USB Type-C Port Manager Coming To Linux 4.12

[Vistaus]

Not only that, but adding HID classes like BT, with specific access only allowed for certain classes. a keyboard can only be a keyboard, storage can only act as storage. It works great. You can even half multiple roles, and at least in android you can limit what profile a device uses, and it shows what profiles a device asks for.

But that's a problem for devices like the MS Surface series, where the trackpad is recognized as a keyboard if you don't have a (recent) Linux kernel with Surface support. If you define HID classes, then you will have the exact same problem as w/o the right kernel: the whole thing is recognized as a keyboard. I know that the Surface series is one of the few where that happens currently but there are more devices like that coming, Lenovo also has a similar device and there are more manufacturers working on these kind of devices.

[dweigert]

I'm actually looking forwards to experimenting with USB 3.1, 10Gbps is a lot of bandwidth, and if it is possible, hooking two machines up and using it for short range Infiniband, or 10G Ethernet equivalent (I know, point to point). 10Gbps switches are still a little too pricey for the home, but I keep hoping...

[starshipeleven]

But that's a problem for devices like the MS Surface series, where the trackpad is recognized as a keyboard if you don't have a (recent) Linux kernel with Surface support. If you define HID classes, then you will have the exact same problem as w/o the right kernel: the whole thing is recognized as a keyboard. I know that the Surface series is one of the few where that happens currently but there are more devices like that coming, Lenovo also has a similar device and there are more manufacturers working on these kind of devices.

Not if this is a standard. HID is a standard.

[torsionbar28]

But USB is insecure.
A device can act as a keyboard and execute commands as if they were typed by a user.

FireWire and Thunderbolt are even worse, they have DMA and can read the RAM.

All true, same goes for SATA and SAS, and don't even get me started on PCI Express!! These interfaces are all so bad, they're racist, and they're causing global warming too.

[Vistaus]

All true, same goes for SATA and SAS, and don't even get me started on PCI Express!! These interfaces are all so bad, they're racist, and they're causing global warming too.

Trump Interfaces???

[LEW21]

All true, same goes for SATA and SAS, and don't even get me started on PCI Express!! These interfaces are all so bad, they're racist, and they're causing global warming too.

There is a big difference between an internal port and an external port.

[eydee]

But USB is insecure.
A device can act as a keyboard and execute commands as if they were typed by a user.

FireWire and Thunderbolt are even worse, they have DMA and can read the RAM.

Users are even more insecure, we still let them touch computers.

[nanonyme]

Having it work kinda like Bluetooth where you need to pair the devices would solve most of the issues. Of course this should be required only by input devices. I don't think there is massive need to encrypt the communication over the wire to avoid hardware keyloggers, but that might be cool to have too.

When you connect a HID device you get a popup saying "hey I detected keyboard with serial number 536827474, authorize its use with your root password" (and you can write your root password with it but it cannot do anything else like using desktop shortcuts or whatever to try to break into your system)

With mice it can show an on-screen keyboard and limit mouse input from that device on that part.

But something like this is not exactly trivial to implement and would break compatibility with most HID devices around. It would be a slaughter the same as the "free upgrade" to Win10.

Fun fact: when pairing bluetooth keyboards they ask you to write the keyboard's own password by typing on it, and it's plain retarded for a security standpoint, but we all know Bluetooth is unsafe anyway.

And the first keyboard would need to skip this check since otherwise you end up with catch-22 where you can't connect a keyboard to the machine because you need an already connected keyboard to type the root password. In reality you can't really solve this except for laptops with builtin keyboards properly since any software solution would be laughably complex and require changes in most parts of display and input stack

[zboson]

I have a Dell Dock WD15 which is a USB type-C to USB type-A hub with ethernet and HDMI/DisplayPort ports. The only thing that really works on with my Ubuntu 16.10 system is the Display Port pass through. At first everything works but then after a say a minute only the video pass through still works. I'm greatly looking forward to better Type-C port support.

[ckruetze]

I have a Dell Dock WD15 which is a USB type-C to USB type-A hub with ethernet and HDMI/DisplayPort ports. The only thing that really works on with my Ubuntu 16.10 system is the Display Port pass through. At first everything works but then after a say a minute only the video pass through still works. I'm greatly looking forward to better Type-C port support.

Interesting. We are using 4 WD15s in the office with Ubuntu 16.10 (and I think 16.04). And for us the only thing that doesn't work always is the display port. Frequently after a wake up from suspend the monitor on the display port connection will flicker. No problems with the other one connected to HDMI.

Using a very recent kernel and MESA version improved that a lot, guessing I'd say there is now a 20% chance that the monitor doesn't wake up properly. However turning off the monitor for a minute while using the dock will in most cases restore everything to working again.