-
Welcome
-
Subscribe to
IOMMU and VFIO track
Friday, October 17, 2014 from 1:00 – 3:45pm
Room 28
Room 28
We are getting more and more IOMMU drivers in the Linux kernel. For those we often have to close the gap between making them look similar through the IOMMU-API while supporting their unique features too. On this Microconference we will discuss the challenges and their solutions to improve the IOMMU code base already in the kernel as well as how we approach new challenges for future features like IO page faulting.
Microconference Leaders
Sessions for this track
* Consolidating IOMMU Drivers
Lots of code in the current IOMMU drivers can be consolidated into core code. This discussion presents some ideas on this.
(slides)
|
IOMMU and VFIO |
Joerg Roedel | |
* Exposing a virtual IOMMU interface to KVM guests
Discussion around exposing a virtual IOMMU interface to a KVM guest using VFIO/KVM ioctls
(slides)
|
IOMMU and VFIO |
Will Deacon | |
* Handling device identity mappings in the IOMMU API
Some devices request identity mapping of ranges via firmware interfaces, which seems contrary to device use via the IOMMU API. Can we support these conflicting requirements?
(slides)
|
IOMMU and VFIO |
Alex Williamson | |
* IOMMU Fault Handling
The IOMMU subsystem currently lacks fault propagation out to drivers or other users of the interface. We'll discuss work to be done in this area.
|
IOMMU and VFIO |
David Woodhouse | |
* IOMMU Page Faulting an Linux MM Integration
Discussion about how to integrate upcoming and existing demand paging IOMMUs into the IOMMU-API
(slides)
|
IOMMU and VFIO |
Joerg Roedel | |
* VFIO and non-DMA devices
This session proposes changes to vfio to remove a current limitation-- allowing the passthrough of non-DMA devices to user space.
|
IOMMU and VFIO |
Stuart Yoder, Bharat Bhushan |
Proposals for this track
* Refactor interrupt remapping interfaces to support hierarchy irqdomain
We are trying to refactor x86 interrupt architecture with hierarchy irqdomain, so changes to current interrupt remapping interfaces are needed to support the refactor.
|
IOMMU and VFIO | 09/01/2014 |
Jiang Liu | ||
* Using IOMMUs for more than just translation
IOMMUs can offer services other than address translation. Do we need to support this in Linux? What are the use-cases?
|
IOMMU and VFIO | 08/26/2014 |
Will Deacon |