Re: [PATCH v1] pc: Support coldplugging of virtio-pmem-pci devices on all buses

From: David Hildenbrand <david@redhat.com>
To: Vivek Goyal <vgoyal@redhat.com>
Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>,
	"Daniel P. Berrange" <berrange@redhat.com>,
	Eduardo Habkost <ehabkost@redhat.com>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	qemu-devel@nongnu.org, Igor Mammedov <imammedo@redhat.com>,
	Paolo Bonzini <pbonzini@redhat.com>,
	Jiri Denemark <jdenemar@redhat.com>,
	Richard Henderson <rth@twiddle.net>
Subject: Re: [PATCH v1] pc: Support coldplugging of virtio-pmem-pci devices on all buses
Date: Tue, 26 May 2020 16:43:35 +0200	[thread overview]
Message-ID: <32210631-5933-088c-52e5-9e9eb2d800e3@redhat.com> (raw)
In-Reply-To: <20200526142219.GE108774@redhat.com>

On 26.05.20 16:22, Vivek Goyal wrote:
> On Tue, May 26, 2020 at 03:44:10PM +0200, David Hildenbrand wrote:
>> On 26.05.20 15:28, Vivek Goyal wrote:
>>> On Mon, May 25, 2020 at 10:45:11AM +0200, David Hildenbrand wrote:
>>>> E.g., with "pc-q35-4.2", trying to coldplug a virtio-pmem-pci devices
>>>> results in
>>>>     "virtio-pmem-pci not supported on this bus"
>>>>
>>>> Reasons is, that the bus does not support hotplug and, therefore, does
>>>> not have a hotplug handler. Let's allow coldplugging virtio-pmem devices
>>>> on such buses. The hotplug order is only relevant for virtio-pmem-pci
>>>> when the guest is already alive and the device is visible before
>>>> memory_device_plug() wired up the memory device bits.
>>>>
>>>> Hotplug attempts will still fail with:
>>>>     "Error: Bus 'pcie.0' does not support hotplugging"
>>>>
>>>> Hotunplug attempts will still fail with:
>>>>     "Error: Bus 'pcie.0' does not support hotplugging"
>>>>
>>>> Reported-by: Vivek Goyal <vgoyal@redhat.com>
>>>> Cc: Pankaj Gupta <pankaj.gupta.linux@gmail.com>
>>>> Cc: Igor Mammedov <imammedo@redhat.com>
>>>> Cc: Paolo Bonzini <pbonzini@redhat.com>
>>>> Cc: Richard Henderson <rth@twiddle.net>
>>>> Cc: Eduardo Habkost <ehabkost@redhat.com>
>>>> Cc: "Michael S. Tsirkin" <mst@redhat.com>
>>>> Cc: Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
>>>> Signed-off-by: David Hildenbrand <david@redhat.com>
>>>> ---
>>>>  hw/i386/pc.c | 18 ++++++++++--------
>>>>  1 file changed, 10 insertions(+), 8 deletions(-)
>>>
>>> Thanks for the patch David. I still seem to face a different error though.
>>>
>>> 2020-05-26T13:26:05.720617Z qemu-system-x86_64: -device virtio-pmem-pci,memdev=pmem1,id=nv1: memory devices (e.g. for memory hotplug) are not enabled, please specify the maxmem option
>>>
>>> Following is my domain xml file.
>>>
>>> Vivek
>>
>> Hi Vivek,
>>
>> you have to declare the maxMemory option. Memory devices like
>> virtio-pmem-pci reside in RAM like a pc-dimm or a nvdimm. If your
>> virtio-pmem device will be 4GB, you have to add that to maxMemory.
>>
>>   <memory unit='GiB'>64</memory>
>>   <maxMemory unit='GiB'>68</maxMemory>
>>   <currentMemory unit='GiB'>64</currentMemory>
>>
>> (you might have to add "slots='0'" or "slots='1'" to maxMemory to make
>> libvirt happy)
> 
> Ok, tried that.
> 
> <maxMemory slots='1' unit='KiB'>134217728</maxMemory>
> 
> And now it complains about.
> 
> error: unsupported configuration: At least one numa node has to be configured when enabling memory hotplug
> 
> So ultimately it seems to be wanting me to somehow enable memory hotplug
> to be able to use virtio-pmem?

That's a libvirt error message. Maybe I am confused how libvirt maps
these parameters to QEMU ...

NVDIMMs under libvirt seem to be easy:

https://www.redhat.com/archives/libvir-list/2016-August/msg00055.html

Maybe the issue is that virtio-pmem has not been properly integrated
into libvirt yet:

https://www.redhat.com/archives/libvir-list/2019-August/msg00007.html

And you attempts to force virtio-pmem in via qemu args does not work
properly.

Maybe maxMemory in libvirt does not directly map to the QEMU variant to
define the maximum physical address space reserved also for any memory
devices (DIMMs, NVDIMMs, virtio-pmem, ...). Any libvirt experts that can
help?

@Pankaj, did you ever get it to run with libvirt?

> 
> Thanks
> Vivek

-- 
Thanks,

David / dhildenb