[pkg-gnupg-maint] Bug#814584: gnupg2: gpg2 --card-status fail on armel / Raspberry Pi - "Card error"

[NIIBE Yutaka]

On 08/03/2016 10:10 PM, Petter Reinholdtsen wrote:
> My original test was done using a Freedombox image, which was armel.
> 
> Todays test was done using some random RPi and SD card I found on my
> desk and I did not notice it was using a different architecture.  I can
> try again on armel, if you believe it mattes.

I see the situation.  I have no idea about the segmentation fault.

I don't know if the difference of armel/armhf matters or not.  My
point was that in order to narrow down an issue (to be fixed), in
general, it would be better not to change other factors.

While I guess that the major problem is hardware related, I modified
the ccid-driver of GnuPG so that the difference between PC/SC service
can be smaller (less factors involved).

Well, if you would like to change other factors to get success anyway,
I'd recommend to take some hardware approach which possibly may
stabilize the USB communication:

  (1) Use good voltage supply to RPi board.

  (2) Only connect the card reader (+ the smart card) to USB port of
      RPi so that we can minimize the load of USB.  If you connect
      keyboard and mouse, try good ones (or none by using the
      network).

  (3) Use a USB HUB with an external voltage supply to connect the
      card reader.

My daily development environment and desktop environment is Wandboard
(armhf).  I test on BeagleBone Green (armhf) and SheevaPlug clone
(armel), too.

For those boards, USB host is not as good as PC, signal-wise and
power-wise.  For example, I can reproduce USB communication failure on
my Wandboard by inserting an additional USB Hub with no voltage supply
(while the USB specification allows many).
--