Re: [PATCH V2] vhost: correctly turn on VIRTIO_F_IOMMU_PLATFORM

[Halil Pasic]

On Wed, 26 Feb 2020 06:44:25 -0500
"Michael S. Tsirkin" <address@hidden> wrote:

> On Wed, Feb 26, 2020 at 06:17:34PM +0800, Jason Wang wrote:
> > 
> > On 2020/2/26 下午5:53, Michael S. Tsirkin wrote:
> > > On Wed, Feb 26, 2020 at 05:43:57PM +0800, Jason Wang wrote:
> > > > We turn on device IOTLB via VIRTIO_F_IOMMU_PLATFORM unconditionally on
> > > > platform without IOMMU support. This can lead unnecessary IOTLB
> > > > transactions which will damage the performance.
> > > > 
> > > > Fixing this by check whether the device is backed by IOMMU and disable
> > > > device IOTLB.
> > > > 
> > > > Reported-by: Halil Pasic <address@hidden>
> > > > Fixes: c471ad0e9bd46 ("vhost_net: device IOTLB support")
> > > Well it's just an optimization, isn't it?
> > 
> > 
> > Kind of, or a fix for the performance.
> > 
> 
> Given this wasn't a regression, it's a valuable enhancement
> but Fixes: seems to agressive.

IMHO Fixes is appropriate. Telling vhost-net F_ACCESS_PLATFORM
when when vdev->dma_as != &address_space_memory results in a severe
performance degradation.

Regards,
Halil

[..]