I remember many years ago Joanna Rutkowska mentioned that a future version of Qubes would be able to have various computers with different functions interacting for performing different purposes.
Is there any progress regarding that, or is it planned?
More practically, to increase security for BTC transactions, I imagine a workaround could be to keep two Qubes installations on two distinct computers, one connected to internet and the other not, and be able to send BTC transactions from the network connected computer to the disconnected one for being signed, and finally moving the signed transaction from the network disconnected computer to the connected one, using this USB cable: USB-A 3.0 Data Transfer Cable for GNU/Linux (TPE-3TRANCBL) | ThinkPenguin.com.
In other words the cable is supposed to connect the sys-usb qube of the network connected computer with the sys-usb qubes of the disconnected computer.
This cable seems the only one able to transfer files between two Linux computers and does only that. All other common and bridge USB cables are unable to do even that on Linux. So, in the very unlikely case that dom0 of the network connected Qubes computer gets compromised, it seems impossible to compromise the sys-usb qube of the the network disconnected computer passing only through this referenced cable.
This cable is creating network connection between two computers.
It’s basically the same as connecting two computers using Ethernet cable.
But using USB devices to create network connection is less secure than connecting using PCI Ethernet controllers and connecting two computers with Ethernet cable. USB is more complex and could contain more vulnerabilities that could be exploited.
You also need to timely update your “offline” computer software/firmware so it wouldn’t have vulnerabilities that could be exploited from the “online” computer.
Yes, that is true. Many thanks for replying. What I was thinking was that the use of this cable is very unknown and difficult to find and so it is extremely unlikely that an hacker tries to find an exploit just for the few people who use this cable. Also how can the hacker know who is using this cable for this purpose, among the multitude of internet users or even the large amount of Qubes users, admitting that it is possible to identify Qubes users from the non-Qubes users?
Your link does not work for me. Thinking about that, a BTC transaction can be transferred to be signed as a file of a certain well defined type, size, etc. I know almost nothing of firewall rules, but maybe it is possible to establish a rule specific enough to only allow to pass BTC transaction files and block everything else.
On the other hand if you use some widely used e.g. Intel Ethernet controller that is much less complex and is thoroughly tested/analyzed for security issues and getting timely firmware updates in case of some issues then I’d think that it’d be harder to find vulnerability there compared to some unpopular USB device that is created by some company that is not really security-oriented and this device being rarely used so it’s not really maintained by the company.
I guess it’s more about protecting from targeted attack when attacker is already inside your online computer and is searching manually what do you have there and what they can do with it.
The link is about unidirectional transfer.
E.g. you’ll initiate the one-way sending of the file with transaction from online computer to the offline computer, sign the transaction in the file on offline computer and send the file back from offline computer to online computer using another one-way channel.
There are examples like using QR codes for transferring files.
Or using unidirectional network protocol with one-way fiber optic or RS-232 serial cable.