Add kapelse readers support in COMPOSITE_AS_MULTISLOT mode

[bmoraine]

Hello,

Please find 2 patches regarding COMPOSITE_AS_MULTISLOT mode to simplify its implementation and to add support for Kapelse readers supports.

Best Regards

Baptiste Moraine

[bmoraine]

Build issue when --enable-strict --enable-composite-as-multislot used for configuration is now fixed by last commit

[LudovicRousseau]

I pushed your fist commit with changes in https://github.com/LudovicRousseau/CCID-debug/

Please re-submit your patch in this -debug repo.

[LudovicRousseau]

Why this test?
In my list the KAP-LINK2 has 3 CCID interfaces https://ccid.apdu.fr/ccid/shouldwork.html#0x29470x0105

Do you have another model with a different configuration but using the same idProduct?

[bmoraine]

This test is needed as KAP-LINK2 is a USB composite device presenting either 3 CCID interfaces or 3 CCID interfaces (the 3 first ones) and a CDC-ACM interface according user configuration

[LudovicRousseau]

I see.
It should work fine if you something like use:

max_interface_number = 3; /* up to 4 interfaces */
num_CCID_interfaces = 3;  /* 3 CCID interfaces */

[LudovicRousseau]

@bmoraine Is it possible for the KAP-LINK2 to NOT have 3 CCID interfaces?

Why can't we use max_interface_number = 2; in all cases without checking the value of config_desc->bNumInterfaces?

[bmoraine]

@LudovicRousseau Effectively, KAP-LINK2 has always 3 CCID interfaces (the 3 first ones) and an optional 4th CDC-ACM interface. So, you are right, we can use max_interface_number = 2; in all cases without checking the value of config_desc->bNumInterfaces

[LudovicRousseau]

Can you also give me the output of the following command for each kapelse readers please

lsusb -d 2947: -v | grep 'bInterfaceClass\|idVendor\|idProduct'

I need to know what other interfaces are present and in which position.

For example the Gemalto Prox DU reader has 2 CCID interfaces and 1 HID interface. The HID interface is in the first position so max_num_interfaces is not the same as the number of "slots" as you suggested in your patch. The code you removed looked like a duplication but was not exactly the same.

[bmoraine]

Please find below the output of the command :
lsusb -d 2947: -v | grep 'bInterfaceClass\|idVendor\|idProduct'

1. For Kap&Link2

• When enumerating 3 CCID interfaces :
  $ lsusb -d 2947: -v | grep 'bInterfaceClass|idVendor|idProduct'
  idVendor 0x2947
  idProduct 0x0105
  bInterfaceClass 11 Chip/SmartCard
  bInterfaceClass 11 Chip/SmartCard
  bInterfaceClass 11 Chip/SmartCard

• When enumerating 3 CCID interfaces and 1 CDC ACM Interface :
  $ lsusb -d 2947: -v | grep 'bInterfaceClass|idVendor|idProduct'
  idVendor 0x2947
  idProduct 0x0105
  bInterfaceClass 11 Chip/SmartCard
  bInterfaceClass 11 Chip/SmartCard
  bInterfaceClass 11 Chip/SmartCard
  bInterfaceClass 2 Communications
  bInterfaceClass 10 CDC Data

2. For KAP-eCV :

$ lsusb -d 2947: -v | grep 'bInterfaceClass|idVendor|idProduct'
idVendor 0x2947
idProduct 0x0112
bInterfaceClass 11 Chip/SmartCard
bInterfaceClass 3 Human Interface Device
bInterfaceClass 2 Communications
bInterfaceClass 10 CDC Data

3. For all others Kapelse readers there are as many CCID interfaces as USB interfaces as for Kap&Link :
   $ lsusb -d 2947: -v | grep 'bInterfaceClass|idVendor|idProduct'
   idVendor 0x2947
   idProduct 0x0101
   bInterfaceClass 11 Chip/SmartCard
   bInterfaceClass 11 Chip/SmartCard

[bmoraine]

It's unclear for me. Do you want me to submit Kapelse reader suport in COMPOSITE_AS_MULTISLOT mode patch in this -debug repo? If yes, via a pull request?

[LudovicRousseau]

I made modifications to your first patch in the CCID-debug repo.
You now have to define 2 values: max_interface_number and num_CCID_interfaces.

Since the CCID interfaces on the kapelse readers are in the first interfaces num_CCID_interfaces should contain the number of CCID interfaces (easy) and max_interface_number should be num_CCID_interfaces -1 because it starts at 0.

Yes, I ask you to propose a Pull Request on this repo to add support of kapelse readers.
I do not have any of the kapelse readers so I can't test the code myself.
I don't want to publish untested code.

[bmoraine]

For information, I have created a Pull Request as recommended :
LudovicRousseau/CCID-debug#1

[LudovicRousseau]

Good.
have you checked your patches works fine on macOS with your readers?

[bmoraine]

No, I tested only on linux using "./configure --enable-strict --enable-composite-as-multislot " before compiling and installing driver.

In my understanding, driver should behave the same as on macos.

[LudovicRousseau]

I was expecting an answer like that :-)
I asked because your patch does not correctly set (I think) the max_interface_number value.
pcsc-lite on GNU/Linux does support composite devices. So you can't check your code on Linux.

You have to use macOS to test your changes work fine on macOS.

Please also test connecting 2 Kapelse readers and check all the CCID interfaces are visible using pcsctest.
The CCID driver has code to reset static_interface value when all the CCID interfaces have been used, so the driver is ready to connect another "composite-as-multislot" reader.
https://github.com/LudovicRousseau/CCID/blob/master/src/ccid_usb.c#L732

[LudovicRousseau]

@bmoraine you can also send me Kapelse sample readers so I can make tests on my side.
Contact me by private email.

[bmoraine]

@LudovicRousseau I will test Kapelse readers on macOS Sonoma and ,in the same time, send you some readers.

Regarding ccid driver generatioin on macos, I always slightly modified ccid driver source in SmartcardCCID :
https://github.com/apple-oss-distributions/SmartcardCCID/ and never update the entire ccid driver source files.

In my undestanding, Apple applies patches on top of cdid driver and use specific parameters when configuring ccid project according :
https://github.com/apple-oss-distributions/SmartcardCCID/blob/main/ccid/Makefile.
I assume that if I replace ccid tar.gz with the content of CCID-debug and I execute the makefile, it should be OK. Do you confirm?

[LudovicRousseau]

I assume that if I replace ccid tar.gz with the content of CCID-debug and I execute the makefile, it should be OK. Do you confirm?

I also assume it should work.
I never used the Apple process to build my CCID driver on macOS, so you are the expert on this point.

[bmoraine]

What is your process to build CCID driver on macOS?

[LudovicRousseau]

https://ccid.apdu.fr/#MacOSX

[LudovicRousseau]

Fixed in 73d9d3a and later patches