Difference between revisions of "Belgian eID"

From YobiWiki
Jump to navigation Jump to search
Line 151: Line 151:
 
I'm using the [[IDream ID-SMID01 SmartCard reader]], bought for 10€
 
I'm using the [[IDream ID-SMID01 SmartCard reader]], bought for 10€
 
<br>So the card is accessed via the USB reader, handled by the libccid, used by the pcscd daemon.
 
<br>So the card is accessed via the USB reader, handled by the libccid, used by the pcscd daemon.
  +
  +
<pre>
  +
Reader 0: iDream ID-SMID01 00 00
  +
Card state: Card inserted,
  +
ATR: 3B 98 13 40 0A A5 03 01 01 01 AD 13 11
  +
  +
ATR: 3B 98 13 40 0A A5 03 01 01 01 AD 13 11
  +
+ TS = 3B --> Direct Convention
  +
+ T0 = 98, Y(1): 1001, K: 8 (historical bytes)
  +
TA(1) = 13 --> Fi=372, Di=4, 93 cycles/ETU (38400 bits/s at 3.57 MHz)
  +
TD(1) = 40 --> Y(i+1) = 0100, Protocol T = 0
  +
-----
  +
TC(2) = 0A --> Work waiting time: 960 x 10 x (Fi/F)
  +
+ Historical bytes: A5 03 01 01 01 AD 13 11
  +
Category indicator byte: A5 (proprietary format)
  +
  +
Possibly identified card (using /usr/share/pcsc/smartcard_list.txt):
  +
3B 98 13 40 0A A5 03 01 01 01 AD 13 11
  +
Belgium Electronic ID card
  +
</pre>
  +
 
==Linux: Government Middleware==
 
==Linux: Government Middleware==
 
===Installation===
 
===Installation===

Revision as of 22:47, 7 April 2008

Belgian eID is part of the efforts of the government for Belgian eGov

News

Officials

Specifications

Usage & Software

I revoked my certificates

Why?

Because at that time I knew too few on the details of the eID architecture and too much about how a new security architecture can have flaws, so better to stay away for a while, especially given the legal implications that the eID can bring.
I knew they were talking about two certificates without understanding their difference, so let's revoke both.
Note that it doesn't mean my eID is not valid, eID card activation is mandatory. The eID card is a proof of identity and residence of a person in Belgium. eID certificates activation is a choice of the holder of the eID (opt-in), he/she can decide to activate or revoke the certificates. (cf FAQ)

How?

It was quite epic.
I was still a bit prepared, hopefully, so I had printed the Annexes 3 & 10, the legal forms to ask either to renounce to have the certificates or to revoke them just after activation, as well as the relevant parts of the User Manual for the civil servants :-)
I printed both as for me it was not clear from the User Manual how to renounce for the certificates.

  • [Me] Good morning, I come for my eID and... I want to get rid of the certificates
  • [Her] You what? Is it possible? Never heard about that
  • [Me] Yes, yes, see (and I show her what's in her User Manual)
  • ... (takes a while to digest the info)
  • [Her] Ha ok, you've to fill Annex 3, (shouting behind her shoulder) JOSETTE, DO WE HAVE ANNEX 3???
  • [Josette] ANNEX WHAT??
  • [Me] No prob, look (and I pull out my own copy of Annex 3 (pdf))
  • ... (meanwhile she has no idea what to do on the eID to fulfill renouncement and I agree with her, the Manual was unclear so we went for activation+revocation)
  • [Me] No prob, look (and I pull out my own copy of Annex 10 (pdf))
  • [Her] Can I take your copies before you fill them? I'll make copies for ourselves.

So today the official Annex 3&10 forms of this civil office are mine :-D
Who knows, maybe I crafted the form... ;-)

It is "funny" to see that everywhere the activation of the certificates is presented as optional, the choice being left to the citizen, blablabla, but in reality most of the citizen just don't know about this possibility (and most of the civil servants, well I hope that since my story, things have evolved a bit) and the civil servants doesn't ask you the question.

According to Danny de Cock you can revoke them by phone via the eID card stop service: call +32-2-518.21.16 or +32.2.518.21.17,in French or Dutch, respectively (there is a 7-day period prior to definitive revocation, I'm not sure how secure is the procedure...)

Some details about those certificates

The eID contains 2 signature certificates, so you cannot encrypt with them, just sign.

  • One labeled "Authentication" is for daily use
  • One labeled "Signature" is for special cases
    • It has a equivalent legal value as a hand-written signature and is referred as non-repudiation signature.
    • According to CertiPost applicability guidelines (pdf), the electronic signature still cannot replace a hand-written signature in a number of cases
    • It's automatically revoked for minors under 18 years as they cannot sign legally yet.
    • From Certipost FAQ:
      Non-repudiation guarantees that one cannot deny having performed an act. E.g. any message signed using a person's digital signature can only have come from this person. The signing person can not claim that the message was not originated by him.
      In other words, non-repudiation means that information cannot be disclaimed, similar to a witnessed handwritten signature on a paper document.

      to be checked in the law
    • Normal pkcs software doesn't seem to be able to use it (?)
    • Should be used only through the Government software which prompts you with a special GUI and warnings about the legal power of this signature
  • CertiPost e-Registered mails is using the non-repudiation signature
  • CertiPost e-Signing is using the non-repudiation signature
  • Both are protected by the same PIN which is typed on your (unsafe) PC

What says the law

Sorry, here are french version abstracts.
9 JUILLET 2001. - Loi fixant certaines règles relatives au cadre juridique pour les signatures électroniques et les services de certification, date de publication au Moniteur: 29 septembre 2001
Morceaux choisis:

  • Art. 4. § 1er. A défaut de dispositions légales contraires, nul ne peut être contraint de poser un acte juridique par voie électronique.
  • § 4. Sans préjudice des articles 1323 et suivants du Code civil, une signature électronique avancée réalisée sur la base d'un certificat qualifié et conçue au moyen d'un dispositif sécurisé de création de signature électronique, est assimilée à une signature manuscrite, qu'elle soit réalisée par une personne physique ou morale.
  • § 5. Une signature électronique ne peut être privée de son efficacité juridique et ne peut être refusée comme preuve en justice au seul motif :
    • que la signature se présente sous forme électronique, ou
    • qu'elle ne repose pas sur un certificat qualifié, ou
    • qu'elle ne repose pas sur un certificat qualifié délivré par un prestataire accrédité de service de certification, ou
    • qu'elle n'est pas créée par un dispositif sécurisé de création de signature.
  • Wait a minute, ANY electronic signature NOT being based on a qualified certificate and NOT created in a secure environment CANNOT be refused as a legal proof???
    Maybe it's me or the triple-negation sentences (lawyers, lawyers...) but it looks like § 5 goes much further than § 4

What do I think today?

Catastrophe scenario

  • Someone captures your PIN while you're using it for "just authentication" via e.g. malware, virus, trojan, worm on the PC
    • you know, the kind of stuff that never happens, anyway it's now your problem even if the official middleware is not signed
    • And apparently even some readers with integrated keypads are not safer :-(
    • And apparently even commercial standalone smatcard terminals are not safer :-( (pdf)
    • And, oh, I did a small test with lkl, a userland keylogger and of course the PIN typed into the beid graphical prompt could be easily captured.
  • He gets physical access to your eID, even briefly if he has e.g. a PDA with Internet & smartcard reader.
  • Now he can sign with your legal signature anything you can imagine... and you cannot repudiate what he does.
  • The fact that maybe legal signatures have to be crafted through CertiPost (cf e-signing below) doesn't change anything to this risk.

So what will I do next time?

  • I'll probably accept the Authentication certificate to be able to play around with it.
  • For sure I'll revoke the Signature certificate unless they change their architecture.
    • Not the same PIN than the other certificate.
      UPDATE: some discussions here
      UPDATE: Public user specification BELPIC application v2.0 mention 2 different PINs with their own ids (01 for auth, 04 for non-repud) and according to Danny the new cards will have 2 PINs
    • Better than 4 digits (& 3 attempts so 3 chances over 10000)
      UPDATE In fact we can change the PIN to up to a 12-digit value.
    • Probably they limited themselves to one single PIN in order to pass the Kafka test ;-)
    • Even then:
      • Then I could use it but only on trusted devices, that's another story...
      • You get the PIN & PUK by post if I remember well, this could be eavesdropped but you can change your PIN... So as the PUK can unlock the PIN 12 times, the attacker has 36 chances over 10000, one over 278, mmm... And who said humans can generate random PINs? ;-)
        UPDATE: Actually you get only half of the PUK, 6 digits, and you need to go to the municipality to have your eID card unblocked... This unblocking consists of sending 12 PUK digits to your eID card: you provide 6 PUK digits, and the National Register provides the other 6... I.e., it is impossible for a citizen to unblock his/her eID card without presenting him/herself to the municipality... Thanks for the info Danny!

Privacy and other security considerations

  • Another consideration I didn't talk about yet: PRIVACY
  • Whoever sees your public certificate (which happens e.g. if you log to a SSL website with your card or if you simply send a signed email) sees your RRN ("rijksregistratienummer")
    BTW here is how it's constructed: It is a total of 11 numbers of which the first 6 are your birthdate JJMMDD followed by three numbers to distinguish the people that were born on the same day (even for men, uneven for women) and the last two numbers are a control number. So if you know the birthdate and the sex of the person, and you know how the control sum is done (97-JJMMDDXXX%97), you only miss two and a half numbers coming from a linear incremental counter (001, 003,...) to recompile his national register number. For mine it takes less than 40 attempts it you proceed logically...
  • Tax-on-web announce it but what about the others and your mail correspondants?
    Suite à l'utilisation de votre moyen d'authentification (carte d'identité électronique ou token citoyen), le SPF Finances a connaissance de votre numéro de registre national.
    Conformément à l'arrêté royal du 25/04/1986 autorisant certaines autorités du Ministère des Finances à utiliser le numéro d'identification du registre national des personnes physiques, votre numéro de registre national n'est utilisé dans ce contexte qu'à des fins d'identification pour l'accès aux applications du SPF Finances.
    La loi du 8 décembre 1992 relative à la protection de la vie privée à l'égard des traitements de données à caractère personnel s'applique à ce traitement d'identification dont le responsable est le SPF Finances, Boulevard Albert II, 33 à 1030 Bruxelles.
  • Other "funny" facts: Belgium government doesn't bother about cross-certification with any common Root CA so when you want to visit an official site supposedly secured such as https://ccff02.minfin.fgov.be/CCFF_Authentication/choseLoginMethod.do or https://mijndossier.rrn.fgov.be you're kindly asked to blindly trust the certificate, years after phishing was invented, sigh...
    You can import Belgium Root CA signed by GlobalSign Root CA here (the belgiumrs*.crt)
  • And even "better": for Firefox, CertiPost e-Signing requires you to download and install their CA certificate and to trust it for identifying everything: web sites, mail users and software! Here the download of the CA certificate is done... on pure http, sigh...
  • ADAPID project is a consortium of researchers and industry representatives in Flanders decided to take action in an attempt to help avoid a national privacy calamity. After a first (non-public) report, the ADAPID project won the financial support of IWT-Flanders. ADAPID officially started July 1st, 2005 and will run until June 30th, 2009.
  • Normally the third and definitive version of the eID should have been rolled out begin of this year (2008) but I've no idea what are the changes.
  • Ethical-ID is a software which presents to the e.g. swimming pool employee the only relevant data e.g. the residence city.
  • And what about administrative errors? ;-)

Linux: Drivers

If you want to try also make sure you're using Linux :-D

I'm using the IDream ID-SMID01 SmartCard reader, bought for 10€
So the card is accessed via the USB reader, handled by the libccid, used by the pcscd daemon.

 Reader 0: iDream ID-SMID01 00 00
  Card state: Card inserted, 
  ATR: 3B 98 13 40 0A A5 03 01 01 01 AD 13 11

ATR: 3B 98 13 40 0A A5 03 01 01 01 AD 13 11
+ TS = 3B --> Direct Convention
+ T0 = 98, Y(1): 1001, K: 8 (historical bytes)
  TA(1) = 13 --> Fi=372, Di=4, 93 cycles/ETU (38400 bits/s at 3.57 MHz)
  TD(1) = 40 --> Y(i+1) = 0100, Protocol T = 0 
-----
  TC(2) = 0A --> Work waiting time: 960 x 10 x (Fi/F)
+ Historical bytes: A5 03 01 01 01 AD 13 11
  Category indicator byte: A5 (proprietary format)

Possibly identified card (using /usr/share/pcsc/smartcard_list.txt):
3B 98 13 40 0A A5 03 01 01 01 AD 13 11
        Belgium Electronic ID card

Linux: Government Middleware

Installation

The Belgian government is providing a Linux middleware to access the eID.
The sources are accessible here (fr) or there (nl)
But thanks to Wouter Verhelst, there are also Debian packages (2.6.0-3 in Lenny as I'm writing):

apt-get install beidgui beid-tools
=> libopenct1 libpcsclite1 libbeidlibopensc2 libbeid2 beid-tools beidgui libccid pcscd 

Some interesting documentation once it's installed: /usr/share/doc/libbeidlibopensc2/README.Debian

A short introduction to the middleware is available here (fr, pdf) or here (nl, pdf)

Belpic version of OpenSC

The middleware is a modified version of OpenSC, talking to pcscd.
I recently saw that my ~/.xsession-errors logfile was full of Error: can't open /var/run/openct/status...
It happens whenever icedove/iceweasel are open (so when the libbeidpkcs11.so is loaded) I found a bugreport on Ubuntu and the proposed fix works so I opened a Debian bugreport: #469485:

OpenSC has support for three driver types : PCSC, OpenCT and CT-API. Belpic only needs PC/SC, and will produce errors/warnings if you leave support for OpenCT enabled.

Edit /etc/beidbase.conf, and insert a statement that limits the use of drivers to pcsc. Right before the reader_driver config feels like an OK place to do this :

## specify driver family pcsc.
# Others (openct, ..) are not needed for Belpic and
# may produce errors/warnings

       reader_drivers = pcsc ;

reader_driver pcsc {
     ....

GUI

The GUI application (beidgui) works well, including OCSP communication, showing me that my eID certificates are revoked, excellent!
You can easily change your PIN here, with a PIN between 4 and 12 digits!
I don't remember of having read that PINs bigger than 4-digit were possible...

beidcrld

Part of beid-tools
It's an optional daemon, supposed to download automatically the CRLs.
TODO: where are those CRLs stored locally? How to check the status? /usr/share/beid/crl

beidpcscd

Part of beid-tools
It's an optional daemon.
The privacy filter monitors all commands sent to the card. When an application requests to read the identity data, address or photo from the eID card, the filter will display a message and ask the user's consent
it's listening on port tcp 2500 and beidcrld, Iceweasel & Icedove (through the PKCS#11 module we'll install later) are constantly speaking with it...
And if it's not running, Iceweasel & Icedove will poll every second on that port 2500, no matter if you are really using the eID at that moment or not, erk!

beid-pkcs11-tool

Part of beid-tools
For a little demo...

$ beid-pkcs11-tool --list-slots
Available slots:
Slot 0           iDream ID-SMID01 00 00
  manufacturer:  Zetes
  hardware ver:  1.0
  firmware ver:  1.0
  flags:         token present, removable device, hardware slot
  token label:   BELPIC (Basic PIN)
  token manuf:   Axalto
  token model:   Belgium eID
  token flags:   rng, PIN initialized, token initialized
[...]
$ beid-pkcs11-tool --list-objects
Private Key Object; RSA 1024 bits
  label:      Authentication
  ID:         02
  Usage:      sign
Certificate Object, type = X.509 cert
  label:      Authentication
  ID:         02
Public Key Object; RSA 1024 bits
  label:      Authentication
  ID:         02
  Usage:      encrypt, verify
Certificate Object, type = X.509 cert
  label:      CA
  ID:         00
Public Key Object; RSA 2048 bits
  label:      CA
  ID:         04
  Usage:      encrypt, verify
Certificate Object, type = X.509 cert
  label:      Root
  ID:         00
Public Key Object; RSA 2048 bits
  label:      Root
  ID:         06
  Usage:      encrypt, verify
Private Key Object; RSA 1024 bits
  label:      Signature
  ID:         03
  Usage:      sign
Certificate Object, type = X.509 cert
  label:      Signature
  ID:         03
Public Key Object; RSA 1024 bits
  label:      Signature
  ID:         03
  Usage:      encrypt, verify
$ beid-pkcs11-tool --list-mechanisms
Supported mechanisms:
  SHA-1, digest
  MD5, digest
  RIPEMD160, digest
  RSA-PKCS, sign, verify, unwrap
  SHA1-RSA-PKCS, sign, verify
  MD5-RSA-PKCS, sign, verify
  RIPEMD160-RSA-PKCS, sign, verify
$ beid-pkcs11-tool --test
C_SeedRandom() and C_GenerateRandom():
  seems to be OK
Digests:
  all 4 digest functions seem to work
  MD5: OK
  SHA-1: OK
  RIPEMD160: OK
Signatures (currently only RSA signatures)
  testing key 0 (Authentication) 
QSettings: failed to open file '/etc/qt3/qt_plugins_3.3rc'
  all 4 signature functions seem to work
  testing signature mechanisms:
    RSA-PKCS: OK
    SHA1-RSA-PKCS: OK
    MD5-RSA-PKCS: OK
    RIPEMD160-RSA-PKCS: OK
  testing key 1 (Signature) with 1 signature mechanism
    RSA-PKCS: OK
Verify (currently only for RSA):
  testing key 0 (Authentication)
    RSA-PKCS: OK
    SHA1-RSA-PKCS: OK
    MD5-RSA-PKCS: OK
    RIPEMD160-RSA-PKCS: OK
  testing key 1 (Signature) with 1 mechanism
    RSA-PKCS: OK
Key unwrap (RSA)
  testing key 0 (Authentication)  -- can't be used to unwrap, skipping
  testing key 1 (Signature)  -- can't be used to unwrap, skipping
[...]

libbeidpkcs11.so

It's a PKCS#11 library which can be used by Firefox/Iceweasel, Thunderbird/Icedove, Iceape, OpenOffice,...
See below for some tests with those applications.

Firefox security module

To add the security module to Firefox:

apt-get install libbeid2-dev libbeidlibopensc2-dev

Visit file:///usr/share/beid/beid-pkcs11-register.html to install the service

Now what?...
cf http://eid.belgium.be/fr_BE/fed_ict/imported_content_eid/pdf/eID-FR-Firefox.pdf
You can see your certificate in Preferences -> Advanced -> Encryption -> View Certificates and you can trust the Belgium Root CA under the "Authorities" tab for e.g. "identifying mail users"

You can then connect to federal sites like Tax-on-web or the RRN, being identified by your card & PIN.
Note that it works only if I start Firefox after having the eID in place in the reader.
Note that I didn't get much further, being redirected to e.g. this nonexistent page but the title speaks for itself ;-) https://mondossier.rrn.fgov.be/CertificateRevoked.html

Thunderbird security module

To add the security module to Firefox:

apt-get install libbeid2-dev libbeidlibopensc2-dev

Menu preferences->advanced->certificates->security devices->load

Module name: Belgium Identity Card PKCS#11
Module filename: /usr/lib/libbeidpkcs11.so

You can see your certificate in Preferences -> Advanced -> Encryption -> View Certificates and you can trust the Belgium Root CA under the "Authorities" tab for e.g. "identifying mail users"

Try to sign a first mail:
Menu S-MIME -> Digitally sign this message -> setup certificate -> digital signing -> select your BELPIC auth certif

I could successfully sign (with my PIN) and verify an email but only with the Authentication certificate, not the Signature certificate
According to the snapshots of the official guide of the eID for Outlook, it's ok, the Authentication certificate must be used, the other being reserved for legal signatures.
UPDATE: Well now that I saw that Wouter could sign with the signature certificate I tried again and indeed it works.
TODO: I still would like to understand what went wrong before, why only the "Authentication" certificate worked and not the "Signature" one.

One difficulty is that the certificate is not bound to an email address so the email client tells you sth like it's validly signed but no idea if the certificate owner corresponds to the sender email address.

Signing in OpenOffice

It is using the same certificate set as firefox/iceweasel so signing in OpenOffice works out-of-the-box on my Debian.
If not you can still check this article to debug your situation.

File -> Digital Signatures... -> Add...

This works also with the legal "signature" certificate

Linux: OpenSC Middleware

Installation

belpic, the Belgian middleware, is a modified version of OpenSC, let's try the plain OpenSC:

apt-get install opensc

=> file:///usr/share/doc/opensc/BelgianEid.html
OpenSC 0.10.* will include support for the Belgian eID card, except for legally binding signatures (with the so-called Signature key) as this requires a GUI, which is not yet available/implemented. Till that new release please use the "belpic" software available from the belgian state.

Note that you've to stop the filter daemon (beidpcscd) first

cardos-info

Returns the ATR

$ cardos-info 
3b:98:13:40:0a:a5:03:01:01:01:ad:13:11
Received (SW1=0x6D, SW2=0x00)

opensc-tool

$ opensc-tool -a -v # with debug=1 in /etc/opensc/opensc.conf
[opensc-tool] ctx.c:705:sc_context_create: ===================================
[opensc-tool] ctx.c:706:sc_context_create: opensc version: 0.11.4
[opensc-tool] sc.c:196:sc_detect_card_presence: called
[opensc-tool] sc.c:201:sc_detect_card_presence: returning with: 1
Connecting to card in reader iDream ID-SMID01 00 00...
[opensc-tool] card.c:110:sc_connect_card: called
[opensc-tool] reader-pcsc.c:542:pcsc_connect: After connect protocol = 1
[opensc-tool] reader-pcsc.c:561:pcsc_connect: Requesting reader features ... 
[opensc-tool] card-belpic.c:988:belpic_init: Belpic V1.4
[opensc-tool] card-belpic.c:995:belpic_init: 
[opensc-tool] card.c:221:sc_connect_card: card info: Belpic cards, 12002, 0x0
[opensc-tool] card.c:222:sc_connect_card: returning with: 0
Using card driver Belpic cards.
Card ATR:
3B 98 13 40 0A A5 03 01 01 01 AD 13 11 ;..@.........
[opensc-tool] card.c:236:sc_disconnect_card: called
[opensc-tool] card.c:251:sc_disconnect_card: returning with: 0
[opensc-tool] ctx.c:738:sc_release_context: called

From the ATR:

  • Component code: A5
  • OS number: 03
  • OS version: 01
  • Softmask number: 01
  • Softmask version: 01
  • Applet version: 1.1
$ opensc-tool -n      
Belpic cards
$ opensc-tool -f
3f00 type:  DF, size: 65535
select[N/A] lock[N/A] delete[N/A] create[N/A] rehab[N/A] inval[N/A] list[N/A] 
[opensc-tool] card.c:343:sc_list_files: returning with: Not supported
sc_list_files() failed: Not supported

opensc-explorer

$ opensc-explorer

#PUK: (max trials: 3, length: 4-12)
OpenSC [3F00]> verify CHV1 31:31:31:31
[opensc-explorer] sec.c:201:sc_pin_cmd: returning with: PIN code or key incorrect
Incorrect code, 2 tries left.

OpenSC [3F00]> verify CHV1 31:32:33:34
Code correct.

#PUK: (max trials: 12, length: 12)
OpenSC [3F00]> verify CHV3 31:32:33:34
[opensc-explorer] sec.c:201:sc_pin_cmd: returning with: PIN code or key incorrect
Incorrect code, 11 tries left.

#PINreset (max trials: 10, length: 12)
OpenSC [3F00]> verify CHV2 31:32:33:34
[opensc-explorer] sec.c:201:sc_pin_cmd: returning with: PIN code or key incorrect
Incorrect code, 9 tries left.

OpenSC [3F00]> random 100
00000000: 80 8E DD 53 92 0A FB 12 17 7E 77 49 11 D5 3E 93 ...S.....~wI..>.
00000010: E7 93 CD C1 D8 AB E2 0E 85 34 44 F0 B2 F4 52 8A .........4D...R.
00000020: FD 0A 34 8F A1 16 2C 91 85 18 77 83 F4 EC 2F DB ..4...,...w.../.
00000030: 5D 5A A6 F8 4C 61 21 74 B1 C0 E2 4C FF 7B CF BF ]Z..La!t...L.{..
00000040: 01 A2 06 CB 41 33 EB 75 2E 86 90 A7 E6 FD 0C 8C ....A3.u........
00000050: BF 12 CD CE 32 EB 40 89 D7 98 39 78 30 86 AF 52 ....2.@...9x0..R
00000060: 60 E0 F6 C3                                     `...

OpenSC [3F00]> cd df00

OpenSC [3F00/DF00]> get 5035
Total of 119 bytes read from 5035 and saved to 3F00_DF00_5035.

You can also cat the files:

OpenSC [3F00]> cd df01
OpenSC [3F00/DF01]> cd 4033
OpenSC [3F00/DF01/4033]> cat
00000000: 01 1E 41 76 65 6E 75 65 20 64 65 20 6C 61 20 43 ..Avenue de la C
00000010: 6F 75 72 6F 6E 6E 65 20 34 31 20 2F 62 30 32 37 ouronne 41 /b027
00000020: 02 04 31 30 35 30 03 07 49 78 65 6C 6C 65 73 00 ..1050..Ixelles.
00000030: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
00000040: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
00000050: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
00000060: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 ................
00000070: 00 00 00 00 00                                  .....

The interpretation of the file contents we just extracted can be found in the Belgian Electronic Identity Card content document
Here are all the files you can extract:

3F00_2F00      MF/DIR
3F00_DF00_5031 MF/Belpic/ODF                 (Object Directory File)
3F00_DF00_5032 MF/Belpic/TokenInfo
3F00_DF00_5034 MF/Belpic/AODF                (Authentication Object Directory File)
3F00_DF00_5035 MF/Belpic/PrKDF               (Private Key Directory File)
3F00_DF00_5037 MF/Belpic/CDF                 (Certificate Directory File)
3F00_DF00_5038 MF/Belpic/Cert#2              (auth)
3F00_DF00_5039 MF/Belpic/Cert#3              (non-rep)
3F00_DF00_503A MF/Belpic/Cert#4              (CA)
3F00_DF00_503B MF/Belpic/Cert#6              (Root)
3F00_DF00_503C MF/Belpic/Cert#8              (RRN)
3F00_DF01_4031 MF/ID/ID#RN                   (contains also hash of ID#Photo)
3F00_DF01_4032 MF/ID/SGN#RN                  (signature of ID#RN by RRN)
3F00_DF01_4033 MF/ID/ID#Address
3F00_DF01_4034 MF/ID/SGN#Address             (signature of ID#Address|SGN#RN by RRN)
3F00_DF01_4035 MF/ID/ID#Photo                (140x200 JPEG grayscale)
3F00_DF01_4038 MF/ID/PuK#7 ID                (CA role Hash SHA-1)
3F00_DF01_4039 MF/ID/Preferences

Note that here we could extract the RRN Cert#8, which was not shown by the usual pkcs#15 tools...
Note that the Preferences file is 100-byte zeroes, is customisable just with the cardholder PIN but the update_binary command is not supported, so I wrote a patch

$ xview 3F00_DF01_4035 
3F00_DF01_4035 is a 140x200 JPEG image, color space Grayscale, 1 comp, Huffman coding.
...

So to extract the picture via a simple script:

echo -e "cd df01\nget 4035 mypic.jpg"|opensc-explorer

pkcs11-tool

Differences with beid-pkcs11-tool are highlighted between *stars*

$ pkcs11-tool --list-slots
Available slots:
Slot 0           iDream ID-SMID01 00 00
  token label:   BELPIC (Basic PIN)
 *token manuf:   (unknown)*
  token model:   PKCS #15 SCard
  token flags:   rng, *login required*, PIN initialized, token initialized
 *serial num  :  6CFF252C5F190218*
$ pkcs11-tool --list-objects
$ pkcs11-tool --login --list-objects
Please enter User PIN: 
Private Key Object; RSA 
  label:      Authentication
  ID:         02
  Usage:      sign
Certificate Object, type = X.509 cert
  label:      Authentication
  ID:         02
Public Key Object; RSA 1024 bits
  label:      Authentication
  ID:         02
  Usage:      encrypt, verify
Private Key Object; RSA 
  label:      Signature
  ID:         03
  Usage:      sign
Certificate Object, type = X.509 cert
  label:      Signature
  ID:         03
Public Key Object; RSA 1024 bits
  label:      Signature
  ID:         03
  Usage:      encrypt, verify
Certificate Object, type = X.509 cert
  label:      CA
  ID:         00
Public Key Object; RSA 2048 bits
  label:      CA
  ID:         04
  Usage:      encrypt, verify
Certificate Object, type = X.509 cert
  label:      Root
  ID:         00
Public Key Object; RSA 2048 bits
  label:      Root
  ID:         06
  Usage:      encrypt, verify

Strange, I need to login to extract the objects
Strange, pubkeys can encrypt but privkey cannot decrypt...
Strange, both RootCA and CitizenCA certificates have the same id 0
And what's the format of those certificates when dumped out? Not DER neither PEM

$ pkcs11-tool --login --read-object --id 0 --type cert|xxd
[...]
$ pkcs11-tool --login --read-object --id 2 --type pubkey |xxd
[...]
$ pkcs11-tool --list-mechanisms
Supported mechanisms:
  SHA-1, digest
  *SHA256, digest*
  *SHA384, digest*
  *SHA512, digest*
  MD5, digest
  RIPEMD160, digest
  RSA-PKCS, sign, verify, unwrap, *decrypt*
  SHA1-RSA-PKCS, sign, verify
  MD5-RSA-PKCS, sign, verify
  RIPEMD160-RSA-PKCS, sign, verify
  *RSA-PKCS-KEY-PAIR-GEN, keypairgen*

Extra mechanisms available apparently...
And they work:

phil@mercure:~$ echo -n test|pkcs11-tool --hash --mechanism SHA512 |xxd 
0000000: ee26 b0dd 4af7 e749 aa1a 8ee3 c10a e992  .&..J..I........
0000010: 3f61 8980 772e 473f 8819 a5d4 940e 0db2  ?a..w.G?........
0000020: 7ac1 85f8 a0e1 d5f8 4f88 bc88 7fd6 7b14  z.......O.....{.
0000030: 3732 c304 cc5f a9ad 8e6f 57f5 0028 a8ff  72..._...oW..(..
$ pkcs11-tool --login --test           
Please enter User PIN: 
C_SeedRandom() and C_GenerateRandom():
  not implemented
Digests:
  all 4 digest functions seem to work
  MD5: OK
  SHA-1: OK
  RIPEMD160: OK
Signatures (currently only RSA signatures)
  testing key 0 (Authentication) 
  all 4 signature functions seem to work
  testing signature mechanisms:
    RSA-PKCS: OK
    SHA1-RSA-PKCS: OK
    MD5-RSA-PKCS: OK
    RIPEMD160-RSA-PKCS: OK
  testing key 1 (1024 bits, label=Signature) with 1 signature mechanism
[opensc-pkcs11] sec.c:67:sc_set_security_env: returning with: Not supported
error: PKCS11 function C_Sign failed: rv = CKR_FUNCTION_NOT_SUPPORTED (0x54)

pkcs15-tool

$ pkcs15-tool --dump
PKCS#15 Card [BELPIC]:
        Version        : 1
        Serial number  : 1234567890ABCDEF1234567890ABCDEF
        Manufacturer ID: (unknown)
        Flags          : PRN generation, EID compliant

PIN [Basic PIN]
        Com. Flags: 0x3
        ID        : 01
        Flags     : [0x30], initialized, needs-padding
        Length    : min_len:4, max_len:12, stored_len:8
        Pad char  : 0xFF
        Reference : 1
        Type      : bcd
        Path      : 3f00

Private RSA Key [Authentication]
        Com. Flags  : 3
        Usage       : [0x4], sign
        Access Flags: [0x1D], sensitive, alwaysSensitive, neverExtract, local
        ModLength   : 1024
        Key ref     : 130
        Native      : yes
        Path        : 3f00df00
        Auth ID     : 01
        ID          : 02

Private RSA Key [Signature]
        Com. Flags  : 3
        Usage       : [0x200], nonRepudiation
        Access Flags: [0x1D], sensitive, alwaysSensitive, neverExtract, local
        ModLength   : 1024
        Key ref     : 131
        Native      : yes
        Path        : 3f00df00
        Auth ID     : 01
        ID          : 03

X.509 Certificate [Authentication]
        Flags    : 3
        Authority: no
        Path     : 3f00df005038
        ID       : 02

X.509 Certificate [Signature]
        Flags    : 3
        Authority: no
        Path     : 3f00df005039
        ID       : 03

X.509 Certificate [CA]
        Flags    : 3
        Authority: yes
        Path     : 3f00df00503a
        ID       : 04

X.509 Certificate [Root]
        Flags    : 3
        Authority: yes
        Path     : 3f00df00503b
        ID       : 06
pkcs15-tool --read-certificate 02 > my_auth.crt
pkcs15-tool --read-certificate 03 > my_sign.crt
pkcs15-tool --read-certificate 04 > belgium.crt
pkcs15-tool --read-certificate 06 >> belgium.crt
openssl x509 -in my_auth.crt -text
pkcs15-tool --read-ssh-key 2

pkcs15-crypt

From https://securehomes.esat.kuleuven.be/~decockd/wiki/bin/view.cgi/Main/UsingOpenscTOC#using_opensc_smartcards_to_sign

Signing text and extracting the public certificate:

fortune > data.txt
openssl sha1 -binary data.txt > data.sha1
pkcs15-crypt --key 2 --sign --pkcs1 --sha-1 --input data.sha1 --output data.auth.sig
pkcs15-tool --read-certificate 02 > my_auth.crt

Verifying the signature:

openssl x509 -in my_auth.crt -pubkey -noout > my_auth.pem
openssl dgst -sha1 -verify my_auth.pem -signature data.auth.sig data.txt

I tried to do the same with the signature certificate instead of the authentication certificate but I get an error:

pkcs15-crypt --key 3 --sign --pkcs1 --sha-1 --input data.sha1 --output data.auth.sig
[pkcs15-crypt] sec.c:67:sc_set_security_env: returning with: Not supported
[pkcs15-crypt] pkcs15-sec.c:267:sc_pkcs15_compute_signature: sc_set_security_env() failed: Not supported
Compute signature failed: Not supported

Indeed signing with the signature certificate and without a GUI showing a warning about the legal implication is forbidden.

eidenv

Very interesting one...

$ eidenv | recode UTF8..
BELPIC_CARDNUMBER: 123456789012
BELPIC_CHIPNUMBER: 1234567890ABCDEF1234567890ABCDEF
BELPIC_VALIDFROM: 20.06.2005
BELPIC_VALIDTILL: 20.06.2010
BELPIC_DELIVERINGMUNICIPALITY: Liege
BELPIC_NATIONALNUMBER: 00310100123
BELPIC_NAME: Teuwen
BELPIC_FIRSTNAMES: Philippe Yvon
BELPIC_INITIAL: F
BELPIC_NATIONALITY: Belge
BELPIC_BIRTHLOCATION: Liège
BELPIC_BIRTHDATE: 31 JAN  1900 (or 2000? ;-)
BELPIC_SEX: M
BELPIC_NOBLECONDITION: 
BELPIC_DOCUMENTTYPE: 1
BELPIC_SPECIALSTATUS: 0
BELPIC_STREETANDNUMBER: Rue de l'OpenSource 12 /b012
BELPIC_ZIPCODE: 1050
BELPIC_MUNICIPALITY: Ixelles
$ eidenv --exec /bin/bash
$ echo $BELPIC_NAME
Teuwen
$ exit

But if the filter daemon beidpcscd is running:

$ eidenv 
[eidenv] reader-pcsc.c:534:pcsc_connect: SCardConnect failed: Sharing violation.
[eidenv] card.c:228:sc_connect_card: returning with: Generic reader error
Failed to connect to card: Generic reader error

I expected to get prompted by the filter but nothing like that

patch

The middleware is missing the function update_binary() while the card supports it and provides a writable file EF(Preferences) for the cardholder (you need first to login with your PIN)
So I added it and submitted it in bugreport #470637
Demo: how to install Linux on the eID (hum, so to speak...)

$ opensc-explorer
OpenSC Explorer version 0.11.4
OpenSC [3F00]> cd df01
OpenSC [3F00/DF01]> verify CHV1 31:32:33:34
Code correct.
OpenSC [3F00/DF01]> put 4039 tux.txt
Total of 100 bytes written.
$ opensc-explorer
OpenSC Explorer version 0.11.4
OpenSC [3F00]> cd df01
OpenSC [3F00/DF01]> cd 4039
OpenSC [3F00/DF01/4039]> cat
000000: 5B 47 65 6E 5D 0A 4C 47 3D 66 72 00 5F 20 20 20 [Gen].LG=fr._
000010: 00 00 00 00 00 00 00 00 20 20 20 2E 20 2E 20 20 ........   . .
000020: 00 00 00 00 00 00 00 00 20 20 20 2F 56 5C 20 20 ........   /V\
000030: 00 00 00 00 00 00 00 00 20 20 2F 2F 20 5C 5C 20 ........  // \\
000040: 00 00 00 00 00 00 00 00 20 2F 28 20 20 20 29 5C ........ /(   )\
000050: 00 00 00 00 00 00 00 00 20 20 5E 60 7E 27 5E 20 ........  ^`~'^
000060: 00 00 00 00                                     ....

Linux: to be sorted...

Signing with GpgSM

GpgSM is to X.509 what GnuPG is to OpenPGP, cf http://gnupg.org/aegypten/tech.en.html

apt-get install gpgsm dirmngr gnupg-agent pinentry-qt
~/.gnupg/gpg-agent.conf:
no-grab
default-cache-ttl 1800
ignore-cache-for-signing
allow-mark-trusted
~/.bash_profile: (appending this stuff)
# preparing gpg-agent:
if test -f $HOME/.gpg-agent-info && kill -0 `cut -d: -f 2 $HOME/.gpg-agent-info` 2>/dev/null; then
  GPG_AGENT_INFO=`cat $HOME/.gpg-agent-info`
  export GPG_AGENT_INFO
else
  eval `gpg-agent --daemon`
  echo $GPG_AGENT_INFO >$HOME/.gpg-agent-info
fi
~/.gnupg/scdaemon.conf: (we disable internal CCID support as only libccid supports more or less my crappy reader)
disable-ccid
debug-level none
~/.gnupg/gpgsm.conf:
debug-level none

Acquiring the certificates:

$ gpgsm --learn-card

Actually I had to run it several times, the first time only the Belgium CA was extracted, then the Citizen CA and finally the 2 personal certificates. And the behavior is not really reproductible so you've to run it till you've the 4 certificates:

$ gpgsm --list-keys
/home/phil/.gnupg/pubring.kbx
-----------------------------
     Subject: /CN=Belgium Root CA/C=BE
[...] 
     Subject: /CN=Citizen CA/C=BE/SerialNumber=200507
[...] 
     Subject: /CN=Philippe Teuwen (Authentication)/C=BE/SerialNumber=...
[...] 
     Subject: /CN=Philippe Teuwen (Signature)/C=BE/SerialNumber=...

To sign sth:

$ gpgsm --sign mail.txt

Then I get prompted to trust Belgium CA and gpgsm fails "error creating signature: Certificat révoqué <GpgSM>", normal.
During trusting the Belgium CA, it created automatically a .gnupg/trustlist.txt with

# CN=Belgium Root CA,C=BE
DF:DF:AC:89:47:BD:F7:52:64:A9:23:3A:C1:0E:E3:D1:28:33:DA:CC S

Ok let's try again without the CRLs check:

$ gpgsm --disable-crl-checks --armor --sign --output mail.txt.smime mail.txt
[...]
gpgsm: signature created

I was prompted for my PIN during the process.

And trying to verify, with CRLs:

$ gpgsm --verify --output mail.txt mail.txt.smime
gpgsm: Signature made 2008-02-06 21:42:40 using certificate ID 0x80211056
gpgsm: note: non-critical certificate policy not allowed
dirmngr[8994]: error opening `/home/phil/.gnupg/dirmngr_ldapservers.conf': Aucun fichier ou répertoire de ce type
dirmngr[8994]: permanently loaded certificates: 0
dirmngr[8994]:     runtime cached certificates: 0
dirmngr[8994]: command ISVALID failed: Certificat révoqué
gpgsm: certificate #100000000000E144CBC42E9BB2453EE4/2.5.4.5=#323030353037,CN=Citizen CA,C=BE
gpgsm: certificate has been revoked
gpgsm: invalid certification chain: Certificat révoqué

And without CRLs:

$ gpgsm --disable-crl-checks --verify --output mail.txt mail.txt.smime          
gpgsm: Signature made 2008-02-06 21:42:40 using certificate ID 0x80211056
gpgsm: CRLs not checked due to --disable-crl-checks option
gpgsm: Good signature from "/CN=Philippe Teuwen (Authentication)/C=BE/SerialNumber=...

e-Signing plugin for Firefox

  • pure curiosity...
  • cf http://www.certipost.be/dpsolutions/en/e-signing-faq.html
  • the plugin is signed so you've to install the CA certificate of Certipost first and, cf above in my "funny facts", for Firefox, you've to download and... trust it.
    The certificate is apparently no longer (broken link) and I could not find it in their search tool, the only thing I could find is a copy of the pem version in Google cache(!)
    but that was good enough to be able to install the plugin.
  • Moreover the form to sign wants to directly execute some jar file but under an SSL connection signed by... the so-hard-to-get Certipost CA certificate, itself signed by GTE CyberTrust Global Root, an authority built-in in the browsers.
  • You've to pay to be able to sign with your eID non-repudiation signature to have a valid Qualified Electronic Signature with long term value
  • There are some explanations of the additional services provided by CertiPost:
    The applied XAdES-X-L is an XML signature format according to the recognized XAdES standard (ETSI TS 101 903 standard) that implements measures to satisfy the legal requirements for advanced electronic signatures as defined in the European Directive (EC 1999/93: European Community (EC) DIRECTIVE 1999/93/EC OF THE EUROPEAN PARLIAMENT AND COUNCIL ON A COMMUNITY FRAMEWORK FOR ELECTRONIC SIGNATURES) and for long term non-repudiation. Next to the electronic signature itself, the XAdES-X-Lfile contains the certificate that was used during the signature, the information to proof whether the certificate was valid when signing, and a time stamp to proof that all information used for signing existed and not altered since.
    Moreover CertiPost keeps a copy of the signed information.
  • Timestamps & storage look like additional features, not mandated by the law, so can I create a Qualified Electronic Signature with long term value out of the CertiPost context, so just a plain x509 signature?

SSH

cf OpenSSH#Patch_for_login_with_eID

OpenID

cf OpenID and OpenID-eID where I'm hacking phpMyID to make my own

Cryptonit

From News: OPENTRUST has announced the availability of a new version of Cryptonit. This latest release is fully compatible with the Belgium electronic ID card which when used with Cryptonit enables documents to be digitally signed.
download

Under Debian:

apt-get install cryptonit
Device->Load-> /usr/lib/libbeidpkcs11.so

It works, you can sign files with both certificates.
Note that the first PIN prompt doesn't matter, you'll get prompted directly by the libbeidpkcs11 middleware.
Note that for repetitive Authentication signings (even after close/reopen cryptonit) it doesn't prompt me anymore for the next signatures, but well if I want to use the non-repudiation one.

Linux: TODO

TODO: Login

I tried https://securehomes.esat.kuleuven.be/~decockd/wiki/bin/view.cgi/Main/UsingOpenscTOC#logging_in_with_smartcards but with the eID.

apt-get install libpam-p11

See file:///usr/share/doc/libpam-p11/QuickStart.html
See also file:///usr/share/doc/opensc/PamModules.html
Bad side: it conflicts with xlockmore :-(

openssh way:
Preparing the account with .ssh/authorized_keys, cf SSH auth on this page
Edit /etc/pam.d/login and add before "@include common-auth" sth like:

auth sufficient pam_p11_openssh.so /usr/lib/opensc-pkcs11.so

/var/log/auth.log tells: no certificates found or

auth sufficient pam_p11_openssh.so /usr/lib/libbeidpkcs11.so

/var/log/auth.log tells: fatal: pkcs11_sign failed
before I was even prompted for my PIN

opensc way: same results

auth sufficient pam_p11_opensc.so /usr/lib/opensc-pkcs11.so
auth sufficient pam_p11_opensc.so /usr/lib/libbeidpkcs11.so

preparing the account:

mkdir ~/.eid
chmod 0755 ~/.eid
pkcs15-tool -r 2 > ~/.eid/authorized_certificates
chmod 0644 ~/.eid/authorized_certificates

So I still couldn't find a way.

TODO: SSL Auth

http://blog.eikke.com/index.php/ikke/2007/10/29/using_your_belgian_eid_for_ssl_authentic

apt-get install libengine-pkcs11-openssl

To generate a request, open a console and launch openssl. Once at the OpenSSL prompt, issue these 2 commands:

engine -t dynamic -pre SO_PATH:/usr/lib/engines/engine_pkcs11.so -pre ID:pkcs11 -pre LIST_ADD:1 -pre LOAD -pre MODULE_PATH:/usr/lib/opensc-pkcs11.so

Adjust paths if necessary, of course. This loads the pkcs11 engine inside OpenSSL.

req -engine pkcs11 -new -days 100 -key id_02 -keyform engine -out myrequest.csr -subj "/C=BE/ST=O-VL/O=My Organisation/CN=My Name/emailAddress=my@email.tld"

Adjust the days, out and subj parameters, at least. The key ID can be found using

pkcs15-tool -c

Use the ID of the Authentication X509 certificate.

See also file:///usr/share/doc/opensc/QuickStart.html

TODO: Apache SSL Reverse Proxy

cf http://www.belgium.be/zip/eid_authentication_proxy_fr.html
and https://securehomes.esat.kuleuven.be/~decockd/wiki/bin/view.cgi/EidForums/ForumEidCardReverseProxy0004

TODO: OpenGPG & x509

Old dream is to sign an OpenPGP key with the eID, but even if technically possible, it probably breaks the validation chain as what Citizen CA signed was the entire certificate, not just the key/uid.
Sth to check: OpenPGP Signatures Incorporating X.509 Certificates
The solution is apparently to extend OpenGPG to allow special signatures on UIDs with sub-packets containing the entire certificate, the UID being build from the DN fields of the certificate (CN, EMAIL,...)
Apparently PGP supports it already?

TODO: OpenVPN Auth

http://christophe.vandeplas.com/2008/02/03/openvpn-belgian-eid
But Debian openvpn 2.1_cr4 doesn't support yet --show-pkcs11-ids
UPDATE openvpn 2.1~rc7-1 is available soon on Debian

TODO: Acrobat Reader

acroread can deals with PKCS#11 modules...

Document -> Security Settings -> Digital IDs -> PKCS#11 -> Attach module -> 

I tried:

/usr/lib/libbeidpkcs11.so         -> fails to load the module
/usr/lib/onepin-opensc-pkcs11.so  -> ok
/usr/lib/opensc-pkcs11.so         -> ok

But with those last 2 modules I don't know how to access the eID.

See also https://securehomes.esat.kuleuven.be/~decockd/wiki/bin/view.cgi/EidForums/ForumEidCards0068?t=2008-04-04T18:04:26Z and try with http://itext.ugent.be/articles/eid-pdf/

TODO: Misc