Add ccid apps to the usbip runner

[sosthene-nitrokey]

This PR cleans up usbd-ccid a bit and adds the CCID apps (opcard) to the usbip runner.

Closes #140, #150 and #151

On my machine I enconter 2 bugs:

• pcscd and this somehow manage to trigger a kernel bug:

  1. `cargo run --release --features alpha
  2. usbip attach -r localhost -b 1-1 (enough times for it to not return immediately)
  3. opgpcard status
  4. now lsusb hangs and dmesg shows an error: kernel BUG at mm/usercopy.c:101!

• gpg still fails to detect this as an openpgp smartcard for some reason

[robin-nitrokey]

Regarding the first problem: I think I can reproduce this. opgpcard hangs, and subsequent usbip calls don’t seem to work at all.

[sosthene-nitrokey]

Does lsusb work after that?

[robin-nitrokey]

No.

[sosthene-nitrokey]

Anything suspicious in dmesg?

[robin-nitrokey]

Yes, basically the same one you had:

usercopy: Kernel memory exposure attempt detected from SLUB object 'kmalloc-16' (offset 0, size 64)!
kernel BUG at mm/usercopy.c:99

[sosthene-nitrokey]

There appear to be a similar bug already reported: https://bugzilla.kernel.org/show_bug.cgi?id=215487

The symptoms are the same, but the dmesg error is not.

[sosthene-nitrokey]

I included a fix for #151

[sosthene-nitrokey]

Includes a fix for #150

[robin-nitrokey]

LGTM!

[sosthene-nitrokey]

It looks like ddce9f8 fixing #151 actually breaks the firmware.

[sosthene-nitrokey]

Ok, so the the way it was done before the fix for #151 is correct, because the response blocks can be sent in multiple parts and it's just that the gnuk test suite doesn't support this.

I still don't understand exactly how my fix made thing worse for the firmware itself, but for now I'll revert it. We should consider fixing the gnuk test suite to use it.

Edit: see #151 (comment)

[robin-nitrokey]

Not tested yet, but the code changes look good to me.

[sosthene-nitrokey]

I think we should merge this now. It works well on the firmware itself. I'll then upstream the usbd-ccid changes into https://github.com/trussed-dev/usbd-ccid

[ZenithalHourlyRate]

Regarding usbip, I think I have found a bug in your usbip server, see https://lore.kernel.org/lkml/ZBHxfUX60EyCMw5l@Sun/

[robin-nitrokey]

@ZenithalHourlyRate Thank you for looking into the bug report! I will try to fix the issue you pointed out.