When I first started working in the Xen community, I purchased the David Chisnall book about Xen to learn more. I received a link from the publisher to showcase a chapter from the book. Click here to read Chapter 6, Understanding Device Drivers. For more information on this book you can visit the publishers site here.
Quote from Simon Crosby on this book:
“The Xen hypervisor has become an incredibly strategic resource for the industry, as the focal point of innovation in cross-platform virtualization technology. David’s book will play a key role in helping the Xen community and ecosystem to grow.”
David also has a recent posting on the HURD kernel worth reading for those with an interest in OS history.
(Note: I will be promoting other Xen books over time to ensure that all authors writing Xen books get publicity within the community)
Is this information up to date? I thought ring buffers were going away due to the inefficiencies. I’d gladly buy a Xen book if I didn’t feel much of the information would be too old.
FWIW, icing on the Xen book cake (to me) would be XSM, mini-os, vTPM, shim domains (cool!), and clean descriptions of the existing and proposed IVC methods. yeah, I know, most these either don’t exist yet or barely exist
.
> Is this information up to date? I thought ring buffers were going
> away due to the inefficiencies. I’d gladly buy a Xen book if I
> didn’t feel much of the information would be too old.
Which ring buffers? Ring buffers in general are here to stay in Xen as they’re an efficient way of doing asynchronous shared memory communication.
In some places the drivers are being restructured with respect to ring buffers (e.g. I think the netchannel2 work is using a different ring buffer set up to the original virtual network). In other places the type of data passed by the rings may be changed to improve efficiency. But device driver shared memory rings themselves are a Good Thing and, as far as I’m aware, will stay in the code forever.
Hope that helps,
Cheers,
Mark