On 2012-06-10 19:02, Alex Williamson wrote: > msix_init has very little configurability as to how it lays out > MSI/X for a device. It claims to resize BARs, but doesn't > actually do this anymore. This patch allows MSI/X to be fully > specified, which is necessary both for emulated devices trying > to match the physical layout of a hardware device as well as for > any kind of device assignment. > > The original intent of msix_init seems to have been to allow > completely virtual devices so enable MSI/X without knowing > anything about it. A sort of "here's a BAR and a device, add > MSI/X to it". We've already dropped the resize support with > the memory API, and I think it makes sense to tune the interface > to something that still makes it easy for virtual devices to use, > but allows the full specification for realistically creating > emulated devices using MSI/X. > > Signed-off-by: Alex Williamson <alex.william...@redhat.com> > --- > > This needs more testing, but I wanted to send it out in case someone > else is working on something similar. In order to not modify the > generic capability adding code VFIO needs to tell qemu about ever > capability for a device. The main roadblock to doing that is that > msix_init does not provide sufficient specification of the MSI/X area > to match a physical device. This solves that.
Reminds me that you asked for my version: http://git.kiszka.org/?p=qemu.git;a=shortlog;h=refs/heads/queues/msi specifically the top two commits. WIP, just picked the old commits, result untested. Still need to look into your version. Jan
signature.asc
Description: OpenPGP digital signature
