[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250529005147.GC192517@ziepe.ca>
Date: Wed, 28 May 2025 21:51:47 -0300
From: Jason Gunthorpe <jgg@...pe.ca>
To: Alex Williamson <alex.williamson@...hat.com>
Cc: David Hildenbrand <david@...hat.com>, lizhe.67@...edance.com,
kvm@...r.kernel.org, linux-kernel@...r.kernel.org,
muchun.song@...ux.dev
Subject: Re: [PATCH v3] vfio/type1: optimize vfio_pin_pages_remote() for huge
folio
On Wed, May 28, 2025 at 02:09:41PM -0600, Alex Williamson wrote:
> To be fair to libvirt, we'd really like libvirt to make use of iommufd
> whenever it's available, but without feature parity this would break
> users.
If people ask there should be no issue with making API functionality
discoverable with a query command of some kind. Alot of new stuff is
already discoverable by invoking an ioctl with bogus arguments and
checking for EOPNOTSUPP/ENOTTY.
But most likely P2P will use the same ioctl as memfd so it will not
work that way.
So for now libvirt could assume no P2P support in iommufd. A simple
algorithm would be to look for more than 1 VFIO device. Or add an xml
"disable P2P" which is a useful thing anyhow.
> Zhe, so if you have no dependencies on P2P DMA within your device
> assignment VMs, the options above may be useful or at least a data
> point for comparison of type1 vs IOMMUFD performance. Thanks,
Even if Zhe does have P2P DMA I have a feeling the OOT P2P patch may
be workable <shrug>
Jason
Powered by blists - more mailing lists