Memory unplug question

[Vasilis Liaskovitis <vasilis.liaskovitis@profitbricks.com>]

Hi,

I am trying to get memory-unplug to work on a kvm host/guest scenario.
This is for an x86_64 kvm Linux guest running 3.1.0 kernel. The host is running a
modified qemu-kvm and seabios to support memory hotplug (see
https://patchwork.kernel.org/patch/1057612/ for details)

I have managed to hotplug memory in 128MB chunks, above the 4GB limit.
E.g. adding a 128MB memory range at 4GB physical offset results in the following
dmesg output in the guest:

[   42.028288] Hotplug Mem Device 
[   42.028580] init_memory_mapping: 0000000100000000-0000000108000000
[   42.028633]  0100000000 - 0108000000 page 2M

And the memory device can be onlined and normally used.
On memory unplug, I always get the followings failure 

[   71.907689] memory offlining 100000 to 108000 failed
[   71.908438] ACPI:memory_hp:Disable memory device

(I have not onlined the memory device in the OS when trying the unplug)

Specifically the callpath is:
acpi_memory_disable_device()
    remove_memory()
        offline_pages()
            check_pages_isolated() returns a negative value.

Is this expected behaviour? Can the pluggable memory pages be moved to a different
memory zone so that they can be later isolated? I 've seen mention of
ZONE_MOVABLE or a retry parameter in old memory-unplug patchsets.

thanks for any suggestions,

- Vasilis

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>