nitrokey / nitrokey-storage-firmware

Firmware for the Nitrokey Storage device

Nitrokey Storage Firmware

Firmware verification

To verify firmware flashed on device, please use the Nitrokey Storage Firmware Checksum Tool from the tools directory.

Building

Windows

Note: Tested with Windows 7

Install the following tools in this order:

1. avr32-gnu-toolchain-2.4.2-setup.exe
2. avr32studio-ide-installer-2.5.0.35-win32.win32.x86.exe
3. AvrStudio4Setup.exe
4. AVRStudio4.18SP2.exe

Linux

GUI

The compile procedure is as follows (tested on ArchLinux but should work on any other GNU/Linux OS):

1. Clone this git project (git clone https://github.com/Nitrokey/nitrokey-storage-firmware.git)
2. Download and extract AVR32 Studio. Example archive filename: avr32studio-ide-2.6.0.753-linux.gtk.x86_64.zip. At the moment no newer version seems to be available for Linux.
3. Start AVR32 Studio by executing avr32studio in the extracted folder.
4. Import project into AVR32 Studio: File | Import... | General | Existing Projects into Workspace | Choose the folder of downloaded git project.
5. Rename pm_240.h in the git project folder to pm_231.h. Make a backup of as4e-ide/plugins/com.atmel.avr.toolchains.linux.x86_64_3.0.0.201009140852/os/linux/x86_64/avr32/include/avr32/pm_231.h. Replace that pm_231.h with the renamed pm_240.h.
6. In AVR32 Studio select Project | Build All.
7. The builded file is now in the folder Debug of the git project folder.

Command Line

# setup
TOOLCHAIN_PATH=as4e-ide
T_SUBPATH=plugins/com.atmel.avr.toolchains.linux.x86_64_3.0.0.201009140852/os/linux/x86_64/avr32/include/avr32
mv $(TOOLCHAIN_PATH)/$(T_SUBPATH)/pm_231.h{,~} -v
cp pm_240.h $(TOOLCHAIN_PATH)/$(T_SUBPATH)/pm_231.h
# firmware build
cd src && make CC=path/to/avr32-gcc

Converting to .HEX file

Before flashing there may be a need to convert binary file to .hex. If it was not done automatically execute the following in Debug or Release directory:

avr-objcopy -R .eeprom -O ihex USB_MASS.elf firmware.hex

or, while in ./Debug directory:

make firmware.hex

Flashing the Firmware to Device

• Latest firmware releases
• General documentation
• Firmware upgrade instructions:
  • Windows
  • macOS
  • Linux
  • (previous documentation)

Debugging

Note: To connect an external debugger as described here, you will need a development version of the Nitrokey Storage that makes the JTAG pins available (pictured below). This version is currently not for sale. The PCB layout for the developer version can be found in the Nitrokey Storage Hardware repository, inside the dev folders.

Compatible Debuggers

This has been tested with the AVR JTAGICE XPII, however the more recent Atmel ICE and any other AVR UC3 compatible debugger should work as well.

Prepare connections

The JTAG connections on the PCB have a pitch of 1.27mm. To ease connecting and disconnecting, it is easiest to solder a pin header to the PCB and use a pin socket to quickly attach the device to the debugger. It is recommended to use the following parts for that (however any connector with a 1.27mm pitch will work):

Part	Digikey Part Number
7-pin THT Pin header, 1.27mm Pitch	S9014E-07-ND
7-pin THT Pin header, 2.54mm Pitch	S1012EC-07-ND
7-pin Socket, 1.27mm Pitch	S9008E-07-ND
1.27mm Ribbon Cable, ca. 15cm
Heatshrink

• Solder the 1.27mm Pin header to the board
• Solder the 1.27mm socket and 2.54mm header to the cable and isolate individual contacts with heatshrink

Connect Debugger interface to the Nitrokey

Use jumper wires to connect the cable from the Nitrokey to the Debugger interface connector as pictured below:

Nitrokey Side	AVR JTAG Connector Side
RST	nSRST
TCK	TCK
TDI	TDI
TDO	TDO
TMS	TMS
GND	GND
VDD	VTref

The device still needs to be powered via USB during debugging. For an initial function test, you can issue the following commands from the AVR32Studio home directory:

cd /plugins/com.atmel.avr.utilities.linux.x86_64_3.0.0.201009140848/os/linux/x86_64/bin
./avr32program --part UC3A3256S cpuinfo

if the device is connected correctly, this should yield an output similar to this:

Connected to JTAGICE mkII version 6.6, 6.6 at USB.

Device information:
Device Name                                   UC3A3256S 
Device Revision                               H
JTAG ID                                       0x7202003f
SRAM size                                     128 kB
Flash size                                    256 kB

Using the debugger in AVR32Studio

To enable the debugger, follow these steps inside the IDE:

• Enable the "AVR Targets" dialog under Window -> Show View -> AVR Targets
• Right click inside the "AVR Targets" window and select Scan Targets. Your debugger should now be shown as a target.
• Right click on the debugger entry and select Properties
• Select the "Details" tab. Under "Device", select AVR UC3A Series -> AT32UC3A3256S

The debugger should now be available. Configure the debugging environment by following these steps:

• Open Run -> Debug Configurations
• If there is no entry under AVR Application, create one by double clicking on it. Otherwise select the existing entry.
• Under "File", select Debug/USB_MASS.elf
• Under "Target" select JTAGICE mkII (or your correspondig JTAG debugger)
• Under "Erase Options" select Erase sectors
• Under "Run Options" select Reset MCU
• Apply Settings and close the dialog window

Congratulations, your IDE should now be ready for debugging. Set breakpoints as needed and start a JTAG debugging session by pressing F11.

Defaulting to USB DFU

If you accidentally erased the DFU bootloader from the chip or run into any trouble, the stick can always be reprogrammed like this:

• In the "AVR Targets" dialog, right click on your JTAG debugger and select Program Bootloader
• Leave all the entries in their default state and reprogram the bootloader by clicking Finish

The device will now start in DFU mode and can be programmed as described above in the Firmware Upgrade Instructions

nitrokey-storage-firmware's Issues

smartcard.Exceptions.CardConnectionException: Unable to connect with protocol: T0 or T1. Sharing violation

I am following raymii's guide for deleting keys

$ python2 tool/gnuk_remove_keys.py -p
returns

Admin password:
Traceback (most recent call last):
  File "tool/gnuk_remove_keys.py", line 112, in <module>
    main(passwd)
  File "tool/gnuk_remove_keys.py", line 86, in main
    gnuk.connection.connect()
  File "/usr/lib64/python2.7/site-packages/smartcard/CardConnectionDecorator.py", line 54, in connect
    self.component.connect(protocol, mode, disposition)
  File "/usr/lib64/python2.7/site-packages/smartcard/pcsc/PCSCCardConnection.py", line 128, in connect
    SCardGetErrorMessage(hresult))
smartcard.Exceptions.CardConnectionException: Unable to connect with protocol: T0 or T1. Sharing violation.

If I unplug and plug the key, I get different errors

$ python2 tool/gnuk_remove_keys.py -p
Admin password:
Token: Nitrokey Nitrokey Pro (xxxxxxxx) 00 00
ATR: xxxxxxxxx
Traceback (most recent call last):
  File "tool/gnuk_remove_keys.py", line 112, in <module>
    main(passwd)
  File "tool/gnuk_remove_keys.py", line 90, in main
    gnuk.cmd_verify(BY_ADMIN, passwd)
  File "tool/gnuk_remove_keys.py", line 54, in cmd_verify
    raise ValueError, ("%02x%02x" % (sw1, sw2))
ValueError: 6a88

python-2.7.13-1.fc25.x86_64 pyscard-1.9.0-5.fc25.x86_64 on Fedora 25

Insert static string on special-key double press

Allow to configure and insert a password string on a special-key double press, similarly to current HOTP functionality.

Device lockup during storage initialization

In the middle of random prefill the device turned unresponsive. Right
after reconnecting the device, Nitrokey application informed me of
underlying storage being fully prefilled, which wasn't true.

Excerpt from dmesg:
[179124.436222] usb 1-1.2: reset high-speed USB device number 102 using ehci-pci
[179124.724359] usb 1-1.2: device descriptor read/64, error -71
[179139.902709] usb 1-1.2: device descriptor read/64, error -110
[179140.078757] usb 1-1.2: reset high-speed USB device number 102 using ehci-pci
[179155.157304] usb 1-1.2: device descriptor read/64, error -110
[179170.335703] usb 1-1.2: device descriptor read/64, error -110
[179170.511653] usb 1-1.2: reset high-speed USB device number 102 using ehci-pci
[179180.921144] usb 1-1.2: device not accepting address 102, error -110
[179180.993440] usb 1-1.2: reset high-speed USB device number 102 using ehci-pci
[179191.402844] usb 1-1.2: device not accepting address 102, error -110
[179191.403670] usb 1-1.2: USB disconnect, device number 102
[179191.483159] usb 1-1.2: new high-speed USB device number 103 using ehci-pci
[179206.557575] usb 1-1.2: device descriptor read/64, error -110
[179221.735995] usb 1-1.2: device descriptor read/64, error -110
[179221.912112] usb 1-1.2: new high-speed USB device number 104 using ehci-pci

Crash on Password Safe status request

Storage v0.45, v0.38
App v1.0-beta.1
Windows 8.1
App cross-compiled with MXE (qMake)

Device crashes (device is not replying to commands, OS is detaching it, green LED stays on) after requesting Password Safe status just after mounting encrypted/hidden volume.

Old App (tested on v0.4) is not causing this behavior since it is not sending any additional commands directly after 'unlock' command. New App (v1.0-beta.1) sends GET_PW_SAFE_SLOT_STATUS which results in locking the device.

Scenario

It seems to be easily reproducible on Windows (MXE i686 cross-compiled binary).

1. Unlock encrypted volume,
2. Set up hidden volume and
3. Try unlock it
4. (edit) App sends GET_PW_SAFE_SLOT_STATUS to update PWS status.

Somewhere on this path the device crash is expected. If not on the first try, then after 3rd or 4th repeating.

App log: storage_crash.zip, crash time: 09:29:15

Workaround

Looks like adding STATUS request before asking for GET_PW_SAFE_SLOT_STATUS avoids the crash.
Simple 5 seconds delay suffices before next request. Lower values not always work.

Lockup on hidden volume activation

Sometimes device is locking up on hidden volume activation - either red LED stays on or off or LED stays on when connection to device is lost through HID interface. It is occurring during test mentioned in #20.
Issue might occur more frequent with lower pooling and sending delays (20 ms) between command packets.
Scenario and environment details same as in #20

Frequency: sometimes
Ubuntu 16.04
Nitrokey Storage v0.43
Delay between sending command and receiving response: 200/20 ms
Delay between retrying: 200/20 ms
Encrypted volume is not formatted
Stick is initialized with random data
Stick was reinserted just before test started

Firmware password change possible only once

Discovered while working on issue Nitrokey/nitrokey-app#97
It looks like device returns success only on first correct change (=> first change after factory reset, device reinsertion and application restart are not helping). Later password changes are always reported by device as failed.
Tested on both Windows 8.0 (compiled with Qt's mingw 5.3) and Ubuntu 16.04. I was using branch issue_97_firmware_password_feedback (the only differences in code with mainline are debug prints and feedback logic).

Prereq:
Firmware 0.36
App Nitrokey/nitrokey-app@bc3ce95#diff-3301dce0d4fa3f36ee73aa1a548416e0R2401
Both Ubuntu and Windows
Device is just after factory reset

Scenario:

1. Run application ./nitrokey-app -d -a
2. Initialize device (generate keys but do not clear SD, disable warning)
3. Change firmware password to asdasdasd using as current 12345678 (device returned TRUE as result so application shows success message)
4. Repeat step 3 with current asdasdasd and new one 123123123 (application shows error)

Any subsequent changes are not working.

Attaching output from app and debug log.
issue_97.txt
issue_97_debug_log.txt

As for App output clarification:

Old:  "12345678"
New1:  "asdasdasd"
New2:  "asdasdasd"
Old: "12345678"
New: "asdasdasd"

The last two lines (which are printed from function sending command to device) are just confirmation of the first 3 lines (which is Qt's form output). In case it would be confusing please let me know so I could simplify it.

write-protected append-only log file

Constant auto mounting the device under macOS 10.12.6

FW v0.46+ don't allow to unmount/eject the device through the finder. When you try to do so, the device disappears as expected, but show up again within a few seconds. So unplug of the device could destroy the device as it is still attached.

Workaround: sudo diskutier unmountDisk /dev/diskx where disk is the disk representing NK Storage.
FW v0.45 has not that behavior.

`No TDs queued` messages on USB3

Users reports about dmesg messages similar to the one below while using USB3:

[ 578.959764] xhci_hcd 0000:00:14.0: WARN Event TRB for slot 2 ep 10 with no TDs queued?

prodinfo command removes unencrypted volume from system

Occurency: always
Ubuntu 16.04
Nitrokey Storage
Prodinfo string:
CPU:0xf4780feb,SC:0x000000dc,SD:0x035503a4,SCM:0x1b,SCO:0x534d,DAT:2012.14,Speed:8820,Size:8,Firmware:0.34 - 5
Nitrokey app sha# Nitrokey/nitrokey-app@0dd1630

After issuing:

./nitrokey-app --cmd prodinfo

unencrypted volume disappears from the system and green LED blinks all the time.
Device physical removal and reinsertion restores correct functionality.

Attached parted output: parted -l
parted_before.txt parted_after.txt
And dmesg: dmesg.txt

Not updating HOTP counter while editing OTP slot

Storage v0.45
test: test_edit_OTP_slot, test_pro.py in libnitrokey test suite
Ubuntu 16.10

Device is not updating HOTP counter while editing HOTP slot. To test TOTP variant too.
More details in issue linked below.

TOTP not working for time bigger than int32 / after year 2038

TOTP returns wrong code with time set bigger than int32 max value (int32_max = 2 ** 31 - 1) according to both RFC spec and handcrafted test. NK Pro passes this. Attaching full log from test run:
TOTP_64bit_issue_log.txt (most interesting part is at the end, but the rest might be helpful too).

Date for max int32 is:

In [42]: datetime.datetime.fromtimestamp(2**31-1)                                                           
Out[42]: datetime.datetime(2038, 1, 19, 4, 14, 7)

Creation/unlocking of hidden volume not possible after sleep

Description

Creation of hidden volume is not possible: device returns status: 6.
In the firmware's source code it is defined as:

#define OUTPUT_CMD_STICK20_STATUS_NO_USER_PASSWORD_UNLOCK   6

After device reinsertion creation of hidden volume works. My guess is this is caused by Smartcard being powered off during PC's sleep mode stage or some other power management issue.

Environment

Storage v0.46
OSX Sierra 10.12.3 / 10.12.5 (latest)
Macbook Pro
Nitrokey App v1.1

Scenario

It looks like this is reproducible by following scenario:
Precond:

• Application opened v1.1, v0.46
• Hidden volume set up with known password, 70%-90%, slot 0, exFAT+MBR

1. Enable encrypted volume
2. Go into sleep mode for 30 seconds, wake up
3. Unlock hidden volume set up earlier

Currently this will result in device responding with status 6.
Reinserting the device fixes the issue.

Expected behaviour

Expected behaviour is to unlock/create the hidden volume without issues.

Export firmware command saves data on unencrypted storage

Export firmware command saves data on unencrypted storage making possibly a security risk since it could not be easily securely wiped-out from this volume. In case attacker steals device and identifies firmware using found data on unencrypted volume he can attack using known vulnerabilities if firmware is not up to date.

Allow to set unblock PIN (Reset Code) and to unblock device through HID

The OpenPGP Card allows to specify an unblock code which can be used if both User PIN and Admin PIN are blocked. The firmware should enable setting the unblock code and unblocking the device using it. This is similar as using User PIN and Admin PIN.

partitions are mounted multiple times

This issue happens for unencrypted and encryped volumes and occurs on Windows 7. A similar behaviour has been reported for macOS too but affecting hidden and encrypted volumes.

Factory reset sets firmware password to default

Factory reset should not change firmware password.

Ubuntu 16.04, Firmware 0.36, Nitrokey/nitrokey-app@5ec94e6

Scenario:

1. Change firmware password to password
2. Set update mode with password to confirm change
3. Flash device
4. Run Factory reset
5. Set update mode with password (would not work, but should)
6. Set update mode with 12345678 (would work)

prodinfo command never works first time

Occurency: always
Ubuntu 16.04
Nitrokey Storage
Prodinfo string:
CPU:0xf4780feb,SC:0x000000dc,SD:0x035503a4,SCM:0x1b,SCO:0x534d,DAT:2012.14,Speed:8820,Size:8,Firmware:0.34 - 5
Nitrokey app sha# Nitrokey/nitrokey-app@0dd1630

Command:

./nitrokey-app --cmd prodinfo

never works first time. Running second time works.

Multiple file copy to encrypted volume fails

Description

Copying files to freshly formatted encrypted volume fails when the device is after libnitrokey's full test suite.
Device is reset by the kernel and further not accessible, it responds to the HID commands though.

Occurrence: very frequent

Reproduction scenario

As of now issue cause has not been isolated yet and full test suite has to be run to provoke issue occurrence.

Environment

Ubuntu 16.10, Storage v0.46 RC,
libnitrokey: v2.0-147-gec500f1, branch: wip-show_information_about_invalid_CRC
Device connected to USB 2.0/3.0
SD card initialization already done

Steps

1. Run libnitrokey full test suite: pro + storage (without SD card initialization test - default), with --randomly-seed=1498651223
2. Reinsert the device
3. Run Nitrokey App
4. Unlock encrypted volume and format it to Fat32 (or ext4)
5. Mount the encrypted volume
6. Schedule a copy of 2GB of files with 5MB of size on average (about
   400 files) and another 1.5GB (1500 files).

Copy stopped after 500th MB (sometimes it stops after 4GB). After a brief moment the device was reset
(a couple of times) by the kernel. It was still responding to Nitrokey App commands. No LED was flashing.

Logs

Attached full system log.
dmesg.zip
Relevant messages shown:

Jun 28 17:32:25 feather kernel: FAT-fs (sdd): Directory bread(block 5217440) failed
Jun 28 17:32:59 feather kernel: FAT-fs (sdd): unable to read inode block for updating (i_pos 83479173)
Jun 28 17:32:25 feather kernel: FAT-fs (sdd): FAT read failed (blocknr 5105)

Workaround

To clear the issue it suffices to regenerate AES keys (e.g. by selecting Destroy Encrypted Data from Nitrokey App's tray menu) make a factory reset.

Device is connected after 20 minutes on latest Windows 10 "Creators Update"

Windows 10 Creators Update
Details: https://support.nitrokey.com/t/nitrokey-storage-extrem-slow-after-windows-10-creators-update/431/4
Quote:

After upgrading my computer to Windows 10 Creators Update (version 1703) mounting Nitrokey Storage is extremly slow.
It tooks ~15-20 minutes after insserting the stick to recognise the unencrypted volume and also to recognise the stick in the nitro key app.
After this time i can unlock the encrypted volume....and have to wait again to mount the unlocked volume.

enable security bit while keeping firmware updates possible

Issues with unlocking encrypted volume on macOS

Unlocking the encrypted volume fails, the device keeps returning busy status for a time and it unlocks the volume after all (Nitrokey App shows error and reconnects; sometimes returning WRONG_PASSWORD error code), but with constantly flashing both red and green LEDs.
After locking the device and unlocking the volume it works (as have in previous releases), but setting up hidden volume causes constant BUSY status replies with no LED activity.
Occasionally device returns BUSY right after insertion.

OS: Sierra 10.12.6
Storage: v0.48
Frequency: always
Nitrokey App v1.1

Adding high priority since issue blocks the use of main functionality.

Scenario:

1. Run Nitrokey App
2. Insert Storage device
3. Wait 10 seconds for device initialization
4. Try to unlock encrypted volume

Logs:
Issue occurrence - almost 1 minute of BUSY status: issue_44-encrypted_volume_unlock-3.txt
Retest on v0.47: issue_44-encrypted_volume_unlock-v0.47.txt

Initiating keys with wrong password results in flag StickKeysNotInitiated change

StickKeysNotInitiated is changing after sending command STICK20_CMD_GENERATE_NEW_KEYS to the device with incorrect admin password.

Prereq:
Nitrokey Storage, v0.35
Ubuntu 16.04
Initialization scenario:

1. Flash device
2. Make factory reset with application nitrokeyapp --cmd factoryReset

Test Scenario:

1. Run Nitrokey Application (in debug mode)
2. Select Initialize device from application context menu
3. Accept keys generation
4. Enter incorrect password for Admin, like 123123123
5. Accept further messages (click Ok), but do not agree for SD card clearing
6. Initialize device menu item is not shown (since it is depending on StickKeysNotInitiated flag)

Keys should not be generated (since incorrect password has been passed) but the flag StickKeysNotInitiated is changing. Please check is that expected behavior.

Before scenario:

GetStick20Configuration
HID_GetStick20Configuration
MagicNumber_StickConfig_u16      : 12824
ReadWriteFlagUncryptedVolume_u8  : 0
ReadWriteFlagCryptedVolume_u8    : 0
VersionInfo_au8[4]               : 0 35 0 6
ReadWriteFlagHiddenVolume_u8     : 0
FirmwareLocked_u8                : 0
NewSDCardFound_u8                : 1
SDFillWithRandomChars_u8         : 0
ActiveSD_CardID_u32              : 0x035503A4
VolumeActiceFlag_u8              : 1
NewSmartCardFound_u8             : 0
UserPwRetryCount                 : 3
AdminPwRetryCount                : 3
ActiveSmartCardID_u32            : 0x3B60
StickKeysNotInitiated            : 1

After scenario:

GetStick20Configuration
HID_GetStick20Configuration
MagicNumber_StickConfig_u16      : 12824
ReadWriteFlagUncryptedVolume_u8  : 0
ReadWriteFlagCryptedVolume_u8    : 0
VersionInfo_au8[4]               : 0 35 0 6
ReadWriteFlagHiddenVolume_u8     : 0
FirmwareLocked_u8                : 0
NewSDCardFound_u8                : 1
SDFillWithRandomChars_u8         : 0
ActiveSD_CardID_u32              : 0x035503A4
VolumeActiceFlag_u8              : 1
NewSmartCardFound_u8             : 0
UserPwRetryCount                 : 3
AdminPwRetryCount                : 2
ActiveSmartCardID_u32            : 0x3B60
StickKeysNotInitiated            : 0

Same after both application restart and device reinsertion:

GetStick20Configuration
HID_GetStick20Configuration
MagicNumber_StickConfig_u16      : 12824
ReadWriteFlagUncryptedVolume_u8  : 0
ReadWriteFlagCryptedVolume_u8    : 0
VersionInfo_au8[4]               : 0 35 0 6
ReadWriteFlagHiddenVolume_u8     : 0
FirmwareLocked_u8                : 0
NewSDCardFound_u8                : 1
SDFillWithRandomChars_u8         : 0
ActiveSD_CardID_u32              : 0x035503A4
VolumeActiceFlag_u8              : 1
NewSmartCardFound_u8             : 0
UserPwRetryCount                 : 3
AdminPwRetryCount                : 2
ActiveSmartCardID_u32            : 0x3B60
StickKeysNotInitiated            : 0

exported firmware doesn't match imported firmware

Perhaps because of a different bootloader...

Set unencrypted volume writable with Admin PIN?

This idea needs to be confirmed!

Once the Nitrokey App is preinstalled on the unencrypted volume, it should be set read-only by default and only allow the Admin PIN to change it.

HOTP codes not sent on special key double-press

Windows 8.1, Ubuntu 16.10
App v1.0-beta.2, v0.6.3
Storage firmware v0.38, v0.45

My Storage sticks were not sending HOTP codes on special key double-press despite enabling this in configuration. Pro device works as expected in same environments.

Enhancement: Encrypt unencrypted volume or format SD card

Objectives:

1. Enable reliable wiping of data on the unencrypted volume
2. Prevent evil maid attacks which physically remove the SD card storage and write to the unencrypted volume.

Possible Solutions:
a. Encrypt the "unencrypted" volume and unlock it when booting the device (without requiring a PIN).
b. Add the SD Card Formatter functionality into the firmware. This would not achieve objective 2.

Further readings on SD cards:

• http://www.bunniestudios.com/blog/?p=3554
• http://flash-extractor.com/
• http://recoverfab.com/

Enabling hidden volume results in stick freeze

Stick appears to be halted after enabling hidden volume in Nitrokey App and checking partitions using parted. Green Led stayed on. Application reported sticks connection and disconnection mutliple times. Attaching log.

Occurence: once
Nitrokey Storage 0.34
Ubuntu 16.04
App Nitrokey/nitrokey-app@0dd1630
hidden-volume-issue.txt

Scenario:

1. Unlock encrypted volume
2. Create partition table and format as fat32
3. Unlock hidden volume
4. Create partition table and format as fat32
5. Lock device
6. Unlock encrypted and hidden device (Unlock of hidden device is unsuccessful)
7. Create another hidden volume in different slot with same password
8. Unlock hidden volume
9. Check partition list with sudo parted /dev/sdc and command print

Support for new authorization mechanism

Recently new authorization mechanism has been introduced to NK Pro. Briefly it consist of:

• removing CMD_AUTHORIZE completely
• adding temporary password directly to command packets which needs to be authorized

Links:

• Nitrokey/nitrokey-pro-firmware#25
• Nitrokey/nitrokey-pro-firmware#29 (updated WRITE_TO_SLOT command)

Some commands combinations crashes the device

v0.45
App v1.1
Windows 7

Some device commands sequential combinations crashes the device or makes the device BUSY (as it is reporting through HID interface).
Example sequence for the latter case are GET_STATUS (Pro status command) and GET_DEVICE_STATUS (Storage status command) invoked one after another - in some cases it makes the device BUSY for 20 seconds, but later it functions properly.

Another case is getting Password Safe status just after unlocking encrypted volume (issue 31).

Log:
windows7-devicebusy-log.txt

Improvement - Create more Password Slots with dynamic structure to store them

You currently use a fixed size for each password slot you store on NK Storage. This will only give you 16 passwords and also a fixed size for each field inside the slot ( which could limit some usage case and also waste space ).

Proposal : Instead of using a fixed slot size (64 byte) for each password slot, a dynamic password slot size would be preferable.

Maybe the following:

{
u8 SlotStatusAndNextSlot_u8; // AND 0x11000000 ==> Status AND 0x0011 1111 ==> start of next slot
u8 SlotNameLength_u8; // No of bytes used after this byte for slot name
u8 SlotNameStartPtr_u8; // Placeholder : Pointer to the first byte of the slot name
u8 SlotPasswordLength_u8; // No of bytes used for password length
u8 SlotPasswordStartPtr_u8; //Placeholder : Pointer to the first byte of the password of that slot
u8 SlotLoginNameLength_u8: // No of Bytes used for User/Login Name
u8 SlotLoginNameStartPtr_u8; // Placeholder : Pointer to the first 
}

of course this is not the complete structure , but maybe give you an idea how I would structure the slots to squeeze the maximum out of the men space - maybe the read & write routines are a bit more complex. The 1st byte would not even need the start of the next slot coded as this would again limit the slot spaces down from 256 to 64 like your slots today ( but with a more flexible internal structure within the 64 bytes).

With 28 chrs for Slot name , password and login name + overhead of 4 bytes you would get 32 slots. I am sure there are different password requirements ( e.g. for a forum vs an bank account ) that would benefit from such a flexible slot handling.

Please use github release feature for binary releases

Hi,

can you please use GitHub release feature for your binary releases?
It is easier to stay up to date and get informed about new releases.

If you'd like to include binary files along with your release, such as compiled programs, drag and drop or select files manually in the binaries box.

https://help.github.com/articles/creating-releases/

Thank you.

New command: Blink LEDs

A command for blinking LEDs would be useful for signalizing from the host (Nitrokey App / libnitrokey) that the device needs action or has finished its time-requiring operation.

"New SD card" flag still set after SD being cleared

Nitrokey App v0.6.3
Storage v0.45
Ubuntu 16.10

Device is not clearing new SD card flag in its internal state after initialization. In consequence "New SD card found" message (depending directly on the byte-flag named similarly in the device's status) is still shown in about window after executing factory reset and clearing SD card with random data.

Not every HID action makes the LED blinking

At the moment not every HID action processing is signalized. Perhaps on finishing processing of a HID command a LED could blink?
With current behavior getting data from PWS is not signalized on the device's LED panel.

Hidden volumes not always unlocking after creation

Hidden volumes are not always unlocking after creation. Device returns wrong password error code. In test where 4 hidden volumes are created and later on unlocked each (by going through lock device) it failed 7/10. Encrypted volume is not formatted so OS should not interfere with any mounting or any write operations. It fails sometimes on unlocking first volume, sometimes later (mentioned in logs, each volume has password consisting of fixed string and its number).

Frequency 7/10
Ubuntu 16.04
Nitrokey Storage v0.43
Delay between sending command and receiving response: 200 ms
Delay between retrying: 200ms
Encrypted volume is not formatted
Stick is initialized with random data
Stick was reinserted just before test started
Test time: 218 seconds / 10 iterations

Test scenario (from unit test, where status 4 means wrong password):

    def test_encrypted_volume_setup_multiple_hidden(C):
        hidden_volume_password = 'hiddenpassword'
        p = lambda i: hidden_volume_password + str(i)
        assert C.NK_lock_device() == DeviceErrorCode.STATUS_OK
        assert C.NK_unlock_encrypted_volume(DefaultPasswords.USER) == DeviceErrorCode.STATUS_OK
        for i in range(4):
            assert C.NK_create_hidden_volume(i, 20+i*10, 20+i*10+i+1, p(i) ) == DeviceErrorCode.STATUS_OK
        for i in range(4):
            assert C.NK_lock_device() == DeviceErrorCode.STATUS_OK
            assert C.NK_unlock_encrypted_volume(DefaultPasswords.USER) == DeviceErrorCode.STATUS_OK
>           assert C.NK_unlock_hidden_volume(p(i)) == DeviceErrorCode.STATUS_OK
E           assert 4 == 0
E            +  where 4 = <cdata 'int(*)(char *)' 0x7f06c35100b0>('hiddenpassword0')
E            +    where <cdata 'int(*)(char *)' 0x7f06c35100b0> = <cffi.api.FFILibrary_../build/libnitrokey.so object at 0x7f06c37f6f50>.NK_unlock_hidden_volume
E            +    and   'hiddenpassword0' = <function <lambda> at 0x7f06c3801b18>(0)
E            +  and   0 = DeviceErrorCode.STATUS_OK

Attached detailed log: issue_unstable_hidden_volumes.txt

Support bigger OTP secrets (up to 40 bytes)

In NK Pro OTP secrets have been extended to support AWS.

Links:

• Nitrokey/nitrokey-pro-firmware#29

No serial numbers presented on prodinfo results

79e07a7
Nitrokey/nitrokey-app@0a89ca5
Ubuntu 16.04
Frequency: always

One of my sticks suddenly stopped sending serial numbers. This is an excerpt from prodlog.txt:

timestamp:25.06.2016 13:21:21,CPU:0x00000000,SC:0x00000000,SD:0x035503a4,SCM:0x1b,SCO:0x534d,DAT:2012.14,Size:8,Firmware:0.36 - 7

Full file: prodlog.txt

I am checking will second stick behave same after firmware upgrade.

Device crash/ejection on unencrypted volume unlock / WUDFHost.exe

Description

Issue frequency: once
During the tests suddenly the device stopped responding. Additionally:

• WUDFHost.exe started to take 100% of one of the cores,
• Application has shown tray message about the device being disconnected, unencrypted
  volume vanished,
• usbview shows device's general failure,

Environment

Windows 8.1
USB 2.0
App v1.1
Storage v0.46
Device state: after libnitrokey tests (done on a Ubuntu a minute before)

Scenario

1. Check if unencrypted volume shows up and is accessible
2. Check PWS/OTP functionality
3. Try to unlock encrypted volume

Logs

Excerpt from system log below (looks like issue was caused by CCID, but possibly linked with device ejection by the OS)

- <Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
- <System>
  <Provider Name="WudfUsbccidDriver"
Guid="{22C370A7-A3DB-4390-ADE5-3A1ACCF4B5D5}" />
  <EventID>7</EventID>
  <Version>0</Version>
  <Level>2</Level>
  <Task>1</Task>
  <Opcode>10</Opcode>
  <Keywords>0x8000000000000000</Keywords>
  <TimeCreated SystemTime="2017-06-15T11:10:47.423497100Z" />
  <EventRecordID>1901578</EventRecordID>
  <Correlation />
  <Execution ProcessID="5932" ThreadID="7520" />
  <Channel>System</Channel>
  <Computer>sz1070</Computer>
  <Security UserID="S-1-5-19" />
  </System>
- <EventData>
  <Data Name="Name">ReaderCompletionUnknownMsgType</Data>
  <Data Name="Value">0x0</Data>
  </EventData>
  </Event>

Device responded with an error status.
Status: ReaderCompletionUnknownMsgType

OTP SECRETs begining with 0x00 are not stored in the slots

Issue details (same registered on NK Pro):
Nitrokey/nitrokey-pro-firmware#19
Confirmed on line:
https://github.com/Nitrokey/nitrokey-storage-firmware/blob/master/src/OTP/hotp.c#L1100
Proposed solution:
Nitrokey/nitrokey-pro-firmware#20

Rejected IOCTL on Windows 8.1

Description

While reading event log on Windows 8.1 I have encountered the following error event:

Smart Card Reader 'Nitrokey Nitrokey Storage 0' rejected IOCTL 0x313520:
Incorrect function.  If this error persists, your smart card or reader
may not be functioning correctly.

Environment

Windows 8.1
USB 2.0
App v1.1
Storage v0.46

Scenario

Insert the stick and check event log.

Present unlocked state by blinking LEDs

Device could present when it is in unlocked state (PWS or Encrypted/Hidden volumes are available) through blinking with one of the LEDs. This would remind user about locking it after use.

Device is not accepting firmware password after changing

Discovered while working on issue Nitrokey/nitrokey-app#97
Might be connected with #11

Prereq:
Firmware 0.36
App Nitrokey/nitrokey-app@bc3ce95#diff-3301dce0d4fa3f36ee73aa1a548416e0R2401
Ubuntu 16.04. I was using branch issue_97_firmware_password_feedback (the only differences in code with mainline are debug prints and feedback logic).

Device has just changed firmware password with application from 12345678 to asdasdasd

Scenario:

1. Send update command ./nitrokey-app --cmd setUpdateMode=asdasdasd
2. Check is device in update mode

Reboot command

I would make use of command which reboots stick. Currently I have to remove and insert again same device for achieving same 'start' state. Would that be possible to implement? Or, is there any other way?
I thought it could be possible through dfu-programmer at32uc3a3256s start, but it shows error it can not find any device.

Allow changing of unencrypted/encrypted storage ratio dynamically (without reflashing)

How large is the storage capacity?

Nitrokey Pro, Nitrokey Start, Nitrokey HSM and Nitrokey U2F don't contain storage capability for ordinary data (it can only store cryptographic keys and certificates).
Nitrokey Storage can store and encrypt 8, 32, or 64 GB of data (depending on particular model).

Upon receiving the device this doesn't appear to be the case. I have a 2GB unencrypted partition and after initializing the device there's another ~5.5GB encrypted device that I'm able to format to whatever filesystem I wish. My question is, how do I reclaim some of that 2GB and shrink it. Or in fact, how can I expand it -- for instance, to install Arch+i3+GRUB etc (might need a bit more than 2GB).

Improve performance

Firmware 0.48 provides about >3 MB/s write performance only.

AES keys regeneration should not force SD card random fill

Hi!
Currently AES keys regeneration command sets SDFillWithRandomChars_u8 to 0 which forces user to make SD card random fill each time.
SD card random overwrite should be done only on first use of given SD card and should not be connected to AES keys regeneration.

For more details please check: Nitrokey/nitrokey-app#45 (comment) and Nitrokey/nitrokey-app#45 (comment)

Attaching logs: after_generating_new_AES_keys.log.txt
before_generating_new_AES_keys.log.txt

ReadMinWatermark > ReadMaxWatermark

Windows
Storage v0.43
Nitrokey App v0.6.3

While setting up hidden volume application has shown that it can be placed between 50th (ReadMinWatermark) and 49th (ReadMaxWatermark) percentile of the SD storage which is not possible. These two values are shown directly as the device has sent to the Application (except when it is x<10 or x>90). It would be great to add proper check in firmware so the watermarks values could be in proper relation to each other.

This situation (ReadMin>ReadMax) could also mean, that the user has overwritten whole SD card and he cannot create hidden volume without losing existing data on encrypted volume. Please clarify.

Details: https://support.nitrokey.com/t/hidden-volume-and-plausible-deniability/388/3

New status: smartcard in use

Firmware v0.48
OS: Mac OS 10.12.6

Since many HID commands accesses the smartcard currently they can fail with unknown reason (WRONG_PASSWORD or device would send constantly a BUSY status) while the smartcard is in use by CCID interface/endpoint.
I would like to add new status CCID_IN_USE to show that requested command cannot be applied due to constant parallel usage, which cannot be interrupted/concurrently executed.
This looks to be especially important on v0.48 and Mac OS, where CCID endpoint seems to be in use, blocking HID commands from execution. New device status would then show the cause of their failure.
It should be returned after maximum of 10 seconds of retrying access request to smartcard (on contrary to returning BUSY infinitely).

Responses for certain commands contain wrong CRC

Certain command responses have a CRC value in them that does not match the actual content. See for instance here where this has been mentioned for the send password case. I have my own library and see the same issue happening for the response of the unlock encrypted volume command.

Can this please be looked into? That is no state to work with. For that reason libnitrokey essentially cuts down on CRC checks completely which circumvents the entire concept of providing such check codes in the first place.