Re: [PATCH 2/2] virtio-blk: set NUMA affinity for a tagset

From: Max Gurtovoy <mgurtovoy@nvidia.com>
To: Stefan Hajnoczi <stefanha@redhat.com>
Cc: <mst@redhat.com>, <virtualization@lists.linux-foundation.org>,
	<kvm@vger.kernel.org>, <oren@nvidia.com>, <nitzanc@nvidia.com>,
	<israelr@nvidia.com>, <hch@infradead.org>,
	<linux-block@vger.kernel.org>, <axboe@kernel.dk>
Subject: Re: [PATCH 2/2] virtio-blk: set NUMA affinity for a tagset
Date: Mon, 27 Sep 2021 20:39:30 +0300	[thread overview]
Message-ID: <21295187-41c4-5fb6-21c3-28004eb7c5d8@nvidia.com> (raw)
In-Reply-To: <YVF8RBZSaJs9BScd@stefanha-x1.localdomain>

On 9/27/2021 11:09 AM, Stefan Hajnoczi wrote:
> On Sun, Sep 26, 2021 at 05:55:18PM +0300, Max Gurtovoy wrote:
>> To optimize performance, set the affinity of the block device tagset
>> according to the virtio device affinity.
>>
>> Signed-off-by: Max Gurtovoy <mgurtovoy@nvidia.com>
>> ---
>>   drivers/block/virtio_blk.c | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
>> index 9b3bd083b411..1c68c3e0ebf9 100644
>> --- a/drivers/block/virtio_blk.c
>> +++ b/drivers/block/virtio_blk.c
>> @@ -774,7 +774,7 @@ static int virtblk_probe(struct virtio_device *vdev)
>>   	memset(&vblk->tag_set, 0, sizeof(vblk->tag_set));
>>   	vblk->tag_set.ops = &virtio_mq_ops;
>>   	vblk->tag_set.queue_depth = queue_depth;
>> -	vblk->tag_set.numa_node = NUMA_NO_NODE;
>> +	vblk->tag_set.numa_node = virtio_dev_to_node(vdev);
>>   	vblk->tag_set.flags = BLK_MQ_F_SHOULD_MERGE;
>>   	vblk->tag_set.cmd_size =
>>   		sizeof(struct virtblk_req) +
> I implemented NUMA affinity in the past and could not demonstrate a
> performance improvement:
> https://lists.linuxfoundation.org/pipermail/virtualization/2020-June/048248.html
>
> The pathological case is when a guest with vNUMA has the virtio-blk-pci
> device on the "wrong" host NUMA node. Then memory accesses should cross
> NUMA nodes. Still, it didn't seem to matter.

I think the reason you didn't see any improvement is since you didn't 
use the right device for the node query. See my patch 1/2.

I can try integrating these patches in my series and fix it.

BTW, we might not see a big improvement because of other bottlenecks but 
this is known perf optimization we use often in block storage drivers.

>
> Please share your benchmark results. If you haven't collected data yet
> you could even combine our patches to see if it helps. Thanks!
>
> Stefan