#3854 closed enhancement (duplicate)
PCI passthrough
| Reported by: | Stephan.Ritscher | Owned by: | |
|---|---|---|---|
| Component: | other | Version: | VirtualBox 2.2.0 |
| Keywords: | PCI | Cc: | |
| Guest type: | Windows | Host type: | Linux |
Description
PCI passthrough for Linux hosts as in Xen or KVM would be great.
The following scenario is (unfortunately) still quite common: For new hardware often there are no drivers, e.g. since the specifications were not released. So it would be great if one could use the devices in an virtualized Windows, instead. I myself have a Ricoh card reader and only can use it for SD/MMC cards, but not xD or MS/MS Pro. But since it is a PCI device, I cannot use it in VirtualBox, either. But it seems technically possible and would be a great enhancement (I prefer VirtualBox over other virtualization software for other reasons as usability). Thanks Stephan
Change History (4)
comment:1 by , 15 years ago
comment:3 by , 14 years ago
| Resolution: | → duplicate |
|---|---|
| Status: | new → closed |
comment:4 by , 13 years ago
And where is original ticket? #5252 (it looks not quite the same) or anything else ? Sorry for noob questions, but I really need this
I have found two alternative interim workarounds to this issue. I too like the ability of some hosts to offer MMC/MS-Pro/SD/XD slots over USB, and it's frustrating that my company laptop offers these via the PCI-attached Ricoh controller.
a) use shared folders. great if you are able to have a placeholder drive letter present for an un-inserted memory card. Which works on some flavours of Windows but not so far on any linux host I have tried. I don't know what the guest will see if you click on the shared folder when no card is in the slot!
b) create raw VMDK file pointing at the drive and attach this before booting the guest. This one requires that the card is present... and is not switched out for a different card during operation - because the VMDK has the capacity/geometry specifications for the "drive" fixed from bootup.
Therefore I think the "easy" way around this for the developers of VirtualBox is to come out with a new variant of raw VMDK files - one which is specifically for removable disks, and which will present these disks as removable drives to the guest, and can tolerate geometry changes/drive not present. This would save a lot of development work on trying to architect paravirtualized PCI which is a tough ask at best.