Re: [Xen-devel] Stub domain crash on Xen v4.6.1

[Fanny Dwargee <fdwargee6@xxxxxxxxx>]

Wei, thanks for your help.

After following your advice I added to the DomU config the 'device_model_version = "qemu-xen-traditional"' (and removing 'device_model_stubdomain_override'), the DomU started successfully as without stub domain:

(d18) Press F12 for boot menu.

(d18)

(d18) Booting from Hard Disk...

(XEN) irq.c:275: Dom18 PCI link 0 changed 5 -> 0

(XEN) irq.c:275: Dom18 PCI link 1 changed 10 -> 0

(XEN) irq.c:275: Dom18 PCI link 2 changed 11 -> 0

(XEN) irq.c:275: Dom18 PCI link 3 changed 5 -> 0

Just for the sake of remember this is the log of the DomU with the initial configuration (no 'device_model_version' and no 'device_model_stubdomain_override'):

(d19) Press F12 for boot menu.

(d19)

(d19) Searching bootorder for: HALT

(d19) drive 0x000f64f0: PCHS=16383/16/63 translation=lba LCHS=1024/255/63 s=20971520

(d19) Space available for UMB: ca800-ef000, f5f10-f64f0

(d19) Returned 258048 bytes of ZoneHigh

(d19) e820 map has 6 items:

(d19)   0: 0000000000000000 - 000000000009fc00 = 1 RAM

(d19)   1: 000000000009fc00 - 00000000000a0000 = 2 RESERVED

(d19)   2: 00000000000f0000 - 0000000000100000 = 2 RESERVED

(d19)   3: 0000000000100000 - 000000007efff000 = 1 RAM

(d19)   4: 000000007efff000 - 000000007f000000 = 2 RESERVED

(d19)   5: 00000000fc000000 - 0000000100000000 = 2 RESERVED

(d19) enter handle_19:

(d19)   NULL

(d19) Booting from Hard Disk...

(d19) Booting from 0000:7c00

(XEN) irq.c:275: Dom19 PCI link 0 changed 5 -> 0

(XEN) irq.c:275: Dom19 PCI link 1 changed 10 -> 0

(XEN) irq.c:275: Dom19 PCI link 2 changed 11 -> 0

(XEN) irq.c:275: Dom19 PCI link 3 changed 5 -> 0

...and this is the log with 'device_model_stubdomain_override = 1' (the DomU hangs starting Windows7):

(d20) Press F12 for boot menu.

(d20)

(d20) Booting from Hard Disk...

(XEN) irq.c:275: Dom20 PCI link 0 changed 5 -> 0

(XEN) irq.c:275: Dom20 PCI link 1 changed 10 -> 0

(XEN) irq.c:275: Dom20 PCI link 2 changed 11 -> 0

(XEN) irq.c:275: Dom20 PCI link 3 changed 5 -> 0

(XEN) grant_table.c:525:d0v0 Bad flags (0) or dom (0). (expected dom 0)

(d21) read error -1 on /local/domain/21/device/vbd/768 at offset 0, num bytes 512

[...]

     [several logs with the same 2 previous lines, snipped for brevity]

[...]

(XEN) grant_table.c:525:d0v0 Bad flags (0) or dom (0). (expected dom 0)

(d21) read error -1 on /local/domain/21/device/vbd/768 at offset 0, num bytes 512

(d21) close(11)

As you can see the error message appears only with the stub domain config.

Regards,

Fanny

On Mon, 11 Apr 2016 at 12:06:38 +0100, Wei Liu wrote:

>

> Sorry for the late reply.

> 

> In the case successfully booting DomU, do you see similar error like

> this in xl dmesg?

> 

> (d21) read error -1 on /local/domain/21/device/vbd/768 at offset 0, num bytes 

> 512

> (XEN) grant_table.c:525:d0v0 Bad flags (0) or dom (0). (expected dom 0)

> (d21) read error -1 on /local/domain/21/device/vbd/768 at offset 0, num bytes 

> 512

> (XEN) grant_table.c:525:d0v0 Bad flags (0) or dom (0). (expected dom 0)

> 

> Another thing to try is to add to your config without

> device_model_override=xxx,

> 

>   device_model_version="qemu-xen-traditional"

> 

> and see if it still works.

> 

> Here is why: the default device model is upstream qemu, but when you

> need to use stubdom, the device model is another version of qemu

> (qemu-trad). The newly added config option forces Xen to use the same

> qemu that would be used when stubdom is in use. It's useful to figure

> out if qemu-trad is not functioning.

>

>

>

> Wei.

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