpidump in a tar.xz file to
> ease bug hunting.
> 
> Xen debug files:
> 
> minicom-normalboot.cap: this is how the system normally boots to prompt
> without the dom0pvh=1 parameter. This is more-or-less a fresh debian 8.2
> system.

You seem to be getting IOMMU errors while booting, which is certainly
not expected:

[...]
(XEN) [VT-D]DMAR:[DMA Write] Request device [0000:09:00.1] fault addr
21e91b000, iommu reg = ffff82c000201000
[...]

And AFAICT 0x21e91b000 is inside a usable memory region according to the
memory map. Anyway, it doesn't seem to stop Xen/Linux from working, so
let's leave it apart.

> 
> minicom-debugenabled.cap: as above, but with the pvh kernel parameter.
> 
> minicom-wotmem.cap: as above, but the boot parameters of transactional
> memory had been removed (I read somewhere that it can cause problems).
> Interestingly, in this case I receive no crash dump, but a silent
> complete reboot without a notice.
> 
> minicom-woextrabootparams.cap: as above, but I removed ALL extra boot
> parameters that can influence startup (of course except the dom0pvh=1).
> This startup has gone further then the Åprevious one, but seemes to me
> rebooting due to the same issue (ACPI MEMORY RESET REG).

Can you pick this one and add "iommu=debug sync_console" to the Xen
command line? The reboot without any kind of message from Xen is
certainly weird.

Roger.


_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users