From ametzler at bebt.de  Tue Sep  2 20:07:51 2014
From: ametzler at bebt.de (Andreas Metzler)
Date: Tue, 2 Sep 2014 20:07:51 +0200
Subject: [gnutls-devel] broken OCSP response parser with some CAs
Message-ID: <20140902180751.GC26285@downhill.g.la>

Hello,

this is <https://bugs.debian.org/759161>. I have not much to add,
except for the fact that it still applies to 3.3.7.

----- Forwarded message from Alessandro Ghedini <ghedo at debian.org> -----
From: Alessandro Ghedini <ghedo at debian.org>
Subject: Bug#759161: libgnutls-deb0-28: broken OCSP response parser with some
	CAs

Package: libgnutls-deb0-28
Version: 3.3.6-2
Severity: normal

Hi,

I've been playing with gnutls OCSP support but I noticed that it fails
to parse many apparently valid OCSP responses.

E.g. using gnutls-cli with the --ocsp option:

    % gnutls-cli --ocsp facebook.com 443
    [...]
    importing response: ASN1 parser: Error in DER parsing.

I noticed that many of the rejected OCSP responses come from either
digicert.com or GlobalSign (e.g. other than facebook.com, try also
cloudflare.com, wikipedia.org, github.com, bitbucket.org, imgur.com,
...). Note that openssl works with them just fine.

Other CAs work fine (e.g. try yahoo.com, namecheap.com,
shipit.ubuntu.com, ...).

Cheers

-- System Information:
Debian Release: jessie/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (1, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 3.14-2-amd64 (SMP w/4 CPU cores)
Locale: LANG=it_IT.UTF-8, LC_CTYPE=it_IT.UTF-8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash

Versions of packages libgnutls-deb0-28 depends on:
ii  libc6              2.19-9
ii  libgmp10           2:6.0.0+dfsg-6
ii  libhogweed2        2.7.1-3
ii  libnettle4         2.7.1-3
ii  libp11-kit0        0.20.3-2
ii  libtasn1-6         4.1-1
ii  multiarch-support  2.19-9
ii  zlib1g             1:1.2.8.dfsg-2

libgnutls-deb0-28 recommends no packages.

Versions of packages libgnutls-deb0-28 suggests:
ii  gnutls-bin  3.3.6-2

----- End forwarded message -----


From jyhjeff at gmail.com  Thu Sep  4 06:31:02 2014
From: jyhjeff at gmail.com (Jeff Lee)
Date: Thu, 4 Sep 2014 12:31:02 +0800
Subject: [gnutls-devel] _gnutls_vasprintf needs to update INOUT variable
	strp prior return
Message-ID: <CAP5ZYKh57_-ejZxAX4v7n8-tj5acoxPCBvn6AYDwHsejkPSarg@mail.gmail.com>

Hi,

As the subject, please review the implementation of _gnutls_vasprintf for
systems without the suport of vasprintf.

Best Regards,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: </pipermail/attachments/20140904/c1dc812f/attachment.html>

From nmav at gnutls.org  Thu Sep  4 10:29:20 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 4 Sep 2014 10:29:20 +0200
Subject: [gnutls-devel] broken OCSP response parser with some CAs
In-Reply-To: <20140902180751.GC26285@downhill.g.la>
References: <20140902180751.GC26285@downhill.g.la>
Message-ID: <CAJU7za+9xw2Gp-0rHfKeVUQ4x+beALQDc=gN-mikAjdH2QWLzw@mail.gmail.com>

On Tue, Sep 2, 2014 at 8:07 PM, Andreas Metzler <ametzler at bebt.de> wrote:
> Hello,
>
> this is <https://bugs.debian.org/759161>. I have not much to add,
> except for the fact that it still applies to 3.3.7.

Thanks for reporting that. It seems like a regression in libtasn1 3.6.
That should be fixed in
http://git.savannah.gnu.org/cgit/libtasn1.git/commit/?id=9e832a6d5072263a1fea554c654e29ccf918e059

regards,
Nikos


From nmav at gnutls.org  Thu Sep  4 14:00:40 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 4 Sep 2014 14:00:40 +0200
Subject: [gnutls-devel] _gnutls_vasprintf needs to update INOUT variable
 strp prior return
In-Reply-To: <CAP5ZYKh57_-ejZxAX4v7n8-tj5acoxPCBvn6AYDwHsejkPSarg@mail.gmail.com>
References: <CAP5ZYKh57_-ejZxAX4v7n8-tj5acoxPCBvn6AYDwHsejkPSarg@mail.gmail.com>
Message-ID: <CAJU7zaJof3XX2jwFQMcEc+M-BjDgBSaZqi7e_tWGx7kDwVz0Pg@mail.gmail.com>

On Thu, Sep 4, 2014 at 6:31 AM, Jeff Lee <jyhjeff at gmail.com> wrote:
> Hi,
> As the subject, please review the implementation of _gnutls_vasprintf for
> systems without the suport of vasprintf.

Thanks. Updated.

regards,
Nikos


From ametzler at bebt.de  Thu Sep  4 19:24:34 2014
From: ametzler at bebt.de (Andreas Metzler)
Date: Thu, 4 Sep 2014 19:24:34 +0200
Subject: [gnutls-devel] broken OCSP response parser with some CAs
In-Reply-To: <CAJU7za+9xw2Gp-0rHfKeVUQ4x+beALQDc=gN-mikAjdH2QWLzw@mail.gmail.com>
References: <20140902180751.GC26285@downhill.g.la>
 <CAJU7za+9xw2Gp-0rHfKeVUQ4x+beALQDc=gN-mikAjdH2QWLzw@mail.gmail.com>
Message-ID: <20140904172434.GB5602@downhill.g.la>

On 2014-09-04 Nikos Mavrogiannopoulos <nmav at gnutls.org> wrote:
> On Tue, Sep 2, 2014 at 8:07 PM, Andreas Metzler <ametzler at bebt.de> wrote:
> > Hello,
> >
> > this is <https://bugs.debian.org/759161>. I have not much to add,
> > except for the fact that it still applies to 3.3.7.

> Thanks for reporting that. It seems like a regression in libtasn1 3.6.
> That should be fixed in
> http://git.savannah.gnu.org/cgit/libtasn1.git/commit/?id=9e832a6d5072263a1fea554c654e29ccf918e059

Also works for me. Thank you!

cu Andreas
-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'


From dbaryshkov at gmail.com  Sat Sep  6 12:40:22 2014
From: dbaryshkov at gmail.com (Dmitry Eremin-Solenikov)
Date: Sat, 6 Sep 2014 14:40:22 +0400
Subject: [gnutls-devel] GnuTLS test suite
Message-ID: <CALT56yPYzYt4SqvwOn6LGF55xS2WDhgX3MusZLHM_76=-eBbKA@mail.gmail.com>

Hello,

I have noticed several problems with GnuTLS test suite.

First, mini-record-2 runs indefinitely. Is this an expected failure?

Also I got two valgrind errors from hostname-check and from
nul-in-x509-names tests (logs are in attachment).

-- 
With best wishes
Dmitry
-------------- next part --------------
A non-text attachment was scrubbed...
Name: hostname-check.log
Type: text/x-log
Size: 5509 bytes
Desc: not available
URL: </pipermail/attachments/20140906/314f3412/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: nul-in-x509-names.log
Type: text/x-log
Size: 1107 bytes
Desc: not available
URL: </pipermail/attachments/20140906/314f3412/attachment-0003.bin>

From nmav at gnutls.org  Sat Sep  6 13:36:25 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Sat, 06 Sep 2014 13:36:25 +0200
Subject: [gnutls-devel] GnuTLS test suite
In-Reply-To: <CALT56yPYzYt4SqvwOn6LGF55xS2WDhgX3MusZLHM_76=-eBbKA@mail.gmail.com>
References: <CALT56yPYzYt4SqvwOn6LGF55xS2WDhgX3MusZLHM_76=-eBbKA@mail.gmail.com>
Message-ID: <1410003385.3910.2.camel@nomad.lan>

On Sat, 2014-09-06 at 14:40 +0400, Dmitry Eremin-Solenikov wrote:
> Hello,
> 
> I have noticed several problems with GnuTLS test suite.
> First, mini-record-2 runs indefinitely. Is this an expected failure?

It should take quite long time when run under valgrind, but it shouldn't
be indefinitely.

> Also I got two valgrind errors from hostname-check and from
> nul-in-x509-names tests (logs are in attachment).

These are warnings due to the usage of IDNA. However, they should be
silenced by this line of suppressions.valgrind.

{
   libidn-strlen
   Memcheck:Addr4
   fun:idna_to_ascii_4z  
   fun:idna_to_ascii_8z  
   fun:gnutls_x509_crt_check_hostname2
   fun:doit
   fun:main
}

I'm wondering whether it is the addition crt_check_hostname that causes
that issue. Would adding:
{
   libidn-strlen
   Memcheck:Addr4
   fun:idna_to_ascii_4z  
   fun:idna_to_ascii_8z  
   fun:gnutls_x509_crt_check_hostname2
   fun:gnutls_x509_crt_check_hostname
   fun:doit
   fun:main
}

solve the issue you see?




From INVALID.NOREPLY at gnu.org  Sun Sep  7 12:22:40 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sun, 07 Sep 2014 10:22:40 +0000
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <20140825-203439.sv707.32318@savannah.gnu.org>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
Message-ID: <20140907-132239.sv707.3610@savannah.gnu.org>

Follow-up Comment #2, sr #108634 (project gnutls):

> For example, I want to verify a certificate, and if the 
> certificate is invalid because the issuer is not found (this I 
> can do with gnutls_certificate_verify_peers), I want to 
> re-validate it with different flags to see whether there are 
> other issues with the certificate (for example, expired).

I have misread you message. Which function has that limitation and would you
suggest to address that issue?

> Or after I set the trusted certificates with 
> gnutls_certificate_set_x509_system_trust(), I want to present a 
> list of these certificates in a user interface, so it would be 
> good if I could obtain the list of trusted CAs from the 
> gnutls_certificate_credentials_t structure.

That as I understand would be possible by a function that enumerates all
certificates in a trust list. Is that correct?


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108634>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From INVALID.NOREPLY at gnu.org  Sun Sep  7 22:04:09 2014
From: INVALID.NOREPLY at gnu.org (Armin Burgmeier)
Date: Sun, 07 Sep 2014 20:04:09 +0000
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <20140907-132239.sv707.3610@savannah.gnu.org>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
Message-ID: <20140907-200409.sv96363.42296@savannah.gnu.org>

Follow-up Comment #3, sr #108634 (project gnutls):

>  I have misread you message. Which function has that limitation and would
you suggest to address that issue? 

All the certificate verification functions I think. The limitation is in the
internal function "verify_crt" in x509/verify.c. If, for example, the verify
result is GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE or
GNUTLS_CERT_SIGNER_NOT_FOUND, then the certificate is not checked for other
problems, such as whether it is expired. Other errors, such as
GNUTLS_CERT_SIGNATURE_FAILURE, do not stop the procedure and certificate
checking continues.

Whether this is supposed to be like this or not I don't know, so maybe it is a
different issue from what I mentioned in the caption of this issue.

>  That as I understand would be possible by a function that enumerates all
certificates in a trust list. Is that correct? 

In addition I would need to get the trust list from the
gnutls_certificate_credentials_t structure, I think. It would also be nice to
obtain the private key (gnutls_x509_privkey_t) and certificate(s)
(gnutls_x509_crt_t) from that structure, not only the trust list.

At the moment in my program I am storing gnutls_x509_privkey_t,
gnutls_x509_crt_t (a client certificate), and the
gnutls_certificate_credentials_t itself. I need both: The certificate I need
to extract the common name to show in the user interface, and the credentials
I need to set it with the GnuTLS session (gnutls_credentials_set). This
basically means I need to store the key and certificate twice in my program:
Once in my data own structures, and once in the
gnutls_certificate_credentials_t structure.

What I would like to do is to only keep the gnutls_certificate_credentials_t
structure, and when I need to show the common name in the user interface, I
would get the certificate from the gnutls_certificate_credentials_t, and then
get the name from the certificate. Then I would only end up with the private
key and certificate once in my program's memory.

Is it clear?

If you agree this would be good to have I can try to come up with a patch for
it.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108634>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From nmav at gnu.org  Mon Sep  8 11:23:07 2014
From: nmav at gnu.org (Nikos Mavrogiannopoulos)
Date: Mon, 8 Sep 2014 11:23:07 +0200
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <20140907-200409.sv96363.42296@savannah.gnu.org>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
Message-ID: <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>

On Sun, Sep 7, 2014 at 10:04 PM, Armin Burgmeier
<INVALID.NOREPLY at gnu.org> wrote:
> Follow-up Comment #3, sr #108634 (project gnutls):
>>  I have misread you message. Which function has that limitation and would
> you suggest to address that issue?
> All the certificate verification functions I think. The limitation is in the
> internal function "verify_crt" in x509/verify.c. If, for example, the verify
> result is GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE or
> GNUTLS_CERT_SIGNER_NOT_FOUND, then the certificate is not checked for other
> problems, such as whether it is expired. Other errors, such as
> GNUTLS_CERT_SIGNATURE_FAILURE, do not stop the procedure and certificate
> checking continues.

Indeed, if the constraints fail it would not proceed to the
verification itself. I don't know if that's useful to proceed. If you
could make a description of your use-case it may help to figure out
whether some changes should be done.

>>  That as I understand would be possible by a function that enumerates all
> certificates in a trust list. Is that correct?
> In addition I would need to get the trust list from the
> gnutls_certificate_credentials_t structure, I think. It would also be nice to
> obtain the private key (gnutls_x509_privkey_t) and certificate(s)
> (gnutls_x509_crt_t) from that structure, not only the trust list.

That would be indeed useful. The certificate_credentials_t
functionality typically wraps over the functionality of trusted_list,
so that would be ok.

The data should be extracted from a trust list/credentials however in
a way that don't assume much about the internal structure of it, in
order to allow a revamp without affecting the API.
A way for example to export the trusted CAs from the trusted list
would be using an iterator similarly to
gnutls_x509_crl_iter_crt_serial().
>From the certificate_credentials_t structure you could export the
private key as a gnutls_privkey_t (and assume that this will be the
format used internally by gnutls), but for the certificates, you only
have gnutls_pubkey_t and a datum with the DER data held internally. I
don't know if that would suffice.

> What I would like to do is to only keep the gnutls_certificate_credentials_t
> structure, and when I need to show the common name in the user interface, I
> would get the certificate from the gnutls_certificate_credentials_t, and then
> get the name from the certificate. Then I would only end up with the private
> key and certificate once in my program's memory.

That would be tricky as gnutls doesn't hold internally a
gnutls_x509_crt_t. You could of course add functionality to get the
DER encoding of the certificate from the credentials, decode it and
get an x509_crt_t, but that you can already do with
gnutls_certificate_get_ours(). I'm not sure I have a good suggestion
for that case.

regards,
Nikos


From armin at arbur.net  Mon Sep  8 15:59:56 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Mon, 08 Sep 2014 09:59:56 -0400
Subject: [gnutls-devel] [sr #108634] Getter functions for
 gnutls_certificate_credentials_t
In-Reply-To: <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
Message-ID: <1410184796.1723.20.camel@waverley>

On Mon, 2014-09-08 at 11:23 +0200, Nikos Mavrogiannopoulos wrote:
> On Sun, Sep 7, 2014 at 10:04 PM, Armin Burgmeier
> <INVALID.NOREPLY at gnu.org> wrote:
> > Follow-up Comment #3, sr #108634 (project gnutls):
> >>  I have misread you message. Which function has that limitation and would
> > you suggest to address that issue?
> > All the certificate verification functions I think. The limitation is in the
> > internal function "verify_crt" in x509/verify.c. If, for example, the verify
> > result is GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE or
> > GNUTLS_CERT_SIGNER_NOT_FOUND, then the certificate is not checked for other
> > problems, such as whether it is expired. Other errors, such as
> > GNUTLS_CERT_SIGNATURE_FAILURE, do not stop the procedure and certificate
> > checking continues.
> 
> Indeed, if the constraints fail it would not proceed to the
> verification itself. I don't know if that's useful to proceed. If you
> could make a description of your use-case it may help to figure out
> whether some changes should be done.

My use-case is the following. I'm verifying the certificate of a peer.
If the certificate validates, I want to just go on with the connection.
However, if the signer of the certificate is not in my trust list, I
want to present a dialog to the user asking whether the certificate
should be trusted, for example if it is a self-signed certificate.
However, if the certificate is invalid beyond the signer not trusted
error (signature failure, or expired), then I want to reject the
connection without any user interaction.

At the moment I am achieving this by verifying the certificate twice:
Once "normally", and if that returns GNUTLS_CERT_SIGNER_NOT_FOUND, then
I am verifying it again but this time I add the certificate's issuer (as
sent by the peer) in the trust list artificially, so that the signer is
always found and the certificate is checked for other flaws.

In the end this is a bit like SSH-like trust-on-first-use... I
understand there is actually a API in gnutls for something very similar,
but from what I have seen this is on top of the "issuer trusted" check,
and not replacing it.

> >>  That as I understand would be possible by a function that enumerates all
> > certificates in a trust list. Is that correct?
> > In addition I would need to get the trust list from the
> > gnutls_certificate_credentials_t structure, I think. It would also be nice to
> > obtain the private key (gnutls_x509_privkey_t) and certificate(s)
> > (gnutls_x509_crt_t) from that structure, not only the trust list.
> 
> That would be indeed useful. The certificate_credentials_t
> functionality typically wraps over the functionality of trusted_list,
> so that would be ok.
> 
> The data should be extracted from a trust list/credentials however in
> a way that don't assume much about the internal structure of it, in
> order to allow a revamp without affecting the API.
> A way for example to export the trusted CAs from the trusted list
> would be using an iterator similarly to
> gnutls_x509_crl_iter_crt_serial().

Okay, this sounds reasonable.

> >From the certificate_credentials_t structure you could export the
> private key as a gnutls_privkey_t (and assume that this will be the
> format used internally by gnutls), but for the certificates, you only
> have gnutls_pubkey_t and a datum with the DER data held internally. I
> don't know if that would suffice.
> 
> > What I would like to do is to only keep the gnutls_certificate_credentials_t
> > structure, and when I need to show the common name in the user interface, I
> > would get the certificate from the gnutls_certificate_credentials_t, and then
> > get the name from the certificate. Then I would only end up with the private
> > key and certificate once in my program's memory.
> 
> That would be tricky as gnutls doesn't hold internally a
> gnutls_x509_crt_t. You could of course add functionality to get the
> DER encoding of the certificate from the credentials, decode it and
> get an x509_crt_t, but that you can already do with
> gnutls_certificate_get_ours(). I'm not sure I have a good suggestion
> for that case.

For my case, just exporting the DER-encoded data would be fine.
Basically just like gnutls_certificate_get_ours(), but with a
gnutls_certificate_credentials_t instead of a gnutls_session_t.

In terms of API flexibility however, it might make sense to construct a
new gnutls_x509_crt_t and return that, i.e. doing the DER-decoding
inside gnutls, and documenting that the caller should
gnutls_x509_crt_free() the result. In this case, the API would always
return a newly constructed gnutls_x509_crt_t, but it is independent of
the actual storage in the credentials structure, as long as what is
stored can be converted to a gnutls_x509_crt_t.

What do you think?

Cheers,
Armin



From nmav at gnu.org  Wed Sep 10 10:34:17 2014
From: nmav at gnu.org (Nikos Mavrogiannopoulos)
Date: Wed, 10 Sep 2014 10:34:17 +0200
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <1410184796.1723.20.camel@waverley>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
Message-ID: <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>

On Mon, Sep 8, 2014 at 3:59 PM, Armin Burgmeier <armin at arbur.net> wrote:

>> Indeed, if the constraints fail it would not proceed to the
>> verification itself. I don't know if that's useful to proceed. If you
>> could make a description of your use-case it may help to figure out
>> whether some changes should be done.
> My use-case is the following. I'm verifying the certificate of a peer.
> If the certificate validates, I want to just go on with the connection.
> However, if the signer of the certificate is not in my trust list, I
> want to present a dialog to the user asking whether the certificate
> should be trusted, for example if it is a self-signed certificate.
> However, if the certificate is invalid beyond the signer not trusted
> error (signature failure, or expired), then I want to reject the
> connection without any user interaction.
> At the moment I am achieving this by verifying the certificate twice:
> Once "normally", and if that returns GNUTLS_CERT_SIGNER_NOT_FOUND, then
> I am verifying it again but this time I add the certificate's issuer (as
> sent by the peer) in the trust list artificially, so that the signer is
> always found and the certificate is checked for other flaws.

Ok, so I understand you would prefer to get as many as possible flags
from the verification process.
However, I'm not really sure that you add any value from that
additional check. In the end an attacker would simply need to put an
arbitrary end-user certificate in the chain and you'll revert to trust
on first use just the same.

> In the end this is a bit like SSH-like trust-on-first-use... I
> understand there is actually a API in gnutls for something very similar,
> but from what I have seen this is on top of the "issuer trusted" check,
> and not replacing it.

It's up to you how to use that API. It can be used with PKIX
authentication or entirely separate.

>> > What I would like to do is to only keep the gnutls_certificate_credentials_t
>> > structure, and when I need to show the common name in the user interface, I
>> > would get the certificate from the gnutls_certificate_credentials_t, and then
>> > get the name from the certificate. Then I would only end up with the private
>> > key and certificate once in my program's memory.
>>
>> That would be tricky as gnutls doesn't hold internally a
>> gnutls_x509_crt_t. You could of course add functionality to get the
>> DER encoding of the certificate from the credentials, decode it and
>> get an x509_crt_t, but that you can already do with
>> gnutls_certificate_get_ours(). I'm not sure I have a good suggestion
>> for that case.
>
> For my case, just exporting the DER-encoded data would be fine.
> Basically just like gnutls_certificate_get_ours(), but with a
> gnutls_certificate_credentials_t instead of a gnutls_session_t.
>
> In terms of API flexibility however, it might make sense to construct a
> new gnutls_x509_crt_t and return that, i.e. doing the DER-decoding
> inside gnutls, and documenting that the caller should
> gnutls_x509_crt_free() the result. In this case, the API would always
> return a newly constructed gnutls_x509_crt_t, but it is independent of
> the actual storage in the credentials structure, as long as what is
> stored can be converted to a gnutls_x509_crt_t.
> What do you think?

It seems ok, although there will be a discrepancy. The
gnutls_privkey_t will not be released by the user, while the
x509_crt_t should be released. Maybe for consistency it would make
sense to generate it on the fly, cache in the structure, and
deinitialize it when whole structure gets deinitialized. Such an API
should allow flexibility on the exporting type of the certificate as
well, since openpgp or raw public keys should be able to be exported
as well (even if that functionality is not implemented initially).

regards,
Nikos


From armin at arbur.net  Wed Sep 10 17:15:39 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 10 Sep 2014 11:15:39 -0400
Subject: [gnutls-devel] [sr #108634] Getter functions for
 gnutls_certificate_credentials_t
In-Reply-To: <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
 <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
Message-ID: <1410362139.2056.24.camel@waverley>

On Wed, 2014-09-10 at 10:34 +0200, Nikos Mavrogiannopoulos wrote:
> On Mon, Sep 8, 2014 at 3:59 PM, Armin Burgmeier <armin at arbur.net> wrote:
> 
> >> Indeed, if the constraints fail it would not proceed to the
> >> verification itself. I don't know if that's useful to proceed. If you
> >> could make a description of your use-case it may help to figure out
> >> whether some changes should be done.
> > My use-case is the following. I'm verifying the certificate of a peer.
> > If the certificate validates, I want to just go on with the connection.
> > However, if the signer of the certificate is not in my trust list, I
> > want to present a dialog to the user asking whether the certificate
> > should be trusted, for example if it is a self-signed certificate.
> > However, if the certificate is invalid beyond the signer not trusted
> > error (signature failure, or expired), then I want to reject the
> > connection without any user interaction.
> > At the moment I am achieving this by verifying the certificate twice:
> > Once "normally", and if that returns GNUTLS_CERT_SIGNER_NOT_FOUND, then
> > I am verifying it again but this time I add the certificate's issuer (as
> > sent by the peer) in the trust list artificially, so that the signer is
> > always found and the certificate is checked for other flaws.
> 
> Ok, so I understand you would prefer to get as many as possible flags
> from the verification process.
> However, I'm not really sure that you add any value from that
> additional check. In the end an attacker would simply need to put an
> arbitrary end-user certificate in the chain and you'll revert to trust
> on first use just the same.

I agree. If I understand correctly, however, SSH is in exactly the same
situation.

The value from the additional check is that I don't have to bug my users
with an extra dialog when the server has a certificate that is issued by
a trusted CA. Only when that is not the case I resort to
trust-on-first-use. Then yes, an attacker could present an arbitrary
certificate; but if the user has connected to the server before already
it will detect that the certificate is different from the previous
connection attempt.

The whole situation is obviously not bullet-proof; I'm basically aiming
for a compromise between security and convenience -- don't want to force
people to get a CA-issued certificate to use the application but still
provide as much security as possible.

> > In the end this is a bit like SSH-like trust-on-first-use... I
> > understand there is actually a API in gnutls for something very similar,
> > but from what I have seen this is on top of the "issuer trusted" check,
> > and not replacing it.
> 
> It's up to you how to use that API. It can be used with PKIX
> authentication or entirely separate.
> 
> >> > What I would like to do is to only keep the gnutls_certificate_credentials_t
> >> > structure, and when I need to show the common name in the user interface, I
> >> > would get the certificate from the gnutls_certificate_credentials_t, and then
> >> > get the name from the certificate. Then I would only end up with the private
> >> > key and certificate once in my program's memory.
> >>
> >> That would be tricky as gnutls doesn't hold internally a
> >> gnutls_x509_crt_t. You could of course add functionality to get the
> >> DER encoding of the certificate from the credentials, decode it and
> >> get an x509_crt_t, but that you can already do with
> >> gnutls_certificate_get_ours(). I'm not sure I have a good suggestion
> >> for that case.
> >
> > For my case, just exporting the DER-encoded data would be fine.
> > Basically just like gnutls_certificate_get_ours(), but with a
> > gnutls_certificate_credentials_t instead of a gnutls_session_t.
> >
> > In terms of API flexibility however, it might make sense to construct a
> > new gnutls_x509_crt_t and return that, i.e. doing the DER-decoding
> > inside gnutls, and documenting that the caller should
> > gnutls_x509_crt_free() the result. In this case, the API would always
> > return a newly constructed gnutls_x509_crt_t, but it is independent of
> > the actual storage in the credentials structure, as long as what is
> > stored can be converted to a gnutls_x509_crt_t.
> > What do you think?
> 
> It seems ok, although there will be a discrepancy. The
> gnutls_privkey_t will not be released by the user, while the
> x509_crt_t should be released.

We could instead return a copy of the gnutls_privkey_t, so that it has
to be released as well by the user. This would then again give some
flexibility in case the key storage inside the credentials structure
changes at some point in the future.

>  Maybe for consistency it would make
> sense to generate it on the fly, cache in the structure, and
> deinitialize it when whole structure gets deinitialized. Such an API
> should allow flexibility on the exporting type of the certificate as
> well, since openpgp or raw public keys should be able to be exported
> as well (even if that functionality is not implemented initially).

In that case, the keys and certificates would then be stored twice in
memory again (once they have been queried via the new API). That idea is
not very appealing to me, but maybe it's just a matter of taste...
either way should work for what I would need the new API for.

What I don't understand is, how would returning a new
certificate/private key not allow openpgp or raw keys to be exported?

Cheers,
Armin



From nmav at gnu.org  Wed Sep 10 17:50:02 2014
From: nmav at gnu.org (Nikos Mavrogiannopoulos)
Date: Wed, 10 Sep 2014 17:50:02 +0200
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <1410362139.2056.24.camel@waverley>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
 <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
 <1410362139.2056.24.camel@waverley>
Message-ID: <CAJU7zaKbs66rnXGgxHxRUP841U-b9buxO_OHmk43bmg0Oa5+iA@mail.gmail.com>

On Wed, Sep 10, 2014 at 5:15 PM, Armin Burgmeier <armin at arbur.net> wrote:
>> >> Indeed, if the constraints fail it would not proceed to the
>> >> verification itself. I don't know if that's useful to proceed. If you
>> >> could make a description of your use-case it may help to figure out
>> >> whether some changes should be done.
>> > My use-case is the following. I'm verifying the certificate of a peer.
>> > If the certificate validates, I want to just go on with the connection.
>> > However, if the signer of the certificate is not in my trust list, I
>> > want to present a dialog to the user asking whether the certificate
>> > should be trusted, for example if it is a self-signed certificate.
>> > However, if the certificate is invalid beyond the signer not trusted
>> > error (signature failure, or expired), then I want to reject the
>> > connection without any user interaction.
>> > At the moment I am achieving this by verifying the certificate twice:
>> > Once "normally", and if that returns GNUTLS_CERT_SIGNER_NOT_FOUND, then
>> > I am verifying it again but this time I add the certificate's issuer (as
>> > sent by the peer) in the trust list artificially, so that the signer is
>> > always found and the certificate is checked for other flaws.
>> Ok, so I understand you would prefer to get as many as possible flags
>> from the verification process.
>> However, I'm not really sure that you add any value from that
>> additional check. In the end an attacker would simply need to put an
>> arbitrary end-user certificate in the chain and you'll revert to trust
>> on first use just the same.
> The value from the additional check is that I don't have to bug my users
> with an extra dialog when the server has a certificate that is issued by
> a trusted CA. Only when that is not the case I resort to
> trust-on-first-use. Then yes, an attacker could present an arbitrary
> certificate; but if the user has connected to the server before already
> it will detect that the certificate is different from the previous
> connection attempt.

Ok, I understand. I believe a small modification of the verification
functions would allow such usage. I don't have much time for such a
change but that could be added in  3.4.0 todo list.

>> > For my case, just exporting the DER-encoded data would be fine.
>> > Basically just like gnutls_certificate_get_ours(), but with a
>> > gnutls_certificate_credentials_t instead of a gnutls_session_t.
>> >
>> > In terms of API flexibility however, it might make sense to construct a
>> > new gnutls_x509_crt_t and return that, i.e. doing the DER-decoding
>> > inside gnutls, and documenting that the caller should
>> > gnutls_x509_crt_free() the result. In this case, the API would always
>> > return a newly constructed gnutls_x509_crt_t, but it is independent of
>> > the actual storage in the credentials structure, as long as what is
>> > stored can be converted to a gnutls_x509_crt_t.
>> > What do you think?
>>
>> It seems ok, although there will be a discrepancy. The
>> gnutls_privkey_t will not be released by the user, while the
>> x509_crt_t should be released.
>
> We could instead return a copy of the gnutls_privkey_t, so that it has
> to be released as well by the user. This would then again give some
> flexibility in case the key storage inside the credentials structure
> changes at some point in the future.

That would be quite some work if gnutls_privkey_t is of type
gnutls_pkcs11_privkey_t. There is no implementation of the copy
process there yet, but is certainly doable.

>>  Maybe for consistency it would make
>> sense to generate it on the fly, cache in the structure, and
>> deinitialize it when whole structure gets deinitialized. Such an API
>> should allow flexibility on the exporting type of the certificate as
>> well, since openpgp or raw public keys should be able to be exported
>> as well (even if that functionality is not implemented initially).
> In that case, the keys and certificates would then be stored twice in
> memory again (once they have been queried via the new API). That idea is
> not very appealing to me, but maybe it's just a matter of taste...
> either way should work for what I would need the new API for.

I don't like the caching either much, but would go well with the
static privkey_t. If both are allocated indeed that would be better.

> What I don't understand is, how would returning a new
> certificate/private key not allow openpgp or raw keys to be exported?

I'm not sure I understand you. What I meant above, is that these keys
should be able to be exported too.

regards,
Nikos


From nmav at gnutls.org  Thu Sep 11 22:56:14 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 11 Sep 2014 22:56:14 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <20140524065807.GA2250@downhill.g.la>
References: <20140524065807.GA2250@downhill.g.la>
Message-ID: <1410468974.3121.3.camel@nomad.lan>

On Sat, 2014-05-24 at 08:58 +0200, Andreas Metzler wrote:

> Anyway, this causes hard crashes like in
> <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=746641#37> or
> <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=748742#37>.
> 
> Fixing this in gnutls' source is pretty easy: In gnutls.map move the
> contents of GNUTLS_1_4, GNUTLS_2_8, GNUTLS_2_10 and GNUTLS_2_12 to
> GNUTLS_3_0_0. However it breaks the ABI, everything linking against
> gnutls3 will need to be rebuilt after the change. Afaiu a soname bump
> would therefore be the correct thing.

Hi Andreas,
 I get back on that, as I noticed that now applications linked against
gnutls in fedora wouldn't work in debian and vice-versa. While that may
not be an issue in software compiled in debian for debian, it will be
for any other software. For example I use a debian-based docker
container to test the openconnect server, and that fails to execute the
server if compiled in any other system. Closed source programs using
gnutls will have the same issue.

regards,
Nikos




From nmav at gnu.org  Fri Sep 12 08:41:36 2014
From: nmav at gnu.org (Nikos Mavrogiannopoulos)
Date: Fri, 12 Sep 2014 08:41:36 +0200
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <CAJU7zaKbs66rnXGgxHxRUP841U-b9buxO_OHmk43bmg0Oa5+iA@mail.gmail.com>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
 <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
 <1410362139.2056.24.camel@waverley>
 <CAJU7zaKbs66rnXGgxHxRUP841U-b9buxO_OHmk43bmg0Oa5+iA@mail.gmail.com>
Message-ID: <CAJU7zaLXhsgX2ReU3PocpY5H7SH2qd82KwNFY7aT5o_DX687TQ@mail.gmail.com>

On Wed, Sep 10, 2014 at 5:50 PM, Nikos Mavrogiannopoulos <nmav at gnu.org> wrote:
>> The value from the additional check is that I don't have to bug my users
>> with an extra dialog when the server has a certificate that is issued by
>> a trusted CA. Only when that is not the case I resort to
>> trust-on-first-use. Then yes, an attacker could present an arbitrary
>> certificate; but if the user has connected to the server before already
>> it will detect that the certificate is different from the previous
>> connection attempt.
> Ok, I understand. I believe a small modification of the verification
> functions would allow such usage. I don't have much time for such a
> change but that could be added in  3.4.0 todo list.

I should have added here, that if there is a patch, I'd review it of
course (and something like that would speed up the process).

regards,
Nikos


From armin at arbur.net  Fri Sep 12 17:10:21 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Fri, 12 Sep 2014 11:10:21 -0400
Subject: [gnutls-devel] [sr #108634] Getter functions for
 gnutls_certificate_credentials_t
In-Reply-To: <CAJU7zaLXhsgX2ReU3PocpY5H7SH2qd82KwNFY7aT5o_DX687TQ@mail.gmail.com>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
 <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
 <1410362139.2056.24.camel@waverley>
 <CAJU7zaKbs66rnXGgxHxRUP841U-b9buxO_OHmk43bmg0Oa5+iA@mail.gmail.com>
 <CAJU7zaLXhsgX2ReU3PocpY5H7SH2qd82KwNFY7aT5o_DX687TQ@mail.gmail.com>
Message-ID: <1410534621.1339.7.camel@waverley>

On Fri, 2014-09-12 at 08:41 +0200, Nikos Mavrogiannopoulos wrote:
> On Wed, Sep 10, 2014 at 5:50 PM, Nikos Mavrogiannopoulos <nmav at gnu.org> wrote:
> >> The value from the additional check is that I don't have to bug my users
> >> with an extra dialog when the server has a certificate that is issued by
> >> a trusted CA. Only when that is not the case I resort to
> >> trust-on-first-use. Then yes, an attacker could present an arbitrary
> >> certificate; but if the user has connected to the server before already
> >> it will detect that the certificate is different from the previous
> >> connection attempt.
> > Ok, I understand. I believe a small modification of the verification
> > functions would allow such usage. I don't have much time for such a
> > change but that could be added in  3.4.0 todo list.
> 
> I should have added here, that if there is a patch, I'd review it of
> course (and something like that would speed up the process).

Yes, I'll try to come up with a patch.

While trying to build from git master I noticed that in commit 469f8fb
you added pkcs11x.c to lib/Makefile.am, however that file does not exist
in the repository... did you maybe forget to git add it?

Cheers,
Armin



From ametzler at bebt.de  Fri Sep 12 18:22:27 2014
From: ametzler at bebt.de (Andreas Metzler)
Date: Fri, 12 Sep 2014 18:22:27 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <1410468974.3121.3.camel@nomad.lan>
References: <20140524065807.GA2250@downhill.g.la>
 <1410468974.3121.3.camel@nomad.lan>
Message-ID: <20140912162227.GA5709@downhill.g.la>

On 2014-09-11 Nikos Mavrogiannopoulos <nmav at gnutls.org> wrote:
> On Sat, 2014-05-24 at 08:58 +0200, Andreas Metzler wrote:

>> Anyway, this causes hard crashes like in
>> <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=746641#37> or
>> <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=748742#37>.

>> Fixing this in gnutls' source is pretty easy: In gnutls.map move the
>> contents of GNUTLS_1_4, GNUTLS_2_8, GNUTLS_2_10 and GNUTLS_2_12 to
>> GNUTLS_3_0_0. However it breaks the ABI, everything linking against
>> gnutls3 will need to be rebuilt after the change. Afaiu a soname bump
>> would therefore be the correct thing.

> Hi Andreas,
>  I get back on that, as I noticed that now applications linked against
> gnutls in fedora wouldn't work in debian and vice-versa. While that may
> not be an issue in software compiled in debian for debian, it will be
> for any other software. For example I use a debian-based docker
> container to test the openconnect server, and that fails to execute the
> server if compiled in any other system. Closed source programs using
> gnutls will have the same issue.

Hello,

yes, that is exactly as expected. It sucks but afaict it could not be
avoided. Fixing the symbol versioning breaks the ABI.  The arguments
for not breaking the ABI by changing the symbol-versioning in GnuTLS
upstream *now* are are the same ones as in May.

OTOH we in Debian needed to fix the symbol versioning because we
care about partial upgrades.

It will therefore continue to be broken until GnuTLS upstream bumps
the symbol-versioning (and the soname) and Debian picks up this
version. 

cu Andreas
-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'


From armin at arbur.net  Fri Sep 12 20:16:06 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Fri, 12 Sep 2014 14:16:06 -0400
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
	verifying a certificate
Message-ID: <1410545766.1339.15.camel@waverley>

This allows to check for all possible flaws with a certificate chain with a
single call to gnutls_x509_crt_list_verify and friends.

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/x509/verify.c   | 164 ++++++++++++++++++++++++----------------------------
 tests/test-chains.h |  13 +++--
 2 files changed, 82 insertions(+), 95 deletions(-)

diff --git a/lib/x509/verify.c b/lib/x509/verify.c
index 81b9b4d..26dfd19 100644
--- a/lib/x509/verify.c
+++ b/lib/x509/verify.c
@@ -572,7 +572,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	gnutls_datum_t cert_signature = { NULL, 0 };
 	gnutls_x509_crt_t issuer = NULL;
 	int issuer_version, hash_algo;
-	bool result = 0;
+	bool result = 1;
 	const mac_entry_st * me;
 	unsigned int out = 0, usage;
 	int sigalg, ret;
@@ -581,13 +581,12 @@ verify_crt(gnutls_x509_crt_t cert,
 		*output = 0;
 
 	if (*max_path == 0) {
-		out =
+		out |=
 		    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
 		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
 		gnutls_assert();
 		result = 0;
+		/* bail immediately, to avoid inconistency  */
 		goto cleanup;
 	}
 	(*max_path)--;
@@ -599,31 +598,26 @@ verify_crt(gnutls_x509_crt_t cert,
 	 * authorities.
 	 */
 	if (issuer == NULL) {
-		out = GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
+		out |= GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID;
 		gnutls_assert();
 		result = 0;
-		goto cleanup;
 	}
 
 	if (_issuer != NULL)
 		*_issuer = issuer;
 
-	if (nc != NULL) {
+	if (nc != NULL && issuer != NULL) {
 		/* append the issuer's constraints */
 		ret = gnutls_x509_crt_get_name_constraints(issuer, nc, 
 			GNUTLS_NAME_CONSTRAINTS_FLAG_APPEND, NULL);
 		if (ret < 0 && ret != GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE) {
  nc_fail:
-			out =
+			out |=
 			    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
 			    GNUTLS_CERT_INVALID;
-			if (output)
-				*output |= out;
 			gnutls_assert();
 			result = 0;
-			goto cleanup;
+			goto nc_done;
 		}
 
 		/* only check name constraints in server certificates, not CAs */
@@ -659,26 +653,26 @@ verify_crt(gnutls_x509_crt_t cert,
 			}
 		}
 	}
+ nc_done:
+
+	if(issuer != NULL)
+		issuer_version = gnutls_x509_crt_get_version(issuer);
+	else
+		issuer_version = 0;
 
-	issuer_version = gnutls_x509_crt_get_version(issuer);
 	if (issuer_version < 0) {
 		gnutls_assert();
 		result = 0;
-		goto cleanup;
-	}
-
-	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
-	    ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
-	     || issuer_version != 1)) {
+	} else if (issuer != NULL &&
+                   !(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
+	           ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
+	            || issuer_version != 1)) {
 		if (check_if_ca(cert, issuer, max_path, flags) != 1) {
 			gnutls_assert();
-			out =
+			out |=
 			    GNUTLS_CERT_SIGNER_NOT_CA |
 			    GNUTLS_CERT_INVALID;
-			if (output)
-				*output |= out;
 			result = 0;
-			goto cleanup;
 		}
 
 		ret =
@@ -686,13 +680,10 @@ verify_crt(gnutls_x509_crt_t cert,
 		if (ret >= 0) {
 			if (!(usage & GNUTLS_KEY_KEY_CERT_SIGN)) {
 				gnutls_assert();
-				out =
+				out |=
 				    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
 				    | GNUTLS_CERT_INVALID;
-				if (output)
-					*output |= out;
 				result = 0;
-				goto cleanup;
 			}
 		}
 	}
@@ -703,7 +694,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
+		cert_signed_data.data = NULL;
 	}
 
 	ret =
@@ -712,7 +703,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
+		cert_signature.data = NULL;
 	}
 
 	ret =
@@ -721,75 +712,70 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
 	}
 	sigalg = ret;
 
-	if (is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
-		gnutls_assert();
-		out =
-		    GNUTLS_CERT_INSECURE_ALGORITHM |
-		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
-		result = 0;
-		goto cleanup;
-	}
+	if (sigalg >= 0 && 
+	    cert_signed_data.data != NULL &&
+	    cert_signature.data != NULL) {
+		if (issuer != NULL &&
+		    is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
+			gnutls_assert();
+			out |=
+			    GNUTLS_CERT_INSECURE_ALGORITHM |
+			    GNUTLS_CERT_INVALID;
+			result = 0;
+		}
 
-	/* If the certificate is not self signed check if the algorithms
-	 * used are secure. If the certificate is self signed it doesn't
-	 * really matter.
-	 */
-	if (gnutls_sign_is_secure(sigalg) == 0 &&
-	    is_broken_allowed(sigalg, flags) == 0 &&
-	    is_issuer(cert, cert) == 0) {
-		gnutls_assert();
-		out =
-		    GNUTLS_CERT_INSECURE_ALGORITHM |
-		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
-		result = 0;
-		goto cleanup;
-	}
+		/* If the certificate is not self signed check if the algorithms
+		 * used are secure. If the certificate is self signed it doesn't
+		 * really matter.
+		 */
+		if (gnutls_sign_is_secure(sigalg) == 0 &&
+		    is_broken_allowed(sigalg, flags) == 0 &&
+		    is_issuer(cert, cert) == 0) {
+			gnutls_assert();
+			out |=
+			    GNUTLS_CERT_INSECURE_ALGORITHM |
+			    GNUTLS_CERT_INVALID;
+			result = 0;
+		}
 
-	hash_algo = gnutls_sign_get_hash_algorithm(sigalg);
-	me = mac_to_entry(hash_algo);
-	if (me == NULL) {
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
+		hash_algo = gnutls_sign_get_hash_algorithm(sigalg);
+		me = mac_to_entry(hash_algo);
+		if (me == NULL) {
+			gnutls_assert();
+			result = 0;
+		} else if (issuer != NULL) {
+			ret =
+			    _gnutls_x509_verify_data(me,
+						     &cert_signed_data,
+		        	                     &cert_signature,
+						     issuer);
+			if (ret == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
+				gnutls_assert();
+				out |=
+				    GNUTLS_CERT_INVALID |
+				    GNUTLS_CERT_SIGNATURE_FAILURE;
+				/* error. ignore it */
+				result = 0;
+			} else if (ret < 0) {
+				result = 0;
+				gnutls_assert();
+			}
+		}
 	}
 
-	ret =
-	    _gnutls_x509_verify_data(me,
-				     &cert_signed_data, &cert_signature,
-				     issuer);
-	if (ret == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
-		gnutls_assert();
-		out |= GNUTLS_CERT_INVALID | GNUTLS_CERT_SIGNATURE_FAILURE;
-		/* error. ignore it */
-		if (output)
-			*output |= out;
-		result = 0;
-	} else if (ret < 0) {
-		result = 0;
-		gnutls_assert();
-		goto cleanup;
-	} else if (ret == 1)
-		result = 1;
-
 	/* Check activation/expiration times
 	 */
 	if (!(flags & GNUTLS_VERIFY_DISABLE_TIME_CHECKS)) {
 		/* check the time of the issuer first */
-		if (!(flags & GNUTLS_VERIFY_DISABLE_TRUSTED_TIME_CHECKS)) {
+		if (issuer != NULL &&
+		    !(flags & GNUTLS_VERIFY_DISABLE_TRUSTED_TIME_CHECKS)) {
 			out |= check_time_status(issuer, now);
 			if (out != 0) {
 				gnutls_assert();
 				result = 0;
-				if (output)
-					*output |= out;
 			}
 		}
 
@@ -797,12 +783,13 @@ verify_crt(gnutls_x509_crt_t cert,
 		if (out != 0) {
 			gnutls_assert();
 			result = 0;
-			if (output)
-				*output |= out;
 		}
 	}
 
       cleanup:
+	if (output)
+		*output |= out;
+
 	if (func) {
 		if (result == 0) {
 			out |= GNUTLS_CERT_INVALID;
@@ -1414,7 +1401,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 			*verify |=
 			    GNUTLS_CERT_SIGNER_NOT_FOUND |
 			    GNUTLS_CERT_INVALID;
-		return 0;
 	}
 
 	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
@@ -1424,7 +1410,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 				*verify |=
 				    GNUTLS_CERT_SIGNER_NOT_CA |
 				    GNUTLS_CERT_INVALID;
-			return 0;
 		}
 
 		result =
@@ -1436,7 +1421,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 					*verify |=
 					    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
 					    | GNUTLS_CERT_INVALID;
-				return 0;
 			}
 		}
 	}
@@ -1504,10 +1488,10 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 		*verify |= GNUTLS_CERT_REVOCATION_DATA_SUPERSEDED;
 

-      cleanup:
 	if (verify)
 		*verify |= GNUTLS_CERT_INVALID;
 
+      cleanup:
 	_gnutls_free_datum(&crl_signed_data);
 	_gnutls_free_datum(&crl_signature);
 
diff --git a/tests/test-chains.h b/tests/test-chains.h
index 28974e1..ff9086f 100644
--- a/tests/test-chains.h
+++ b/tests/test-chains.h
@@ -1366,9 +1366,11 @@ static struct
 } chains[] =
 {
   { "CVE-2014-0092", cve_2014_0092_check, &cve_2014_0092_check[1],
-    0, GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
   { "CVE-2008-4989", cve_2008_4989_chain, &cve_2008_4989_chain[2],
-    0, GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
   { "amazon.com ok", verisign_com_chain_g5, &verisign_com_chain_g5[4],
     GNUTLS_VERIFY_DISABLE_TIME_CHECKS | GNUTLS_PROFILE_TO_VFLAGS(GNUTLS_PROFILE_LOW),
     0 },
@@ -1404,9 +1406,10 @@ static struct
     GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
     0 },
   { "rsa-md5 fail", mayfirst_chain, &mayfirst_chain[1],
-    0, GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "rsa-md5 not ok", mayfirst_chain, &mayfirst_chain[1],
-    GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD2,
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS | GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD2,
     GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "rsa-md5 not ok2", mayfirst_chain, &mayfirst_chain[1],
     GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD5,
@@ -1449,7 +1452,7 @@ static struct
     GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD5, 0 },
   { "cacertrsamd5 short-cut not ok", cacertrsamd5, &cacertrsamd5[0],
     GNUTLS_VERIFY_DO_NOT_ALLOW_SAME,
-    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
+    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "cacertrsamd5 short-cut ok", cacertrsamd5, &cacertrsamd5[1],
     0, 0 },
   { "ecc cert ok", ecc_cert, &ecc_cert[1], GNUTLS_PROFILE_TO_VFLAGS(GNUTLS_PROFILE_HIGH), 0 },
-- 
2.1.0




From nmav at gnutls.org  Fri Sep 12 20:36:17 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Fri, 12 Sep 2014 20:36:17 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <20140912162227.GA5709@downhill.g.la>
References: <20140524065807.GA2250@downhill.g.la>
 <1410468974.3121.3.camel@nomad.lan> <20140912162227.GA5709@downhill.g.la>
Message-ID: <1410546977.11023.8.camel@nomad.lan>

On Fri, 2014-09-12 at 18:22 +0200, Andreas Metzler wrote:

> > Hi Andreas,
> >  I get back on that, as I noticed that now applications linked against
> > gnutls in fedora wouldn't work in debian and vice-versa. While that may
> > not be an issue in software compiled in debian for debian, it will be
> > for any other software. For example I use a debian-based docker
> > container to test the openconnect server, and that fails to execute the
> > server if compiled in any other system. Closed source programs using
> > gnutls will have the same issue.
> Hello,
> yes, that is exactly as expected. It sucks but afaict it could not be
> avoided. Fixing the symbol versioning breaks the ABI.  The arguments
> for not breaking the ABI by changing the symbol-versioning in GnuTLS
> upstream *now* are are the same ones as in May.
> OTOH we in Debian needed to fix the symbol versioning because we
> care about partial upgrades.
> 
> It will therefore continue to be broken until GnuTLS upstream bumps
> the symbol-versioning (and the soname) and Debian picks up this
> version. 

Indeed, but the ABI in 3.x is fixed since years and will remain fixed
for years too; there is no plan to change it. Why not change the ABI in
the old releases as a work around? Having a different ABI in the latest
release ensures that debian will remain binary incompatible with any
software build with modern gnutls for the foreseeable future.

regards,
Nikos




From armin at arbur.net  Fri Sep 12 20:39:52 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Fri, 12 Sep 2014 14:39:52 -0400
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410545766.1339.15.camel@waverley>
References: <1410545766.1339.15.camel@waverley>
Message-ID: <1410547192.1339.22.camel@waverley>

This is an attempt at a patch for what we discussed in the other email
thread. I have made sure that the tests still pass (with minor
modifications where additional failure reasons are now expected). Since
this is in a quite sensitive part of the code it would be good to have
this reviewed carefully...

> @@ -1504,10 +1488,10 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
>  		*verify |= GNUTLS_CERT_REVOCATION_DATA_SUPERSEDED;
>  
> 
> -      cleanup:
>  	if (verify)
>  		*verify |= GNUTLS_CERT_INVALID;
>  
> +      cleanup:
>  	_gnutls_free_datum(&crl_signed_data);
>  	_gnutls_free_datum(&crl_signature);

I have two comments to this hunk.

 1) the question where to put the cleanup label is basically what should
happen with the *verify output parameter when the function itself does
not return GNUTLS_E_SUCCESS. Probably it does not matter because when
the function does not run successfully, the verification output is of no
good use anyway.

 2) If I understand the code correctly, then the verify result will
*always* have the GNUTLS_CERT_INVALID flag set, even for perfectly valid
CRLs. I have the impression the check should read "if(verify && result !
= 0)" instead of just "if(verify)"... Is there a unit test for this
function?

Cheers,
Armin



From ametzler at bebt.de  Sat Sep 13 08:04:41 2014
From: ametzler at bebt.de (Andreas Metzler)
Date: Sat, 13 Sep 2014 08:04:41 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <1410546977.11023.8.camel@nomad.lan>
References: <20140524065807.GA2250@downhill.g.la>
 <1410468974.3121.3.camel@nomad.lan>
 <20140912162227.GA5709@downhill.g.la>
 <1410546977.11023.8.camel@nomad.lan>
Message-ID: <20140913060441.GA2429@downhill.g.la>

On 2014-09-12 Nikos Mavrogiannopoulos <nmav at gnutls.org> wrote:
> On Fri, 2014-09-12 at 18:22 +0200, Andreas Metzler wrote:
[Debian GnuTLS v3 package use a different ABI and soname than
upstream]
>> It will therefore continue to be broken until GnuTLS upstream bumps
>> the symbol-versioning (and the soname) and Debian picks up this
>> version. 

> Indeed, but the ABI in 3.x is fixed since years and will remain fixed
> for years too; there is no plan to change it. Why not change the ABI in
> the old releases as a work around? Having a different ABI in the latest
> release ensures that debian will remain binary incompatible with any
> software build with modern gnutls for the foreseeable future.

Hello,

GnuTLS 2.x is part of the current Debian stable release ("wheezy" aka
7.6). There is simply no way to retroactively change the gnutls ABI
in wheezy and rebuild all packages in it. And because of partial
upgrades GnuTLS 2.x in wheezy and GnuTLS 3.x in wheezy+1 (the soon to
be frozen "jessie") will need to be co-installable without breaking
each other.

It is theorethically possible to do a transition and switch back to the
upstream GnuTLS 3.x ABI for the next-next Debian release, but I doubt
that it is really worth the trouble because of the timeframe. Debian
jessie+1 will probably be released in 2017. The additional transition
will also complicate things for Debian derivatives (like Ubuntu) who
have a different release cycle.

cu Andreas
-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'


From nmav at gnutls.org  Sat Sep 13 09:44:26 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Sat, 13 Sep 2014 09:44:26 +0200
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410545766.1339.15.camel@waverley>
References: <1410545766.1339.15.camel@waverley>
Message-ID: <1410594266.2452.17.camel@nomad.lan>

On Fri, 2014-09-12 at 14:16 -0400, Armin Burgmeier wrote:

Thanks. Some comments inline.

> -	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
> -	    ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
> -	     || issuer_version != 1)) {
> +	} else if (issuer != NULL &&
> +                   !(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
> +	           ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
> +	            || issuer_version != 1)) {

After this point wouldn't it make sense to have one large block with an
if (issuer != NULL)? (or at least one that is interrupted at some
point). I'm worried about the multiple checks for if not NULL, as it is
easy to miss one.

>  	if (func) {
>  		if (result == 0) {
>  			out |= GNUTLS_CERT_INVALID;
> @@ -1414,7 +1401,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
>  			*verify |=
>  			    GNUTLS_CERT_SIGNER_NOT_FOUND |
>  			    GNUTLS_CERT_INVALID;
> -		return 0;
>  	}
>  
>  	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
> @@ -1424,7 +1410,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
>  				*verify |=
>  				    GNUTLS_CERT_SIGNER_NOT_CA |
>  				    GNUTLS_CERT_INVALID;
> -			return 0;
>  		}

Shouldn't the if (issuer != NULL) be repeated here as well? Even if
gnutls_x509_crt_get_ca_status() doesn't crash, you'll most probably get
a flag GNUTLS_CERT_SIGNER_NOT_CA, which is misleading as there is no
signer there.

regards,
Nikos




From nmav at gnutls.org  Sat Sep 13 09:48:51 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Sat, 13 Sep 2014 09:48:51 +0200
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410547192.1339.22.camel@waverley>
References: <1410545766.1339.15.camel@waverley>
 <1410547192.1339.22.camel@waverley>
Message-ID: <1410594531.2452.20.camel@nomad.lan>

On Fri, 2014-09-12 at 14:39 -0400, Armin Burgmeier wrote:
> This is an attempt at a patch for what we discussed in the other email
> thread. I have made sure that the tests still pass (with minor
> modifications where additional failure reasons are now expected). Since
> this is in a quite sensitive part of the code it would be good to have
> this reviewed carefully...
> 
> > @@ -1504,10 +1488,10 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
> >  		*verify |= GNUTLS_CERT_REVOCATION_DATA_SUPERSEDED;
> >  
> > 
> > -      cleanup:
> >  	if (verify)
> >  		*verify |= GNUTLS_CERT_INVALID;
> >  
> > +      cleanup:
> >  	_gnutls_free_datum(&crl_signed_data);
> >  	_gnutls_free_datum(&crl_signature);
> 
> I have two comments to this hunk.
> 
>  1) the question where to put the cleanup label is basically what should
> happen with the *verify output parameter when the function itself does
> not return GNUTLS_E_SUCCESS. Probably it does not matter because when
> the function does not run successfully, the verification output is of no
> good use anyway.
> 
>  2) If I understand the code correctly, then the verify result will
> *always* have the GNUTLS_CERT_INVALID flag set, even for perfectly valid
> CRLs. I have the impression the check should read "if(verify && result !
> = 0)" instead of just "if(verify)"... Is there a unit test for this
> function?

That is indeed correct. The check should have been if (verify &&
*verify) ... It seems we need some unit tests for that one.

regards,
Nikos


> 
> Cheers,
> Armin
> 
> 
> _______________________________________________
> Gnutls-devel mailing list
> Gnutls-devel at lists.gnutls.org
> http://lists.gnupg.org/mailman/listinfo/gnutls-devel




From armin at arbur.net  Sat Sep 13 17:45:51 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Sat, 13 Sep 2014 11:45:51 -0400
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410594266.2452.17.camel@nomad.lan>
References: <1410545766.1339.15.camel@waverley>
 <1410594266.2452.17.camel@nomad.lan>
Message-ID: <1410623151.1354.11.camel@waverley>

Thanks for the comments. I'll send an updated patch shortly.

Cheers,
Armin

On Sat, 2014-09-13 at 09:44 +0200, Nikos Mavrogiannopoulos wrote:
> On Fri, 2014-09-12 at 14:16 -0400, Armin Burgmeier wrote:
> 
> Thanks. Some comments inline.
> 
> > -	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
> > -	    ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
> > -	     || issuer_version != 1)) {
> > +	} else if (issuer != NULL &&
> > +                   !(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
> > +	           ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
> > +	            || issuer_version != 1)) {
> 
> After this point wouldn't it make sense to have one large block with an
> if (issuer != NULL)? (or at least one that is interrupted at some
> point). I'm worried about the multiple checks for if not NULL, as it is
> easy to miss one.

My motivation mostly was to avoid too many levels of indentation, to
maybe keep is easier to follow the logic of the code. But I don't have a
strong opinion. I have changed it to use a large issuer != NULL block in
the updated patch.

> >  	if (func) {
> >  		if (result == 0) {
> >  			out |= GNUTLS_CERT_INVALID;
> > @@ -1414,7 +1401,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
> >  			*verify |=
> >  			    GNUTLS_CERT_SIGNER_NOT_FOUND |
> >  			    GNUTLS_CERT_INVALID;
> > -		return 0;
> >  	}
> >  
> >  	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
> > @@ -1424,7 +1410,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
> >  				*verify |=
> >  				    GNUTLS_CERT_SIGNER_NOT_CA |
> >  				    GNUTLS_CERT_INVALID;
> > -			return 0;
> >  		}
> 
> Shouldn't the if (issuer != NULL) be repeated here as well? Even if
> gnutls_x509_crt_get_ca_status() doesn't crash, you'll most probably get
> a flag GNUTLS_CERT_SIGNER_NOT_CA, which is misleading as there is no
> signer there.

Agreed. I added the check there.



From armin at arbur.net  Sat Sep 13 17:46:42 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Sat, 13 Sep 2014 11:46:42 -0400
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
	verifying a certificate
Message-ID: <1410623202.1354.12.camel@waverley>

This allows to check for all possible flaws with a certificate chain with a
single call to gnutls_x509_crt_list_verify and friends.

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/x509/verify.c   | 393 +++++++++++++++++++++++++---------------------------
 tests/test-chains.h |  13 +-
 2 files changed, 194 insertions(+), 212 deletions(-)

diff --git a/lib/x509/verify.c b/lib/x509/verify.c
index 0302973..6c4ef5d 100644
--- a/lib/x509/verify.c
+++ b/lib/x509/verify.c
@@ -572,7 +572,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	gnutls_datum_t cert_signature = { NULL, 0 };
 	gnutls_x509_crt_t issuer = NULL;
 	int issuer_version, hash_algo;
-	bool result = 0;
+	bool result = 1;
 	const mac_entry_st * me;
 	unsigned int out = 0, usage;
 	int sigalg, ret;
@@ -581,13 +581,12 @@ verify_crt(gnutls_x509_crt_t cert,
 		*output = 0;
 
 	if (*max_path == 0) {
-		out =
+		out |=
 		    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
 		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
 		gnutls_assert();
 		result = 0;
+		/* bail immediately, to avoid inconistency  */
 		goto cleanup;
 	}
 	(*max_path)--;
@@ -595,115 +594,16 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (tcas_size >= 1)
 		issuer = find_issuer(cert, trusted_cas, tcas_size);
 
-	/* issuer is not in trusted certificate
-	 * authorities.
-	 */
-	if (issuer == NULL) {
-		out = GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
-	}
-
 	if (_issuer != NULL)
 		*_issuer = issuer;
 
-	if (nc != NULL) {
-		/* append the issuer's constraints */
-		ret = gnutls_x509_crt_get_name_constraints(issuer, nc, 
-			GNUTLS_NAME_CONSTRAINTS_FLAG_APPEND, NULL);
-		if (ret < 0 && ret != GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE) {
- nc_fail:
-			out =
-			    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
-			    GNUTLS_CERT_INVALID;
-			if (output)
-				*output |= out;
-			gnutls_assert();
-			result = 0;
-			goto cleanup;
-		}
-
-		/* only check name constraints in server certificates, not CAs */
-		if (end_cert != 0) {
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DNSNAME, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_RFC822NAME, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DN, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_URI, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_IPADDRESS, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-		}
-	}
-
-	issuer_version = gnutls_x509_crt_get_version(issuer);
-	if (issuer_version < 0) {
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
-	}
-
-	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
-	    ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
-	     || issuer_version != 1)) {
-		if (check_if_ca(cert, issuer, max_path, flags) != 1) {
-			gnutls_assert();
-			out =
-			    GNUTLS_CERT_SIGNER_NOT_CA |
-			    GNUTLS_CERT_INVALID;
-			if (output)
-				*output |= out;
-			result = 0;
-			goto cleanup;
-		}
-
-		ret =
-		    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
-		if (ret >= 0) {
-			if (!(usage & GNUTLS_KEY_KEY_CERT_SIGN)) {
-				gnutls_assert();
-				out =
-				    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
-				    | GNUTLS_CERT_INVALID;
-				if (output)
-					*output |= out;
-				result = 0;
-				goto cleanup;
-			}
-		}
-	}
-
 	ret =
 	    _gnutls_x509_get_signed_data(cert->cert, &cert->der, "tbsCertificate",
 					 &cert_signed_data);
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
+		cert_signed_data.data = NULL;
 	}
 
 	ret =
@@ -712,7 +612,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
+		cert_signature.data = NULL;
 	}
 
 	ret =
@@ -721,75 +621,156 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
 	}
 	sigalg = ret;
 
-	if (is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
-		gnutls_assert();
-		out =
-		    GNUTLS_CERT_INSECURE_ALGORITHM |
-		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
-		result = 0;
-		goto cleanup;
-	}
-
-	/* If the certificate is not self signed check if the algorithms
-	 * used are secure. If the certificate is self signed it doesn't
-	 * really matter.
+	/* issuer is not in trusted certificate
+	 * authorities.
 	 */
-	if (gnutls_sign_is_secure(sigalg) == 0 &&
-	    is_broken_allowed(sigalg, flags) == 0 &&
-	    is_issuer(cert, cert) == 0) {
+	if (issuer == NULL) {
+		out |= GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID;
 		gnutls_assert();
-		out =
-		    GNUTLS_CERT_INSECURE_ALGORITHM |
-		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
 		result = 0;
-		goto cleanup;
-	}
+	} else {
+		if (nc != NULL) {
+			/* append the issuer's constraints */
+			ret = gnutls_x509_crt_get_name_constraints(issuer, nc, 
+				GNUTLS_NAME_CONSTRAINTS_FLAG_APPEND, NULL);
+			if (ret < 0 && ret != GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE) {
+ nc_fail:
+				out |=
+				    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
+				    GNUTLS_CERT_INVALID;
+				gnutls_assert();
+				result = 0;
+				goto nc_done;
+			}
 
-	hash_algo = gnutls_sign_get_hash_algorithm(sigalg);
-	me = mac_to_entry(hash_algo);
-	if (me == NULL) {
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
+			/* only check name constraints in server certificates, not CAs */
+			if (end_cert != 0) {
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DNSNAME, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_RFC822NAME, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DN, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_URI, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_IPADDRESS, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+			}
+		}
+ nc_done:
+
+		issuer_version = gnutls_x509_crt_get_version(issuer);
+
+		if (issuer_version < 0) {
+			gnutls_assert();
+			result = 0;
+		} else if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
+	                   ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
+	                    || issuer_version != 1)) {
+			if (check_if_ca(cert, issuer, max_path, flags) != 1) {
+				gnutls_assert();
+				out |=
+				    GNUTLS_CERT_SIGNER_NOT_CA |
+				    GNUTLS_CERT_INVALID;
+				result = 0;
+			}
+
+			ret =
+			    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
+			if (ret >= 0) {
+				if (!(usage & GNUTLS_KEY_KEY_CERT_SIGN)) {
+					gnutls_assert();
+					out |=
+					    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
+					    | GNUTLS_CERT_INVALID;
+					result = 0;
+				}
+			}
+		}
+
+		if (sigalg >= 0 &&
+		    is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
+			gnutls_assert();
+			out |=
+			    GNUTLS_CERT_INSECURE_ALGORITHM |
+			    GNUTLS_CERT_INVALID;
+			result = 0;
+		}
+
+		hash_algo = gnutls_sign_get_hash_algorithm(sigalg);
+		me = mac_to_entry(hash_algo);
+		if (me == NULL) {
+			gnutls_assert();
+			result = 0;
+		} else if (cert_signed_data.data != NULL &&
+		           cert_signature.data != NULL) {
+			ret =
+			    _gnutls_x509_verify_data(me,
+						     &cert_signed_data,
+		        	                     &cert_signature,
+						     issuer);
+			if (ret == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
+				gnutls_assert();
+				out |=
+				    GNUTLS_CERT_INVALID |
+				    GNUTLS_CERT_SIGNATURE_FAILURE;
+				/* error. ignore it */
+				result = 0;
+			} else if (ret < 0) {
+				result = 0;
+				gnutls_assert();
+			}
+		}
 	}
 
-	ret =
-	    _gnutls_x509_verify_data(me,
-				     &cert_signed_data, &cert_signature,
-				     issuer);
-	if (ret == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
-		gnutls_assert();
-		out |= GNUTLS_CERT_INVALID | GNUTLS_CERT_SIGNATURE_FAILURE;
-		/* error. ignore it */
-		if (output)
-			*output |= out;
-		result = 0;
-	} else if (ret < 0) {
-		result = 0;
-		gnutls_assert();
-		goto cleanup;
-	} else if (ret == 1)
-		result = 1;
+	if (sigalg >= 0) {
+		/* If the certificate is not self signed check if the algorithms
+		 * used are secure. If the certificate is self signed it doesn't
+		 * really matter.
+		 */
+		if (gnutls_sign_is_secure(sigalg) == 0 &&
+		    is_broken_allowed(sigalg, flags) == 0 &&
+		    is_issuer(cert, cert) == 0) {
+			gnutls_assert();
+			out |=
+			    GNUTLS_CERT_INSECURE_ALGORITHM |
+			    GNUTLS_CERT_INVALID;
+			result = 0;
+		}
+	}
 
 	/* Check activation/expiration times
 	 */
 	if (!(flags & GNUTLS_VERIFY_DISABLE_TIME_CHECKS)) {
 		/* check the time of the issuer first */
-		if (!(flags & GNUTLS_VERIFY_DISABLE_TRUSTED_TIME_CHECKS)) {
+		if (issuer != NULL &&
+		    !(flags & GNUTLS_VERIFY_DISABLE_TRUSTED_TIME_CHECKS)) {
 			out |= check_time_status(issuer, now);
 			if (out != 0) {
 				gnutls_assert();
 				result = 0;
-				if (output)
-					*output |= out;
 			}
 		}
 
@@ -797,12 +778,13 @@ verify_crt(gnutls_x509_crt_t cert,
 		if (out != 0) {
 			gnutls_assert();
 			result = 0;
-			if (output)
-				*output |= out;
 		}
 	}
 
       cleanup:
+	if (output)
+		*output |= out;
+
 	if (func) {
 		if (result == 0) {
 			out |= GNUTLS_CERT_INVALID;
@@ -1405,42 +1387,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 	if (tcas_size >= 1)
 		issuer = find_crl_issuer(crl, trusted_cas, tcas_size);
 
-	/* issuer is not in trusted certificate
-	 * authorities.
-	 */
-	if (issuer == NULL) {
-		gnutls_assert();
-		if (verify)
-			*verify |=
-			    GNUTLS_CERT_SIGNER_NOT_FOUND |
-			    GNUTLS_CERT_INVALID;
-		return 0;
-	}
-
-	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
-		if (gnutls_x509_crt_get_ca_status(issuer, NULL) != 1) {
-			gnutls_assert();
-			if (verify)
-				*verify |=
-				    GNUTLS_CERT_SIGNER_NOT_CA |
-				    GNUTLS_CERT_INVALID;
-			return 0;
-		}
-
-		result =
-		    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
-		if (result >= 0) {
-			if (!(usage & GNUTLS_KEY_CRL_SIGN)) {
-				gnutls_assert();
-				if (verify)
-					*verify |=
-					    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
-					    | GNUTLS_CERT_INVALID;
-				return 0;
-			}
-		}
-	}
-
 	result =
 	    _gnutls_x509_get_signed_data(crl->crl, &crl->der, "tbsCertList",
 					 &crl_signed_data);
@@ -1473,21 +1419,54 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 
 	hash_algo = gnutls_sign_get_hash_algorithm(result);
 
-	result =
-	    _gnutls_x509_verify_data(mac_to_entry(hash_algo),
-				     &crl_signed_data, &crl_signature,
-				     issuer);
-	if (result == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
-		gnutls_assert();
-		/* error. ignore it */
-		if (verify)
-			*verify |= GNUTLS_CERT_SIGNATURE_FAILURE;
-		result = 0;
-	} else if (result < 0) {
+	/* issuer is not in trusted certificate
+	 * authorities.
+	 */
+	if (issuer == NULL) {
 		gnutls_assert();
 		if (verify)
-			*verify |= GNUTLS_CERT_INVALID;
-		goto cleanup;
+			*verify |=
+			    GNUTLS_CERT_SIGNER_NOT_FOUND |
+			    GNUTLS_CERT_INVALID;
+	} else {
+		if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
+			if (gnutls_x509_crt_get_ca_status(issuer, NULL) != 1) {
+				gnutls_assert();
+				if (verify)
+					*verify |=
+					    GNUTLS_CERT_SIGNER_NOT_CA |
+					    GNUTLS_CERT_INVALID;
+			}
+
+			result =
+			    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
+			if (result >= 0) {
+				if (!(usage & GNUTLS_KEY_CRL_SIGN)) {
+					gnutls_assert();
+					if (verify)
+						*verify |=
+						    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
+						    | GNUTLS_CERT_INVALID;
+				}
+			}
+		}
+
+		result =
+		    _gnutls_x509_verify_data(mac_to_entry(hash_algo),
+					     &crl_signed_data, &crl_signature,
+					     issuer);
+		if (result == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
+			gnutls_assert();
+			/* error. ignore it */
+			if (verify)
+				*verify |= GNUTLS_CERT_SIGNATURE_FAILURE;
+			result = 0;
+		} else if (result < 0) {
+			gnutls_assert();
+			if (verify)
+				*verify |= GNUTLS_CERT_INVALID;
+			goto cleanup;
+		}
 	}
 
 	{
diff --git a/tests/test-chains.h b/tests/test-chains.h
index 28974e1..ff9086f 100644
--- a/tests/test-chains.h
+++ b/tests/test-chains.h
@@ -1366,9 +1366,11 @@ static struct
 } chains[] =
 {
   { "CVE-2014-0092", cve_2014_0092_check, &cve_2014_0092_check[1],
-    0, GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
   { "CVE-2008-4989", cve_2008_4989_chain, &cve_2008_4989_chain[2],
-    0, GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
   { "amazon.com ok", verisign_com_chain_g5, &verisign_com_chain_g5[4],
     GNUTLS_VERIFY_DISABLE_TIME_CHECKS | GNUTLS_PROFILE_TO_VFLAGS(GNUTLS_PROFILE_LOW),
     0 },
@@ -1404,9 +1406,10 @@ static struct
     GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
     0 },
   { "rsa-md5 fail", mayfirst_chain, &mayfirst_chain[1],
-    0, GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "rsa-md5 not ok", mayfirst_chain, &mayfirst_chain[1],
-    GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD2,
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS | GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD2,
     GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "rsa-md5 not ok2", mayfirst_chain, &mayfirst_chain[1],
     GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD5,
@@ -1449,7 +1452,7 @@ static struct
     GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD5, 0 },
   { "cacertrsamd5 short-cut not ok", cacertrsamd5, &cacertrsamd5[0],
     GNUTLS_VERIFY_DO_NOT_ALLOW_SAME,
-    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
+    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "cacertrsamd5 short-cut ok", cacertrsamd5, &cacertrsamd5[1],
     0, 0 },
   { "ecc cert ok", ecc_cert, &ecc_cert[1], GNUTLS_PROFILE_TO_VFLAGS(GNUTLS_PROFILE_HIGH), 0 },
-- 
2.1.0




From nmav at gnu.org  Mon Sep 15 09:43:10 2014
From: nmav at gnu.org (Nikos Mavrogiannopoulos)
Date: Mon, 15 Sep 2014 09:43:10 +0200
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <1410534621.1339.7.camel@waverley>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
 <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
 <1410362139.2056.24.camel@waverley>
 <CAJU7zaKbs66rnXGgxHxRUP841U-b9buxO_OHmk43bmg0Oa5+iA@mail.gmail.com>
 <CAJU7zaLXhsgX2ReU3PocpY5H7SH2qd82KwNFY7aT5o_DX687TQ@mail.gmail.com>
 <1410534621.1339.7.camel@waverley>
Message-ID: <CAJU7za+pV42HuHn8TjyZV8-YRpHsFvSTzUedQCEPGEwqLuRh0g@mail.gmail.com>

On Fri, Sep 12, 2014 at 5:10 PM, Armin Burgmeier <armin at arbur.net> wrote:
>> I should have added here, that if there is a patch, I'd review it of
>> course (and something like that would speed up the process).
> Yes, I'll try to come up with a patch.
> While trying to build from git master I noticed that in commit 469f8fb
> you added pkcs11x.c to lib/Makefile.am, however that file does not exist
> in the repository... did you maybe forget to git add it?

It depended on a patched p11-kit. I've now added some definitions to
allow compilation with any version of p11-kit. If the problem persists
let me know.

regards,
Nikos


From nmav at gnutls.org  Mon Sep 15 09:46:38 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 15 Sep 2014 09:46:38 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <20140913060441.GA2429@downhill.g.la>
References: <20140524065807.GA2250@downhill.g.la>
 <1410468974.3121.3.camel@nomad.lan>
 <20140912162227.GA5709@downhill.g.la>
 <1410546977.11023.8.camel@nomad.lan>
 <20140913060441.GA2429@downhill.g.la>
Message-ID: <CAJU7zaKYOuRAy8V2ns=eoj0KJ+YfMuUBQ33GR1vfqqH0ZOdiNw@mail.gmail.com>

On Sat, Sep 13, 2014 at 8:04 AM, Andreas Metzler <ametzler at bebt.de> wrote:
> [Debian GnuTLS v3 package use a different ABI and soname than
> upstream]
>>> It will therefore continue to be broken until GnuTLS upstream bumps
>>> the symbol-versioning (and the soname) and Debian picks up this
>>> version.
>> Indeed, but the ABI in 3.x is fixed since years and will remain fixed
>> for years too; there is no plan to change it. Why not change the ABI in
>> the old releases as a work around? Having a different ABI in the latest
>> release ensures that debian will remain binary incompatible with any
>> software build with modern gnutls for the foreseeable future.
> It is theorethically possible to do a transition and switch back to the
> upstream GnuTLS 3.x ABI for the next-next Debian release, but I doubt
> that it is really worth the trouble because of the timeframe. Debian
> jessie+1 will probably be released in 2017. The additional transition
> will also complicate things for Debian derivatives (like Ubuntu) who
> have a different release cycle.

I'm wondering whether there is some way to transition to a new ABI
without breaking the old one. I'll try to see whether the ld script
allows such move.

regards,
Nikos


From nmav at gnutls.org  Mon Sep 15 15:49:05 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 15 Sep 2014 15:49:05 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <CAJU7zaKYOuRAy8V2ns=eoj0KJ+YfMuUBQ33GR1vfqqH0ZOdiNw@mail.gmail.com>
References: <20140524065807.GA2250@downhill.g.la>
 <1410468974.3121.3.camel@nomad.lan>
 <20140912162227.GA5709@downhill.g.la>
 <1410546977.11023.8.camel@nomad.lan>
 <20140913060441.GA2429@downhill.g.la>
 <CAJU7zaKYOuRAy8V2ns=eoj0KJ+YfMuUBQ33GR1vfqqH0ZOdiNw@mail.gmail.com>
Message-ID: <CAJU7zaLvs1S3YyxDVKMut5BpdKdaE+CYQDnxWLihfxkF8pxajg@mail.gmail.com>

On Mon, Sep 15, 2014 at 9:46 AM, Nikos Mavrogiannopoulos
<nmav at gnutls.org> wrote:
> I'm wondering whether there is some way to transition to a new ABI
> without breaking the old one. I'll try to see whether the ld script
> allows such move.

Ok, it seems that adding a new VERSION section with all the symbols in
the beginning of libgnutls.map, makes the default symbols to be the
new, but also exposes the old ones. So we can have a slow migration to
a new ABI, in the 3.4.0 release. However, That is not really
documented in [0], and I wonder whether we can rely on that.

regards,
Nikos

[0]. https://sourceware.org/binutils/docs/ld/VERSION.html#VERSION


From armin at arbur.net  Mon Sep 15 16:57:01 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Mon, 15 Sep 2014 10:57:01 -0400
Subject: [gnutls-devel] [sr #108634] Getter functions for
 gnutls_certificate_credentials_t
In-Reply-To: <CAJU7za+pV42HuHn8TjyZV8-YRpHsFvSTzUedQCEPGEwqLuRh0g@mail.gmail.com>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
 <CAJU7zaLha_d318vXEy1m2d2ncujmbq4S8grARr1fhYJkYTuvUA@mail.gmail.com>
 <1410184796.1723.20.camel@waverley>
 <CAJU7zaJsinGLxHcLtHM5Aaw5b6jMO07vzwGO0dZyiDXk5xJxNg@mail.gmail.com>
 <1410362139.2056.24.camel@waverley>
 <CAJU7zaKbs66rnXGgxHxRUP841U-b9buxO_OHmk43bmg0Oa5+iA@mail.gmail.com>
 <CAJU7zaLXhsgX2ReU3PocpY5H7SH2qd82KwNFY7aT5o_DX687TQ@mail.gmail.com>
 <1410534621.1339.7.camel@waverley>
 <CAJU7za+pV42HuHn8TjyZV8-YRpHsFvSTzUedQCEPGEwqLuRh0g@mail.gmail.com>
Message-ID: <1410793021.1347.12.camel@waverley>

On Mon, 2014-09-15 at 09:43 +0200, Nikos Mavrogiannopoulos wrote:
> On Fri, Sep 12, 2014 at 5:10 PM, Armin Burgmeier <armin at arbur.net> wrote:
> >> I should have added here, that if there is a patch, I'd review it of
> >> course (and something like that would speed up the process).
> > Yes, I'll try to come up with a patch.
> > While trying to build from git master I noticed that in commit 469f8fb
> > you added pkcs11x.c to lib/Makefile.am, however that file does not exist
> > in the repository... did you maybe forget to git add it?
> 
> It depended on a patched p11-kit. I've now added some definitions to
> allow compilation with any version of p11-kit. If the problem persists
> let me know.

Yep, it's working. Thanks!

Armin



From antoine.pierlot-garcin at scle.fr  Tue Sep 16 10:22:46 2014
From: antoine.pierlot-garcin at scle.fr (antoine.pierlot-garcin at scle.fr)
Date: Tue, 16 Sep 2014 10:22:46 +0200
Subject: [gnutls-devel] gnutls 3.1.26 build failures
Message-ID: <E600E31BE48FC746A528F50ED8EB570B3E34B1B15E@xs203448.MELINDA.LOCAL>

Hello,

Recently I had to compile GnuTLS 3.1.26 for a win32 target 
(cross-compile from Debian unstable) and for a Debian stable target. I 
encountered two build failures, which I was able to resolve. Here some 
details and the patches, if by any chance you want to include them in a 
future 3.1.x release.

----
make[4]: Entering directory '/root/src/gnutls/gnutls-3.1.26/src'
/bin/bash ../libtool  --tag=CC   --mode=compile i586-mingw32msvc-gcc 
-std=gnu99 -DHAVE_CONFIG_H -I. -I..  -I./../gl -I./../gl 
-I./../lib/includes -I./../lib/includes -I./../libdane/includes 
-I./../extra/includes -I../src/libopts 
-I/root/src/gnutls/gnutls-3.1.26/../lib_win32/include  -O2 -MT 
libcmd_srp_la-srptool-args.lo -MD -MP -MF 
.deps/libcmd_srp_la-srptool-args.Tpo -c -o libcmd_srp_la-srptool-args.lo 
`test -f 'srptool-args.c' || echo './'`srptool-args.c
libtool: compile:  i586-mingw32msvc-gcc -std=gnu99 -DHAVE_CONFIG_H -I. 
-I.. -I./../gl -I./../gl -I./../lib/includes -I./../lib/includes 
-I./../libdane/includes -I./../extra/includes -I../src/libopts 
-I/root/src/gnutls/gnutls-3.1.26/../lib_win32/include -O2 -MT 
libcmd_srp_la-srptool-args.lo -MD -MP -MF 
.deps/libcmd_srp_la-srptool-args.Tpo -c srptool-args.c  -DDLL_EXPORT 
-DPIC -o .libs/libcmd_srp_la-srptool-args.o
In file included from srptool-args.c:43:0:
srptool-args.h:61:3: error: #error option template version mismatches 
autoopts/options.h header
  # error option template version mismatches autoopts/options.h header
----

There is a mismatch between the generated options.h (autoopts from 
autogen 5.18.3, already in GnuTLS sources) and the embedded libopts at 
src/libopts which is obtained from autogen 5.18.2. I could not 
re-generate the headers with the embedded libopts, so I updated 
src/libopts to 5.18.3. I am not sure this is the right solution, so feel 
free to ignore the patch.

----
i586-mingw32msvc-gcc -std=gnu99 -DHAVE_CONFIG_H -I. -I..  -I./../gl 
-I./../gl -I./../lib/includes -I./../lib/includes 
-I./../libdane/includes -I./../extra/includes -I../src/libopts 
-I/root/src/gnutls/gnutls-3.1.26/../lib_win32/include   -O2 -MT cli.o 
-MD -MP -MF .deps/cli.Tpo -c -o cli.o cli.c
cli.c: In function 'main':
cli.c:844:20: error: storage size of 'new_action' isn't known
    struct sigaction new_action, old_action;
----

The mingw32 compiler does not define struct sigaction and sigaction() in 
<signal.h>. In cli.c, the uses are under #ifndef _WIN32 guards, but not 
the definition. The second patch fixes that.

Antoine.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gnutls_fix_win32_cli_sigaction.patch
Type: text/x-patch
Size: 450 bytes
Desc: gnutls_fix_win32_cli_sigaction.patch
URL: </pipermail/attachments/20140916/7c5935a4/attachment-0002.bin>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: gnutls_update_libopts_to_5.18.3.patch
Type: text/x-patch
Size: 78531 bytes
Desc: gnutls_update_libopts_to_5.18.3.patch
URL: </pipermail/attachments/20140916/7c5935a4/attachment-0003.bin>

From nmav at gnutls.org  Tue Sep 16 13:08:09 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Tue, 16 Sep 2014 13:08:09 +0200
Subject: [gnutls-devel] gnutls 3.1.26 build failures
In-Reply-To: <E600E31BE48FC746A528F50ED8EB570B3E34B1B15E@xs203448.MELINDA.LOCAL>
References: <E600E31BE48FC746A528F50ED8EB570B3E34B1B15E@xs203448.MELINDA.LOCAL>
Message-ID: <CAJU7zaLSK8WFk3ePaBJt2x5-BJqMDSKXt=6UYa68JvTBFs3cWQ@mail.gmail.com>

On Tue, Sep 16, 2014 at 10:22 AM,  <antoine.pierlot-garcin at scle.fr> wrote:
> Hello,
> Recently I had to compile GnuTLS 3.1.26 for a win32 target
> (cross-compile from Debian unstable) and for a Debian stable target. I
> encountered two build failures, which I was able to resolve. Here some
> details and the patches, if by any chance you want to include them in a
> future 3.1.x release.

Thank you. I've updated libopts in the repository and applied the
sigaction patch.

regards,
Nikos


From nmav at gnutls.org  Tue Sep 16 13:32:00 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Tue, 16 Sep 2014 13:32:00 +0200
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410623202.1354.12.camel@waverley>
References: <1410623202.1354.12.camel@waverley>
Message-ID: <CAJU7zaJ-n-r=WZ8DmbELHkdV5wGtgVzWoCWdVxQgxN3bGU3rqQ@mail.gmail.com>

On Sat, Sep 13, 2014 at 5:46 PM, Armin Burgmeier <armin at arbur.net> wrote:
> This allows to check for all possible flaws with a certificate chain with a
> single call to gnutls_x509_crt_list_verify and friends.

> +               if (sigalg >= 0 &&
> +                   is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
> +                       gnutls_assert();
> +                       out |=
> +                           GNUTLS_CERT_INSECURE_ALGORITHM |
> +                           GNUTLS_CERT_INVALID;
> +                       result = 0;
> +               }

Hi,
 The security level check will not be performed if there is no issuer.
That means if the issuer wasn't found you'll not know whether
GNUTLS_CERT_INSECURE_ALGORITHM would have been set for the specified
security level. Would that be acceptable?

Also you added the "if (sigalg >= 0)"... Why is that needed? Isn't
sigalg always positive at this point?

>         {
> diff --git a/tests/test-chains.h b/tests/test-chains.h
> index 28974e1..ff9086f 100644
> --- a/tests/test-chains.h
> +++ b/tests/test-chains.h
> @@ -1366,9 +1366,11 @@ static struct
>  } chains[] =
>  {
>    { "CVE-2014-0092", cve_2014_0092_check, &cve_2014_0092_check[1],
> -    0, GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
> +    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
> +    GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
>    { "CVE-2008-4989", cve_2008_4989_chain, &cve_2008_4989_chain[2],
> -    0, GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
> +    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
> +    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },

Wouldn't here instead of adding GNUTLS_VERIFY_DISABLE_TIME_CHECKS, to
add the GNUTLS_CERT_EXPIRED in the expected result? That way we can
test that the results remain consistent from now on.

Other than these, it looks reasonable. btw. would you like to send a
DCO on the list (as in http://www.gnutls.org/devel.html) ?

regards,
Nikos


From armin at arbur.net  Tue Sep 16 19:57:09 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Tue, 16 Sep 2014 13:57:09 -0400
Subject: [gnutls-devel] [PATCH] Check for all error conditions when
 verifying a certificate
In-Reply-To: <CAJU7zaJ-n-r=WZ8DmbELHkdV5wGtgVzWoCWdVxQgxN3bGU3rqQ@mail.gmail.com>
References: <1410623202.1354.12.camel@waverley>
 <CAJU7zaJ-n-r=WZ8DmbELHkdV5wGtgVzWoCWdVxQgxN3bGU3rqQ@mail.gmail.com>
Message-ID: <1410890229.1622.8.camel@waverley>

On Tue, 2014-09-16 at 13:32 +0200, Nikos Mavrogiannopoulos wrote:
> On Sat, Sep 13, 2014 at 5:46 PM, Armin Burgmeier <armin at arbur.net> wrote:
> > This allows to check for all possible flaws with a certificate chain with a
> > single call to gnutls_x509_crt_list_verify and friends.
> 
> > +               if (sigalg >= 0 &&
> > +                   is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
> > +                       gnutls_assert();
> > +                       out |=
> > +                           GNUTLS_CERT_INSECURE_ALGORITHM |
> > +                           GNUTLS_CERT_INVALID;
> > +                       result = 0;
> > +               }
> 
> Hi,
>  The security level check will not be performed if there is no issuer.
> That means if the issuer wasn't found you'll not know whether
> GNUTLS_CERT_INSECURE_ALGORITHM would have been set for the specified
> security level. Would that be acceptable?

For my own use case yes, but I agree that while we are at it we should
consider this as well. I have changed the function so that it is allowed
to be called with a NULL issuer.

> Also you added the "if (sigalg >= 0)"... Why is that needed? Isn't
> sigalg always positive at this point?

It can be negative if _gnutls_x509_get_signature_algorithm returns an
error code.

> >         {
> > diff --git a/tests/test-chains.h b/tests/test-chains.h
> > index 28974e1..ff9086f 100644
> > --- a/tests/test-chains.h
> > +++ b/tests/test-chains.h
> > @@ -1366,9 +1366,11 @@ static struct
> >  } chains[] =
> >  {
> >    { "CVE-2014-0092", cve_2014_0092_check, &cve_2014_0092_check[1],
> > -    0, GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
> > +    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
> > +    GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
> >    { "CVE-2008-4989", cve_2008_4989_chain, &cve_2008_4989_chain[2],
> > -    0, GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
> > +    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
> > +    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
> 
> Wouldn't here instead of adding GNUTLS_VERIFY_DISABLE_TIME_CHECKS, to
> add the GNUTLS_CERT_EXPIRED in the expected result? That way we can
> test that the results remain consistent from now on.

I agree, that makes sense.

> Other than these, it looks reasonable. btw. would you like to send a
> DCO on the list (as in http://www.gnutls.org/devel.html) ?

Yes, I'll do it. Will also work on a patch for the getter functions for
gnutls_certificate_credentials_t that we discussed about.

> regards,
> Nikos

Cheers,
Armin



From armin at arbur.net  Tue Sep 16 19:58:46 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Tue, 16 Sep 2014 13:58:46 -0400
Subject: [gnutls-devel] DCO
Message-ID: <1410890326.1622.9.camel@waverley>

Hi,

below the DCO for my contributions.

Cheers,
Armin

Developer's Certificate of Origin 1.1

    By making a contribution to this project, I certify that:
    (a) The contribution was created in whole or in part by me and I
    have the right to submit it under the open source license
    indicated in the file; or
    (b) The contribution is based upon previous work that, to the best
    of my knowledge, is covered under an appropriate open source
    license and I have the right under that license to submit that
    work with modifications, whether created in whole or in part
    by me, under the same open source license (unless I am
    permitted to submit under a different license), as indicated
    in the file; or
    (c) The contribution was provided directly to me by some other
    person who certified (a), (b) or (c) and I have not modified
    it.
    (d) I understand and agree that this project and the contribution
    are public and that a record of the contribution (including all
    personal information I submit with it, including my sign-off) is
    maintained indefinitely and may be redistributed consistent with
    this project or the open source license(s) involved.



From armin at arbur.net  Tue Sep 16 20:02:24 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Tue, 16 Sep 2014 14:02:24 -0400
Subject: [gnutls-devel] [PATCH V3] Check for all error conditions when
 verifying a certificate
Message-ID: <1410890544.1622.11.camel@waverley>

This allows to check for all possible flaws with a certificate chain with a
single call to gnutls_x509_crt_list_verify and friends.

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/x509/verify.c   | 463 +++++++++++++++++++++++++---------------------------
 tests/test-chains.h |  13 +-
 2 files changed, 232 insertions(+), 244 deletions(-)

diff --git a/lib/x509/verify.c b/lib/x509/verify.c
index e71832d..53de776 100644
--- a/lib/x509/verify.c
+++ b/lib/x509/verify.c
@@ -423,7 +423,7 @@ int is_broken_allowed(gnutls_sign_algorithm_t sig, unsigned int flags)
  * according to verification profile specified.
  *
  * @crt: a certificate
- * @issuer: the certificates issuer
+ * @issuer: the certificates issuer (allowed to be NULL)
  * @sigalg: the signature algorithm used
  * @flags: the specified verification flags
  */
@@ -469,11 +469,6 @@ int hash;
 				return gnutls_assert_val(0);
 			}
 
-			if (gnutls_x509_crt_get_version(issuer) != 3) {
-				_gnutls_debug_log("SUITEB: certificate's issuer uses an unacceptable version number\n");
-				return gnutls_assert_val(0);
-			}
-
 			if (sigalg != GNUTLS_SIGN_ECDSA_SHA256 && sigalg != GNUTLS_SIGN_ECDSA_SHA384) {
 				_gnutls_debug_log("SUITEB: certificate is not signed using ECDSA-SHA256 or ECDSA-SHA384\n");
 				return gnutls_assert_val(0);
@@ -498,38 +493,12 @@ int hash;
 			curve = params.flags;
 			gnutls_pk_params_release(&params);
 
-			ret = _gnutls_x509_crt_get_mpis(issuer, &params);
-			if (ret < 0) {
-				_gnutls_debug_log("SUITEB: cannot read certificate params\n");
-				return gnutls_assert_val(0);
-			}
-
-			issuer_curve = params.flags;
-			gnutls_pk_params_release(&params);
-
-			if (issuer_curve != GNUTLS_ECC_CURVE_SECP256R1 &&
-				issuer_curve != GNUTLS_ECC_CURVE_SECP384R1) {
-				_gnutls_debug_log("SUITEB: certificate's issuer ECC params do not contain SECP256R1 or SECP384R1\n");
-				return gnutls_assert_val(0);
-			}
-
 			if (curve != GNUTLS_ECC_CURVE_SECP256R1 &&
 				curve != GNUTLS_ECC_CURVE_SECP384R1) {
 				_gnutls_debug_log("SUITEB: certificate's ECC params do not contain SECP256R1 or SECP384R1\n");
 				return gnutls_assert_val(0);
 			}
 
-			if (issuer_curve < curve) {
-				_gnutls_debug_log("SUITEB: certificate's issuer ECC params are weaker than the certificate's\n");
-				return gnutls_assert_val(0);
-			}
-
-			if (sigalg == GNUTLS_SIGN_ECDSA_SHA256 && 
-				issuer_curve == GNUTLS_ECC_CURVE_SECP384R1) {
-				_gnutls_debug_log("SUITEB: certificate is signed with ECDSA-SHA256 when using SECP384R1\n");
-				return gnutls_assert_val(0);
-			}
-
 			if (profile == GNUTLS_PROFILE_SUITEB192) {
 				if (curve != GNUTLS_ECC_CURVE_SECP384R1) {
 					_gnutls_debug_log("SUITEB192: certificate does not use SECP384R1\n");
@@ -537,6 +506,39 @@ int hash;
 				}
 			}
 
+			if (issuer != NULL) {
+				if (gnutls_x509_crt_get_version(issuer) != 3) {
+					_gnutls_debug_log("SUITEB: certificate's issuer uses an unacceptable version number\n");
+					return gnutls_assert_val(0);
+				}
+
+				ret = _gnutls_x509_crt_get_mpis(issuer, &params);
+				if (ret < 0) {
+					_gnutls_debug_log("SUITEB: cannot read certificate params\n");
+					return gnutls_assert_val(0);
+				}
+
+				issuer_curve = params.flags;
+				gnutls_pk_params_release(&params);
+
+				if (issuer_curve != GNUTLS_ECC_CURVE_SECP256R1 &&
+					issuer_curve != GNUTLS_ECC_CURVE_SECP384R1) {
+					_gnutls_debug_log("SUITEB: certificate's issuer ECC params do not contain SECP256R1 or SECP384R1\n");
+					return gnutls_assert_val(0);
+				}
+
+				if (issuer_curve < curve) {
+					_gnutls_debug_log("SUITEB: certificate's issuer ECC params are weaker than the certificate's\n");
+					return gnutls_assert_val(0);
+				}
+
+				if (sigalg == GNUTLS_SIGN_ECDSA_SHA256 && 
+					issuer_curve == GNUTLS_ECC_CURVE_SECP384R1) {
+					_gnutls_debug_log("SUITEB: certificate is signed with ECDSA-SHA256 when using SECP384R1\n");
+					return gnutls_assert_val(0);
+				}
+			}
+
 			break;
 		}
 	}
@@ -572,7 +574,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	gnutls_datum_t cert_signature = { NULL, 0 };
 	gnutls_x509_crt_t issuer = NULL;
 	int issuer_version, hash_algo;
-	bool result = 0;
+	bool result = 1;
 	const mac_entry_st * me;
 	unsigned int out = 0, usage;
 	int sigalg, ret;
@@ -581,13 +583,12 @@ verify_crt(gnutls_x509_crt_t cert,
 		*output = 0;
 
 	if (*max_path == 0) {
-		out =
+		out |=
 		    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
 		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
 		gnutls_assert();
 		result = 0;
+		/* bail immediately, to avoid inconistency  */
 		goto cleanup;
 	}
 	(*max_path)--;
@@ -595,115 +596,16 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (tcas_size >= 1)
 		issuer = find_issuer(cert, trusted_cas, tcas_size);
 
-	/* issuer is not in trusted certificate
-	 * authorities.
-	 */
-	if (issuer == NULL) {
-		out = GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
-	}
-
 	if (_issuer != NULL)
 		*_issuer = issuer;
 
-	if (nc != NULL) {
-		/* append the issuer's constraints */
-		ret = gnutls_x509_crt_get_name_constraints(issuer, nc, 
-			GNUTLS_NAME_CONSTRAINTS_FLAG_APPEND, NULL);
-		if (ret < 0 && ret != GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE) {
- nc_fail:
-			out =
-			    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
-			    GNUTLS_CERT_INVALID;
-			if (output)
-				*output |= out;
-			gnutls_assert();
-			result = 0;
-			goto cleanup;
-		}
-
-		/* only check name constraints in server certificates, not CAs */
-		if (end_cert != 0) {
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DNSNAME, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_RFC822NAME, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DN, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_URI, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-
-			ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_IPADDRESS, cert);
-			if (ret == 0) {
-				gnutls_assert();
-				goto nc_fail;
-			}
-		}
-	}
-
-	issuer_version = gnutls_x509_crt_get_version(issuer);
-	if (issuer_version < 0) {
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
-	}
-
-	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
-	    ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
-	     || issuer_version != 1)) {
-		if (check_if_ca(cert, issuer, max_path, flags) != 1) {
-			gnutls_assert();
-			out =
-			    GNUTLS_CERT_SIGNER_NOT_CA |
-			    GNUTLS_CERT_INVALID;
-			if (output)
-				*output |= out;
-			result = 0;
-			goto cleanup;
-		}
-
-		ret =
-		    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
-		if (ret >= 0) {
-			if (!(usage & GNUTLS_KEY_KEY_CERT_SIGN)) {
-				gnutls_assert();
-				out =
-				    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
-				    | GNUTLS_CERT_INVALID;
-				if (output)
-					*output |= out;
-				result = 0;
-				goto cleanup;
-			}
-		}
-	}
-
 	ret =
 	    _gnutls_x509_get_signed_data(cert->cert, &cert->der, "tbsCertificate",
 					 &cert_signed_data);
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
+		cert_signed_data.data = NULL;
 	}
 
 	ret =
@@ -712,7 +614,7 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
+		cert_signature.data = NULL;
 	}
 
 	ret =
@@ -721,75 +623,160 @@ verify_crt(gnutls_x509_crt_t cert,
 	if (ret < 0) {
 		result = 0;
 		gnutls_assert();
-		goto cleanup;
 	}
 	sigalg = ret;
 
-	if (is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
-		gnutls_assert();
-		out =
-		    GNUTLS_CERT_INSECURE_ALGORITHM |
-		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
-		result = 0;
-		goto cleanup;
-	}
-
-	/* If the certificate is not self signed check if the algorithms
-	 * used are secure. If the certificate is self signed it doesn't
-	 * really matter.
+	/* issuer is not in trusted certificate
+	 * authorities.
 	 */
-	if (gnutls_sign_is_secure(sigalg) == 0 &&
-	    is_broken_allowed(sigalg, flags) == 0 &&
-	    is_issuer(cert, cert) == 0) {
+	if (issuer == NULL) {
+		out |= GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID;
 		gnutls_assert();
-		out =
-		    GNUTLS_CERT_INSECURE_ALGORITHM |
-		    GNUTLS_CERT_INVALID;
-		if (output)
-			*output |= out;
 		result = 0;
-		goto cleanup;
-	}
+	} else {
+		if (nc != NULL) {
+			/* append the issuer's constraints */
+			ret = gnutls_x509_crt_get_name_constraints(issuer, nc, 
+				GNUTLS_NAME_CONSTRAINTS_FLAG_APPEND, NULL);
+			if (ret < 0 && ret != GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE) {
+ nc_fail:
+				out |=
+				    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE |
+				    GNUTLS_CERT_INVALID;
+				gnutls_assert();
+				result = 0;
+				goto nc_done;
+			}
 
-	hash_algo = gnutls_sign_get_hash_algorithm(sigalg);
-	me = mac_to_entry(hash_algo);
-	if (me == NULL) {
-		gnutls_assert();
-		result = 0;
-		goto cleanup;
+			/* only check name constraints in server certificates, not CAs */
+			if (end_cert != 0) {
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DNSNAME, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_RFC822NAME, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_DN, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_URI, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+
+				ret = gnutls_x509_name_constraints_check_crt(nc, GNUTLS_SAN_IPADDRESS, cert);
+				if (ret == 0) {
+					gnutls_assert();
+					goto nc_fail;
+				}
+			}
+		}
+ nc_done:
+
+		issuer_version = gnutls_x509_crt_get_version(issuer);
+
+		if (issuer_version < 0) {
+			gnutls_assert();
+			result = 0;
+		} else if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN) &&
+	                   ((flags & GNUTLS_VERIFY_DO_NOT_ALLOW_X509_V1_CA_CRT)
+	                    || issuer_version != 1)) {
+			if (check_if_ca(cert, issuer, max_path, flags) != 1) {
+				gnutls_assert();
+				out |=
+				    GNUTLS_CERT_SIGNER_NOT_CA |
+				    GNUTLS_CERT_INVALID;
+				result = 0;
+			}
+
+			ret =
+			    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
+			if (ret >= 0) {
+				if (!(usage & GNUTLS_KEY_KEY_CERT_SIGN)) {
+					gnutls_assert();
+					out |=
+					    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
+					    | GNUTLS_CERT_INVALID;
+					result = 0;
+				}
+			}
+		}
+
+		if (sigalg >= 0) {
+			hash_algo = gnutls_sign_get_hash_algorithm(sigalg);
+			me = mac_to_entry(hash_algo);
+		} else {
+			me = NULL;
+		}
+
+		if (me == NULL) {
+			gnutls_assert();
+			result = 0;
+		} else if (cert_signed_data.data != NULL &&
+		           cert_signature.data != NULL) {
+			ret =
+			    _gnutls_x509_verify_data(me,
+						     &cert_signed_data,
+		        	                     &cert_signature,
+						     issuer);
+			if (ret == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
+				gnutls_assert();
+				out |=
+				    GNUTLS_CERT_INVALID |
+				    GNUTLS_CERT_SIGNATURE_FAILURE;
+				/* error. ignore it */
+				result = 0;
+			} else if (ret < 0) {
+				result = 0;
+				gnutls_assert();
+			}
+		}
 	}
 
-	ret =
-	    _gnutls_x509_verify_data(me,
-				     &cert_signed_data, &cert_signature,
-				     issuer);
-	if (ret == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
-		gnutls_assert();
-		out |= GNUTLS_CERT_INVALID | GNUTLS_CERT_SIGNATURE_FAILURE;
-		/* error. ignore it */
-		if (output)
-			*output |= out;
-		result = 0;
-	} else if (ret < 0) {
-		result = 0;
-		gnutls_assert();
-		goto cleanup;
-	} else if (ret == 1)
-		result = 1;
+	if (sigalg >= 0) {
+		if (is_level_acceptable(cert, issuer, sigalg, flags) == 0) {
+			gnutls_assert();
+			out |=
+			    GNUTLS_CERT_INSECURE_ALGORITHM |
+			    GNUTLS_CERT_INVALID;
+			result = 0;
+		}
+
+		/* If the certificate is not self signed check if the algorithms
+		 * used are secure. If the certificate is self signed it doesn't
+		 * really matter.
+		 */
+		if (gnutls_sign_is_secure(sigalg) == 0 &&
+		    is_broken_allowed(sigalg, flags) == 0 &&
+		    is_issuer(cert, cert) == 0) {
+			gnutls_assert();
+			out |=
+			    GNUTLS_CERT_INSECURE_ALGORITHM |
+			    GNUTLS_CERT_INVALID;
+			result = 0;
+		}
+	}
 
 	/* Check activation/expiration times
 	 */
 	if (!(flags & GNUTLS_VERIFY_DISABLE_TIME_CHECKS)) {
 		/* check the time of the issuer first */
-		if (!(flags & GNUTLS_VERIFY_DISABLE_TRUSTED_TIME_CHECKS)) {
+		if (issuer != NULL &&
+		    !(flags & GNUTLS_VERIFY_DISABLE_TRUSTED_TIME_CHECKS)) {
 			out |= check_time_status(issuer, now);
 			if (out != 0) {
 				gnutls_assert();
 				result = 0;
-				if (output)
-					*output |= out;
 			}
 		}
 
@@ -797,12 +784,13 @@ verify_crt(gnutls_x509_crt_t cert,
 		if (out != 0) {
 			gnutls_assert();
 			result = 0;
-			if (output)
-				*output |= out;
 		}
 	}
 
       cleanup:
+	if (output)
+		*output |= out;
+
 	if (func) {
 		if (result == 0) {
 			out |= GNUTLS_CERT_INVALID;
@@ -1407,42 +1395,6 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 	if (tcas_size >= 1)
 		issuer = find_crl_issuer(crl, trusted_cas, tcas_size);
 
-	/* issuer is not in trusted certificate
-	 * authorities.
-	 */
-	if (issuer == NULL) {
-		gnutls_assert();
-		if (verify)
-			*verify |=
-			    GNUTLS_CERT_SIGNER_NOT_FOUND |
-			    GNUTLS_CERT_INVALID;
-		return 0;
-	}
-
-	if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
-		if (gnutls_x509_crt_get_ca_status(issuer, NULL) != 1) {
-			gnutls_assert();
-			if (verify)
-				*verify |=
-				    GNUTLS_CERT_SIGNER_NOT_CA |
-				    GNUTLS_CERT_INVALID;
-			return 0;
-		}
-
-		result =
-		    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
-		if (result >= 0) {
-			if (!(usage & GNUTLS_KEY_CRL_SIGN)) {
-				gnutls_assert();
-				if (verify)
-					*verify |=
-					    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
-					    | GNUTLS_CERT_INVALID;
-				return 0;
-			}
-		}
-	}
-
 	result =
 	    _gnutls_x509_get_signed_data(crl->crl, &crl->der, "tbsCertList",
 					 &crl_signed_data);
@@ -1475,21 +1427,54 @@ gnutls_x509_crl_verify(gnutls_x509_crl_t crl,
 
 	hash_algo = gnutls_sign_get_hash_algorithm(result);
 
-	result =
-	    _gnutls_x509_verify_data(mac_to_entry(hash_algo),
-				     &crl_signed_data, &crl_signature,
-				     issuer);
-	if (result == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
-		gnutls_assert();
-		/* error. ignore it */
-		if (verify)
-			*verify |= GNUTLS_CERT_SIGNATURE_FAILURE;
-		result = 0;
-	} else if (result < 0) {
+	/* issuer is not in trusted certificate
+	 * authorities.
+	 */
+	if (issuer == NULL) {
 		gnutls_assert();
 		if (verify)
-			*verify |= GNUTLS_CERT_INVALID;
-		goto cleanup;
+			*verify |=
+			    GNUTLS_CERT_SIGNER_NOT_FOUND |
+			    GNUTLS_CERT_INVALID;
+	} else {
+		if (!(flags & GNUTLS_VERIFY_DISABLE_CA_SIGN)) {
+			if (gnutls_x509_crt_get_ca_status(issuer, NULL) != 1) {
+				gnutls_assert();
+				if (verify)
+					*verify |=
+					    GNUTLS_CERT_SIGNER_NOT_CA |
+					    GNUTLS_CERT_INVALID;
+			}
+
+			result =
+			    gnutls_x509_crt_get_key_usage(issuer, &usage, NULL);
+			if (result >= 0) {
+				if (!(usage & GNUTLS_KEY_CRL_SIGN)) {
+					gnutls_assert();
+					if (verify)
+						*verify |=
+						    GNUTLS_CERT_SIGNER_CONSTRAINTS_FAILURE
+						    | GNUTLS_CERT_INVALID;
+				}
+			}
+		}
+
+		result =
+		    _gnutls_x509_verify_data(mac_to_entry(hash_algo),
+					     &crl_signed_data, &crl_signature,
+					     issuer);
+		if (result == GNUTLS_E_PK_SIG_VERIFY_FAILED) {
+			gnutls_assert();
+			/* error. ignore it */
+			if (verify)
+				*verify |= GNUTLS_CERT_SIGNATURE_FAILURE;
+			result = 0;
+		} else if (result < 0) {
+			gnutls_assert();
+			if (verify)
+				*verify |= GNUTLS_CERT_INVALID;
+			goto cleanup;
+		}
 	}
 
 	{
diff --git a/tests/test-chains.h b/tests/test-chains.h
index 28974e1..8e250cd 100644
--- a/tests/test-chains.h
+++ b/tests/test-chains.h
@@ -1366,9 +1366,11 @@ static struct
 } chains[] =
 {
   { "CVE-2014-0092", cve_2014_0092_check, &cve_2014_0092_check[1],
-    0, GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_INVALID },
+    0,
+    GNUTLS_CERT_SIGNER_NOT_CA | GNUTLS_CERT_NOT_ACTIVATED | GNUTLS_CERT_INVALID },
   { "CVE-2008-4989", cve_2008_4989_chain, &cve_2008_4989_chain[2],
-    0, GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
+    0,
+    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_EXPIRED | GNUTLS_CERT_INVALID },
   { "amazon.com ok", verisign_com_chain_g5, &verisign_com_chain_g5[4],
     GNUTLS_VERIFY_DISABLE_TIME_CHECKS | GNUTLS_PROFILE_TO_VFLAGS(GNUTLS_PROFILE_LOW),
     0 },
@@ -1404,9 +1406,10 @@ static struct
     GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
     0 },
   { "rsa-md5 fail", mayfirst_chain, &mayfirst_chain[1],
-    0, GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS,
+    GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "rsa-md5 not ok", mayfirst_chain, &mayfirst_chain[1],
-    GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD2,
+    GNUTLS_VERIFY_DISABLE_TIME_CHECKS | GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD2,
     GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "rsa-md5 not ok2", mayfirst_chain, &mayfirst_chain[1],
     GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD5,
@@ -1449,7 +1452,7 @@ static struct
     GNUTLS_VERIFY_ALLOW_SIGN_RSA_MD5, 0 },
   { "cacertrsamd5 short-cut not ok", cacertrsamd5, &cacertrsamd5[0],
     GNUTLS_VERIFY_DO_NOT_ALLOW_SAME,
-    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INVALID },
+    GNUTLS_CERT_SIGNER_NOT_FOUND | GNUTLS_CERT_INSECURE_ALGORITHM | GNUTLS_CERT_INVALID },
   { "cacertrsamd5 short-cut ok", cacertrsamd5, &cacertrsamd5[1],
     0, 0 },
   { "ecc cert ok", ecc_cert, &ecc_cert[1], GNUTLS_PROFILE_TO_VFLAGS(GNUTLS_PROFILE_HIGH), 0 },
-- 
2.1.0




From dkg at fifthhorseman.net  Wed Sep 17 00:31:42 2014
From: dkg at fifthhorseman.net (Daniel Kahn Gillmor)
Date: Tue, 16 Sep 2014 18:31:42 -0400
Subject: [gnutls-devel] certtool --crq-info should show signing algorithm
Message-ID: <5418BA4E.7090602@fifthhorseman.net>

when looking at a human-readable representation of a certificate signing
request (CSR), certtool --crq-info currently doesn't display the
algorithm used for the self-signature of the CSR.  (i.e. it doesn't
indicate if it was RSAwithSHA1 or RSAwithSHA256, etc)

It would occasionally be useful for people examining CSRs to see this
information (i was just asked for it privately, and i think it would be
useful too, e.g. when reviewing a CSR before submitting it to a CA).

Does this seem reasonable?

	--dkg

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 949 bytes
Desc: OpenPGP digital signature
URL: </pipermail/attachments/20140916/8768aab4/attachment.sig>

From nmav at gnutls.org  Wed Sep 17 10:08:17 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Wed, 17 Sep 2014 10:08:17 +0200
Subject: [gnutls-devel] certtool --crq-info should show signing algorithm
In-Reply-To: <5418BA4E.7090602@fifthhorseman.net>
References: <5418BA4E.7090602@fifthhorseman.net>
Message-ID: <CAJU7zaLBtiR-mvPiHP=Yq8a-cHUoHytUX=WG_nJdE8ZEJ_psVQ@mail.gmail.com>

On Wed, Sep 17, 2014 at 12:31 AM, Daniel Kahn Gillmor
<dkg at fifthhorseman.net> wrote:
> when looking at a human-readable representation of a certificate signing
> request (CSR), certtool --crq-info currently doesn't display the
> algorithm used for the self-signature of the CSR.  (i.e. it doesn't
> indicate if it was RSAwithSHA1 or RSAwithSHA256, etc)
> It would occasionally be useful for people examining CSRs to see this
> information (i was just asked for it privately, and i think it would be
> useful too, e.g. when reviewing a CSR before submitting it to a CA).
> Does this seem reasonable?

Hello Daniel,
 It certainly does.

regards,
Nikos


From nmav at gnutls.org  Wed Sep 17 13:12:08 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Wed, 17 Sep 2014 13:12:08 +0200
Subject: [gnutls-devel] [PATCH V3] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410890544.1622.11.camel@waverley>
References: <1410890544.1622.11.camel@waverley>
Message-ID: <CAJU7zaJnKPS=KbxuEQKE34=Fhk_n=Csz6m=sDjj0zT6CG5OVDA@mail.gmail.com>

On Tue, Sep 16, 2014 at 8:02 PM, Armin Burgmeier <armin at arbur.net> wrote:
> This allows to check for all possible flaws with a certificate chain with a
> single call to gnutls_x509_crt_list_verify and friends.

I've noticed that this causes some discrepancies between
pkcs11-chainverify and chainverify in the verification. That is not
much of your concern of course as it is about the PKCS#11 module
verification and I'll handle it, but it uncovered the following issue:
The verification of a chain works like that:
CA -> ICA1 -> ICA2 -> ENDCERT

To prevent a DoS (e.g. by a server or client that sends a list of 1000
certificates) it starts like (see _gnutls_verify_crt_status()):
1. verify: CA -> CA1
if (fail return status)

2. verify: CA1 -> CA2
if (fail return status)

3. verify: CA2 -> ENDCERT
if (fail return status)

So with your patch you'll get the status up to the point of first
failure. If the failure is in step 1 you'll get the full status for
CA->CA1 verification, but no flag will apply on ENDCERT. In your case
I think you verify against the scenario: CA -> ENDCERT, so you get
some reasonable flags. I don't know how reasonable these would be if
you are in a multiple CA scenario. Still it may make sense to do that
(in that case I should document that correctly), and I'm not sure
whether getting the flags of the 3 steps combined would offer much of
an advantage as they refer to multiple certificates. What do you think
of that? Is the current situation reasonable for your use case?

regards,
Nikos


From armin at arbur.net  Wed Sep 17 16:19:19 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 10:19:19 -0400
Subject: [gnutls-devel] [PATCH V3] Check for all error conditions when
 verifying a certificate
In-Reply-To: <CAJU7zaJnKPS=KbxuEQKE34=Fhk_n=Csz6m=sDjj0zT6CG5OVDA@mail.gmail.com>
References: <1410890544.1622.11.camel@waverley>
 <CAJU7zaJnKPS=KbxuEQKE34=Fhk_n=Csz6m=sDjj0zT6CG5OVDA@mail.gmail.com>
Message-ID: <1410963559.1336.11.camel@waverley>

Hi Nikos,

On Wed, 2014-09-17 at 13:12 +0200, Nikos Mavrogiannopoulos wrote:
> On Tue, Sep 16, 2014 at 8:02 PM, Armin Burgmeier <armin at arbur.net> wrote:
> > This allows to check for all possible flaws with a certificate chain with a
> > single call to gnutls_x509_crt_list_verify and friends.
> 
> I've noticed that this causes some discrepancies between
> pkcs11-chainverify and chainverify in the verification. That is not
> much of your concern of course as it is about the PKCS#11 module
> verification and I'll handle it,

Okay.

>  but it uncovered the following issue:
> The verification of a chain works like that:
> CA -> ICA1 -> ICA2 -> ENDCERT
> 
> To prevent a DoS (e.g. by a server or client that sends a list of 1000
> certificates) it starts like (see _gnutls_verify_crt_status()):
> 1. verify: CA -> CA1
> if (fail return status)
> 
> 2. verify: CA1 -> CA2
> if (fail return status)
> 
> 3. verify: CA2 -> ENDCERT
> if (fail return status)
> 
> So with your patch you'll get the status up to the point of first
> failure. If the failure is in step 1 you'll get the full status for
> CA->CA1 verification, but no flag will apply on ENDCERT. In your case
> I think you verify against the scenario: CA -> ENDCERT, so you get
> some reasonable flags. I don't know how reasonable these would be if
> you are in a multiple CA scenario. Still it may make sense to do that
> (in that case I should document that correctly), and I'm not sure
> whether getting the flags of the 3 steps combined would offer much of
> an advantage as they refer to multiple certificates. What do you think
> of that? Is the current situation reasonable for your use case?

Yes, I think it is reasonable. As you say in the scenario with
intermediate CAs, the verification flags would be for multiple
verifications combined, and therefore still lack some information. I
think it is fine if it is documented that:

a) the verification procedure stops if a failure has been found with one
certificate in the chain.
b) in that case all issues with that particular certificate are
reported.
c) the verification order starts from CA->CA1, then CA1->CA2, ..., then
CAn -> ENDCERT
d) if people need to know more details, they should run the verification
for each certificate in the chain individually.

Does it sound good to you?

Cheers,
Armin



From nmav at gnutls.org  Thu Sep 18 14:06:02 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 18 Sep 2014 14:06:02 +0200
Subject: [gnutls-devel] gnutls 3.2.18
Message-ID: <1411041962.22024.2.camel@nomad.lan>

Hello,
 I've just released gnutls 3.2.18. This is a bugfix release on the
current stable branch.

* Version 3.2.18 (released 2014-09-18)

** libgnutls: Fixes in gnutls_x509_crt_set_dn() and friends to properly handle
strings with embedded spaces and escaped commas.

** libgnutls: Corrected gnutls_x509_crl_verify() which would always report
a CRL signature as invalid. Reported by Armin Burgmeier.

** libgnutls: Fixed issue with certificates being sanitized by gnutls prior
to signature verification. That resulted to certain non-DER compliant modifications
of valid certificates, being corrected by libtasn1's parser and restructured as
the original. Issue found and reported by Antti Karjalainen and Matti Kamunen from
Codenomicon.

** API and ABI modifications:
No changes since last version.


Getting the Software
====================

GnuTLS may be downloaded directly from
<ftp://ftp.gnutls.org/gcrypt/gnutls/>.  A list of GnuTLS mirrors can be
found at <http://www.gnutls.org/download.html>.

Here are the XZ and LZIP compressed sources:

  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.lz

Here are OpenPGP detached signatures signed using key 0x96865171:

  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz.sig
  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.lz.sig

Note that it has been signed with my openpgp key:
pub   3104R/96865171 2008-05-04 [expires: 2028-04-29]
uid                  Nikos Mavrogiannopoulos <nmav <at> gnutls.org>
uid                  Nikos Mavrogiannopoulos <n.mavrogiannopoulos <at>
gmail.com>
sub   2048R/9013B842 2008-05-04 [expires: 2018-05-02]
sub   2048R/1404A91D 2008-05-04 [expires: 2018-05-02]

regards,
Nikos





From nmav at gnutls.org  Thu Sep 18 14:10:09 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 18 Sep 2014 14:10:09 +0200
Subject: [gnutls-devel] gnutls 3.3.8
Message-ID: <1411042209.22024.6.camel@nomad.lan>

Hello, 
 I've just released gnutls 3.3.8. This is a bug-fix release on
the next-stable branch. An important aspect of this release is that it
completes support for the p11-kit trust module, allowing gnutls to
utilize attached extensions in the system CA certificates (e.g. to
restrict a CA certificate to certain domain names, or for a specific
scope). I'll provide details on that feature on a follow up e-mail.


* Version 3.3.8 (released 2014-09-18)

** libgnutls: Updates in the name constraints checks. No name constraints
will be checked for intermediate certificates. As our support for name
constraints is limited to e-mail addresses in DNS names, it is pointless
to check them on intermediate certificates.

** libgnutls: Fixed issues in PKCS #11 object listing. Previously multiple
object listing would fail completely if a single object could not be exported.

** libgnutls: Improved the performance of PKCS #11 object listing/retrieving,
by retrieving them in large batches. Report and suggestion by David
Woodhouse.

** libgnutls: Fixed issue with certificates being sanitized by gnutls prior
to signature verification. That resulted to certain non-DER compliant modifications
of valid certificates, being corrected by libtasn1's parser and restructured as
the original. Issue found and reported by Antti Karjalainen and Matti Kamunen from
Codenomicon.

** libgnutls: Fixes in gnutls_x509_crt_set_dn() and friends to properly handle
strings with embedded spaces and escaped commas.

** libgnutls: when comparing a CA certificate with the trusted list compare
the name and key only instead of the whole certificate. That is to handle
cases where a CA certificate was superceded by a different one with the same
name and the same key.

** libgnutls: when verifying a certificate against a p11-kit trusted
module, use the attached extensions in the module to override the CA's
extensions (that requires p11-kit 0.20.7).

** libgnutls: In DTLS prevent sending zero-size fragments in certain cases
of MTU split. Reported by Manuel P?gouri?-Gonnard.

** libgnutls: Added gnutls_x509_trust_list_verify_crt2() which allows
verifying using a hostname and a purpose (extended key usage). That
enhances PKCS #11 trust module verification, as it can now check the purpose
when this function is used.

** libgnutls: Corrected gnutls_x509_crl_verify() which would always report
a CRL signature as invalid. Reported by Armin Burgmeier.

** libgnutls: added option --disable-padlock to allow disabling the padlock
CPU acceleration.

** p11tool: when listing tokens, list their type as well.

** p11tool: when listing objects from a trust module print any attached
extensions on certificates.

** API and ABI modifications:
gnutls_x509_crq_get_extension_by_oid2: Added
gnutls_x509_crt_get_extension_by_oid2: Added
gnutls_x509_trust_list_verify_crt2: Added
gnutls_x509_ext_print: Added
gnutls_x509_ext_deinit: Added
gnutls_x509_othername_to_virtual: Added
gnutls_pkcs11_obj_get_exts: Added


Getting the Software
====================

GnuTLS may be downloaded directly from
<ftp://ftp.gnutls.org/gcrypt/gnutls/>.  A list of GnuTLS mirrors can be
found at <http://www.gnutls.org/download.html>.

Here are the XZ and LZIP compressed sources:

  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.3/gnutls-3.3.8.tar.xz
  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.3/gnutls-3.3.8.tar.lz

Here are OpenPGP detached signatures signed using key 0x96865171:

  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.3/gnutls-3.3.8.tar.xz.sig
  ftp://ftp.gnutls.org/gcrypt/gnutls/v3.3/gnutls-3.3.8.tar.lz.sig

Note that it has been signed with my openpgp key:
pub   3104R/96865171 2008-05-04 [expires: 2028-04-29]
uid                  Nikos Mavrogiannopoulos <nmav <at> gnutls.org>
uid                  Nikos Mavrogiannopoulos <n.mavrogiannopoulos <at>
gmail.com>
sub   2048R/9013B842 2008-05-04 [expires: 2018-05-02]
sub   2048R/1404A91D 2008-05-04 [expires: 2018-05-02]

regards,
Nikos






From nmav at gnutls.org  Thu Sep 18 10:53:46 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 18 Sep 2014 10:53:46 +0200
Subject: [gnutls-devel] [PATCH V3] Check for all error conditions when
 verifying a certificate
In-Reply-To: <1410963559.1336.11.camel@waverley>
References: <1410890544.1622.11.camel@waverley>
 <CAJU7zaJnKPS=KbxuEQKE34=Fhk_n=Csz6m=sDjj0zT6CG5OVDA@mail.gmail.com>
 <1410963559.1336.11.camel@waverley>
Message-ID: <CAJU7zaKgj5j_y0MTiQbYHDifO-tzBhrMpDZpC2=m39o4P5p_=g@mail.gmail.com>

On Wed, Sep 17, 2014 at 4:19 PM, Armin Burgmeier <armin at arbur.net> wrote:

>> So with your patch you'll get the status up to the point of first
>> failure. If the failure is in step 1 you'll get the full status for
>> CA->CA1 verification, but no flag will apply on ENDCERT. In your case
>> I think you verify against the scenario: CA -> ENDCERT, so you get
>> some reasonable flags. I don't know how reasonable these would be if
>> you are in a multiple CA scenario. Still it may make sense to do that
>> (in that case I should document that correctly), and I'm not sure
>> whether getting the flags of the 3 steps combined would offer much of
>> an advantage as they refer to multiple certificates. What do you think
>> of that? Is the current situation reasonable for your use case?
> Yes, I think it is reasonable. As you say in the scenario with
> intermediate CAs, the verification flags would be for multiple
> verifications combined, and therefore still lack some information. I
> think it is fine if it is documented that:
> a) the verification procedure stops if a failure has been found with one
> certificate in the chain.
> b) in that case all issues with that particular certificate are
> reported.
> c) the verification order starts from CA->CA1, then CA1->CA2, ..., then
> CAn -> ENDCERT
> d) if people need to know more details, they should run the verification
> for each certificate in the chain individually.

I've applied the patch as well some documentation in master. The plan
is to be included in the 3.4.0 branch.

regards,
Nikos


From Ondrej.Caletka at cesnet.cz  Thu Sep 18 09:48:04 2014
From: Ondrej.Caletka at cesnet.cz (=?UTF-8?B?T25kxZllaiBDYWxldGth?=)
Date: Thu, 18 Sep 2014 09:48:04 +0200
Subject: [gnutls-devel] DANE: Verification failed for TLSA record with
	usage=0
Message-ID: <541A8E34.1090501@cesnet.cz>

Hello,

I've deployed TLSA records for www.cesnet.cz. I've used swede utility to
constrain allowed CA and according to swede, the record is valid:

$ ./swede verify www.cesnet.cz
Received the following record for name _443._tcp.www.cesnet.cz.:
        Usage:                          0 (CA Constraint)
        Selector:                       0 (Certificate)
        Matching Type:                  1 (SHA-256)
        Certificate for Association:
5c428b013b2e3f0d30abb5bebd92d066dc06dc223329eb0fc735609946cf8e1c
This record is valid (well-formed).
Attempting to verify the record with the TLS service...
Got the following IP: 195.113.144.230
SUCCESS (Usage 0): A certificate in the certificate chain offered by the
server matches the one mentioned in the TLSA record and is a CA certificate
The matched certificate has Subject: /C=NL/O=TERENA/CN=TERENA SSL CA

However, ehen trying to validate this TLSA record using danetool from
GnuTLS 3.3.7, I get validation error:

$ danetool --check www.cesnet.cz --proto tcp --port 443
Resolving 'www.cesnet.cz'...
Obtaining certificate from '2001:718:1:101::4:443'...
Querying DNS for www.cesnet.cz (tcp:443)...
_443._tcp.www.cesnet.cz. IN TLSA ( 00 00 01
5c428b013b2e3f0d30abb5bebd92d066dc06dc223329eb0fc735609946cf8e1c )
Certificate usage: CA (00)
Certificate type:  X.509 (00)
Contents:          SHA2-256 hash (01)
Data:
5c428b013b2e3f0d30abb5bebd92d066dc06dc223329eb0fc735609946cf8e1c

Verification: Verification failed. CA constrains were violated.


I believe this is a bug in GnuTLS. Other DANE implementations sees the
TLSA record as valid.

-- 
Best regards

Ond?ej Caletka
CESNET

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5563 bytes
Desc: Elektronicky podpis S/MIME
URL: </pipermail/attachments/20140918/ed4ff7bb/attachment.bin>

From armin at arbur.net  Wed Sep 17 18:30:44 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 12:30:44 -0400
Subject: [gnutls-devel] [PATCH 4/9] Add a function to obtain the trust list
 of a gnutls_certificate_credentials_t
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <e0647f69b7239f47b79e6d5ba7ebbbb176e9e425.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/gnutls_x509.c          | 18 ++++++++++++++++++
 lib/includes/gnutls/x509.h |  3 +++
 lib/libgnutls.map          |  1 +
 3 files changed, 22 insertions(+)

diff --git a/lib/gnutls_x509.c b/lib/gnutls_x509.c
index b102f4d..f57c6b9 100644
--- a/lib/gnutls_x509.c
+++ b/lib/gnutls_x509.c
@@ -1244,6 +1244,24 @@ gnutls_certificate_set_trust_list(gnutls_certificate_credentials_t res,
 	res->tlist = tlist;
 }
 
+/**
+ * gnutls_certificate_get_trust_list:
+ * @res: is a #gnutls_certificate_credentials_t structure.
+ * @tlist: Location where to store the trust list.
+ *
+ * Obtains the list of trusted certificates stored in @res and writes a
+ * pointer to it to the location @tlist. The pointer will point to memory
+ * internal to @res, and must not be deinitialized. It will be automatically
+ * deallocated when the @res structure is deinitialized.
+ *
+ * Since: 3.4.0
+ */
+void
+gnutls_certificate_get_trust_list(gnutls_certificate_credentials_t res,
+                                  gnutls_x509_trust_list_t *tlist)
+{
+	*tlist = res->tlist;
+}
 
 /**
  * gnutls_certificate_set_x509_key_file:
diff --git a/lib/includes/gnutls/x509.h b/lib/includes/gnutls/x509.h
index d86b0ef..fd02f0c 100644
--- a/lib/includes/gnutls/x509.h
+++ b/lib/includes/gnutls/x509.h
@@ -1354,6 +1354,9 @@ gnutls_x509_trust_list_add_system_trust(gnutls_x509_trust_list_t
 void gnutls_certificate_set_trust_list
     (gnutls_certificate_credentials_t res,
      gnutls_x509_trust_list_t tlist, unsigned flags);
+void gnutls_certificate_get_trust_list
+    (gnutls_certificate_credentials_t res,
+     gnutls_x509_trust_list_t *tlist);
 
 typedef struct gnutls_x509_ext_st {
 	char *oid;
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index f3a0582..b1121d0 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -899,6 +899,7 @@ GNUTLS_3_1_0 {
 	gnutls_alpn_set_protocols;
 	gnutls_session_set_id;
 	gnutls_certificate_set_trust_list;
+	gnutls_certificate_get_trust_list;
 	gnutls_cipher_get_tag_size;
 	gnutls_record_overhead_size;
 	gnutls_handshake_set_hook_function;
-- 
2.1.0



From armin at arbur.net  Thu Sep 18 00:59:29 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 18:59:29 -0400
Subject: [gnutls-devel] [PATCH 7/9] Add functions to obtain X.509 keys and
 certificates from certificate credentials
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <5cdfe2388a105d4df6e0d501f67b77e4a9236b1a.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/gnutls_x509.c               | 96 +++++++++++++++++++++++++++++++++++++++++
 lib/includes/gnutls/gnutls.h.in |  8 ++++
 lib/libgnutls.map               |  2 +
 3 files changed, 106 insertions(+)

diff --git a/lib/gnutls_x509.c b/lib/gnutls_x509.c
index f57c6b9..5b9ab1a 100644
--- a/lib/gnutls_x509.c
+++ b/lib/gnutls_x509.c
@@ -1138,6 +1138,102 @@ gnutls_certificate_set_x509_key(gnutls_certificate_credentials_t res,
 }
 
 /**
+ * gnutls_certificate_get_x509_key:
+ * @res: is a #gnutls_certificate_credentials_t structure.
+ * @index: The index of the key to obtain.
+ * @key: Location to store the key.
+ *
+ * Obtains a X.509 private key that has been stored in @res with one of
+ * gnutls_certificate_set_x509_key(), gnutls_certificate_set_key(),
+ * gnutls_certificate_set_x509_key_file(),
+ * gnutls_certificate_set_x509_key_file2(),
+ * gnutls_certificate_set_x509_key_mem(), or
+ * gnutls_certificate_set_x509_key_mem2(). The returned key must be deallocated
+ * with gnutls_x509_privkey_deinit() when no longer needed.
+ *
+ * If there is no key with the given index,
+ * %GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE is returned. If the key with the
+ * given index is not a X.509 key, %GNUTLS_E_INVALID_REQUEST is returned.
+ *
+ * Returns: %GNUTLS_E_SUCCESS (0) on success, or a negative error code.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_certificate_get_x509_key(gnutls_certificate_credentials_t res,
+                                int index,
+                                gnutls_x509_privkey_t *key)
+{
+	if (index >= res->ncerts) {
+		gnutls_assert();
+		return GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE;
+	}
+
+	return gnutls_privkey_export_x509(res->pkey[index], key);
+}
+
+/**
+ * gnutls_certificate_get_x509_crt:
+ * @res: is a #gnutls_certificate_credentials_t structure.
+ * @index: The index of the certificate list to obtain.
+ * @crt_list: Where to store the certificate list.
+ * @key: Will hold the number of certificates.
+ *
+ * Obtains a X.509 certificate list that has been stored in @res with one of
+ * gnutls_certificate_set_x509_key(), gnutls_certificate_set_key(),
+ * gnutls_certificate_set_x509_key_file(),
+ * gnutls_certificate_set_x509_key_file2(),
+ * gnutls_certificate_set_x509_key_mem(), or
+ * gnutls_certificate_set_x509_key_mem2(). Each certificate in the returned
+ * certificate list must be deallocated with gnutls_x509_crt_deinit(), and the
+ * list itself must be freed with gnutls_free().
+ *
+ * If there is no certificate with the given index,
+ * %GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE is returned. If the certificate
+ * with the given index is not a X.509 certificate, %GNUTLS_E_INVALID_REQUEST
+ * is returned.
+ *
+ * Returns: %GNUTLS_E_SUCCESS (0) on success, or a negative error code.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_certificate_get_x509_crt(gnutls_certificate_credentials_t res,
+                                int index,
+                                gnutls_x509_crt_t **crt_list,
+                                int *crt_list_size)
+{
+	int ret, i;
+
+	if (index >= res->ncerts) {
+		gnutls_assert();
+		return GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE;
+	}
+
+	*crt_list_size = res->certs[index].cert_list_length;
+	*crt_list = gnutls_malloc(
+		res->certs[index].cert_list_length * sizeof (gnutls_x509_crt_t));
+	if (*crt_list == NULL) {
+		gnutls_assert();
+		return GNUTLS_E_MEMORY_ERROR;
+	}
+
+	for (i = 0; i < res->certs[index].cert_list_length; ++i) {
+		ret = gnutls_pcert_export_x509(&res->certs[index].cert_list[i], crt_list[i]);
+		if (ret < 0) {
+			while (i--)
+				gnutls_x509_crt_deinit(*crt_list[i]);
+			gnutls_free(*crt_list);
+			*crt_list = NULL;
+
+			return gnutls_assert_val(ret);
+		}
+	}
+
+	return 0;
+}
+
+/**
  * gnutls_certificate_set_key:
  * @res: is a #gnutls_certificate_credentials_t structure.
  * @names: is an array of DNS name of the certificate (NULL if none)
diff --git a/lib/includes/gnutls/gnutls.h.in b/lib/includes/gnutls/gnutls.h.in
index 17ff8fc..b06381a 100644
--- a/lib/includes/gnutls/gnutls.h.in
+++ b/lib/includes/gnutls/gnutls.h.in
@@ -1429,6 +1429,14 @@ int gnutls_certificate_set_x509_crl(gnutls_certificate_credentials_t res,
 				    gnutls_x509_crl_t * crl_list,
 				    int crl_list_size);
 
+int gnutls_certificate_get_x509_key(gnutls_certificate_credentials_t res,
+                                    int index,
+                                    gnutls_x509_privkey_t *key);
+int gnutls_certificate_get_x509_crt(gnutls_certificate_credentials_t res,
+                                    int index,
+                                    gnutls_x509_crt_t **crt_list,
+                                    int *crt_list_size);
+
   /* OCSP status request extension, RFC 6066 */
 typedef int (*gnutls_status_request_ocsp_func)
  (gnutls_session_t session, void *ptr, gnutls_datum_t * ocsp_response);
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index b64ff13..0e21496 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -1013,6 +1013,8 @@ GNUTLS_3_1_0 {
 	gnutls_openpgp_crt_check_hostname2;
 	gnutls_certificate_verify_peers;
 	gnutls_certificate_get_verify_flags;
+	gnutls_certificate_get_x509_key;
+	gnutls_certificate_get_x509_crt;
 	gnutls_credentials_get;
 	gnutls_x509_crl_iter_crt_serial;
 	gnutls_x509_crl_iter_deinit;
-- 
2.1.0



From armin at arbur.net  Wed Sep 17 18:31:19 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 12:31:19 -0400
Subject: [gnutls-devel] [PATCH 1/9] Fix a documentation typo
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <3f8b5db51a40252d380b189d3e3f977a5e42dd93.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/gnutls_ui.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/gnutls_ui.c b/lib/gnutls_ui.c
index dbee667..03610e8 100644
--- a/lib/gnutls_ui.c
+++ b/lib/gnutls_ui.c
@@ -484,7 +484,7 @@ const gnutls_datum_t *gnutls_certificate_get_ours(gnutls_session_t session)
  * In case of OpenPGP keys a single key will be returned in raw
  * format.
  *
- * Returns: a pointer to a #gnutls_datum_t containing our
+ * Returns: a pointer to a #gnutls_datum_t containing the peer's
  *   certificates, or %NULL in case of an error or if no certificate
  *   was used.
  **/
-- 
2.1.0



From armin at arbur.net  Thu Sep 18 17:22:35 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Thu, 18 Sep 2014 11:22:35 -0400
Subject: [gnutls-devel] [PATCH 9/9] Add an interface to iterate the trusted
 CA certificates in a trust list
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <ee71a78c012e6871ec9e435db509d3a9d6955220.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/includes/gnutls/x509.h |  9 +++++
 lib/libgnutls.map          |  2 ++
 lib/x509/verify-high.c     | 88 ++++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 99 insertions(+)

diff --git a/lib/includes/gnutls/x509.h b/lib/includes/gnutls/x509.h
index fd02f0c..c30eca7 100644
--- a/lib/includes/gnutls/x509.h
+++ b/lib/includes/gnutls/x509.h
@@ -1234,6 +1234,7 @@ gnutls_x509_crt_get_extension_by_oid2(gnutls_x509_crt_t cert,
 				     unsigned int *critical);
 
 typedef struct gnutls_x509_trust_list_st *gnutls_x509_trust_list_t;
+typedef struct gnutls_x509_trust_list_iter *gnutls_x509_trust_list_iter_t;
 
 int
 gnutls_x509_trust_list_init(gnutls_x509_trust_list_t * list,
@@ -1275,6 +1276,14 @@ gnutls_x509_trust_list_add_crls(gnutls_x509_trust_list_t list,
 				unsigned int flags,
 				unsigned int verification_flags);
 
+
+int
+gnutls_x509_trust_list_iter_get_ca(gnutls_x509_trust_list_t list,
+                                   gnutls_x509_trust_list_iter_t *iter,
+                                   gnutls_x509_crt_t *crt);
+
+void gnutls_x509_trust_list_iter_deinit(gnutls_x509_trust_list_iter_t iter);
+
 typedef int gnutls_verify_output_function(gnutls_x509_crt_t cert, gnutls_x509_crt_t issuer,	/* The issuer if verification failed 
 												 * because of him. might be null.
 												 */
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index 3a2263f..1d9aa36 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -788,6 +788,8 @@ GNUTLS_3_1_0 {
 	gnutls_x509_trust_list_add_system_trust;
 	gnutls_x509_trust_list_add_trust_file;
 	gnutls_x509_trust_list_add_trust_mem;
+	gnutls_x509_trust_list_iter_get_ca;
+	gnutls_x509_trust_list_iter_deinit;
 	gnutls_pkcs12_simple_parse;
 	gnutls_privkey_import_openpgp_raw;
 	gnutls_privkey_import_x509_raw;
diff --git a/lib/x509/verify-high.c b/lib/x509/verify-high.c
index 065920d..9b4ac8c 100644
--- a/lib/x509/verify-high.c
+++ b/lib/x509/verify-high.c
@@ -54,6 +54,11 @@ struct node_st {
 
 };
 
+struct gnutls_x509_trust_list_iter {
+	unsigned int node_index;
+	unsigned int ca_index;
+};
+
 #define DEFAULT_SIZE 127
 
 /**
@@ -289,6 +294,89 @@ gnutls_x509_trust_list_add_cas(gnutls_x509_trust_list_t list,
 	return i;
 }
 
+/**
+ * gnutls_x509_trust_list_iter_get_ca:
+ * @list: The structure of the list
+ * @iter: A pointer to an iterator (initially the iterator should be %NULL)
+ * @crt: where the certificate will be copied
+ *
+ * This function obtains a certificate in the trust list and advances the
+ * iterator to the next certificate. The certificate returned in @crt must be
+ * deallocated with gnutls_x509_crt_deinit().
+ *
+ * When past the last element is accessed %GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE
+ * is returned and the iterator is reset.
+ *
+ * After use, the iterator must be deinitialized usin
+ *  gnutls_x509_trust_list_iter_deinit().
+ *
+ * Returns: On success, %GNUTLS_E_SUCCESS (0) is returned, otherwise a
+ *   negative error value.
+ *
+ * Since: 3.4.0
+ **/
+int
+gnutls_x509_trust_list_iter_get_ca(gnutls_x509_trust_list_t list,
+                                   gnutls_x509_trust_list_iter_t *iter,
+                                   gnutls_x509_crt_t *crt)
+{
+	int ret;
+
+	/* advance to next entry */
+	if (*iter == NULL) {
+		*iter = gnutls_malloc(sizeof (struct gnutls_x509_trust_list_iter));
+		if (*iter == NULL)
+			return gnutls_assert_val(GNUTLS_E_MEMORY_ERROR);
+
+		(*iter)->node_index = 0;
+		(*iter)->ca_index = 0;
+	} else {
+		++(*iter)->ca_index;
+	}
+
+	/* skip empty nodes */
+	while ((*iter)->ca_index >= list->node[(*iter)->node_index].trusted_ca_size) {
+		++(*iter)->node_index;
+		(*iter)->ca_index = 0;
+
+		if ((*iter)->node_index >= list->size) {
+			gnutls_free(*iter);
+			*iter = NULL;
+
+			*crt = NULL;
+			return gnutls_assert_val(GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE);
+		}
+	}
+
+	ret = gnutls_x509_crt_init(crt);
+	if (ret < 0)
+		return gnutls_assert_val(ret);
+
+	ret = _gnutls_x509_crt_cpy(*crt, list->node[(*iter)->node_index].trusted_cas[(*iter)->ca_index]);
+	if (ret < 0) {
+		gnutls_x509_crt_deinit(*crt);
+		return gnutls_assert_val(ret);
+	}
+
+	return 0;
+}
+
+/**
+ * gnutls_x509_trust_list_iter_deinit:
+ * @iter: The iterator structure to be deinitialized
+ *
+ * This function will deinitialize an iterator structure.
+ *
+ * Since: 3.4.0
+ **/
+void gnutls_x509_trust_list_iter_deinit(gnutls_x509_trust_list_iter_t iter)
+{
+	if (!iter)
+		return;
+
+	gnutls_free(iter);
+}
+
 static gnutls_x509_crt_t crt_cpy(gnutls_x509_crt_t src)
 {
 gnutls_x509_crt_t dst;
-- 
2.1.0



From armin at arbur.net  Thu Sep 18 17:22:50 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Thu, 18 Sep 2014 11:22:50 -0400
Subject: [gnutls-devel] [PATCH 2/9] Memory leak fix on certificate copy
	failure
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <d421f45dd630036c8ed27f7aff50d54250d5d24a.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/x509/verify-high.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/lib/x509/verify-high.c b/lib/x509/verify-high.c
index e7faccc..065920d 100644
--- a/lib/x509/verify-high.c
+++ b/lib/x509/verify-high.c
@@ -302,6 +302,7 @@ int ret;
 
 	ret = _gnutls_x509_crt_cpy(dst, src);
 	if (ret < 0) {
+		gnutls_x509_crt_deinit(dst);
 		gnutls_assert();
 		return NULL;
 	}
-- 
2.1.0



From armin at arbur.net  Thu Sep 18 16:13:55 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Thu, 18 Sep 2014 10:13:55 -0400
Subject: [gnutls-devel] [PATCH 8/9] Add getter functions for openpgp keys
	and certificates
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <16bf9a14ffd998451b0ea6c0b96bdc46635e6737.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/includes/gnutls/openpgp.h | 10 +++++
 lib/libgnutls.map             |  2 +
 lib/openpgp/gnutls_openpgp.c  | 98 +++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 110 insertions(+)

diff --git a/lib/includes/gnutls/openpgp.h b/lib/includes/gnutls/openpgp.h
index 11958b3..b7b64ad 100644
--- a/lib/includes/gnutls/openpgp.h
+++ b/lib/includes/gnutls/openpgp.h
@@ -333,6 +333,16 @@ int gnutls_certificate_set_openpgp_key
      gnutls_openpgp_crt_t crt, gnutls_openpgp_privkey_t pkey);
 
 int
+gnutls_certificate_get_openpgp_key(gnutls_certificate_credentials_t res,
+                                   int index,
+                                   gnutls_openpgp_privkey_t *key);
+int
+gnutls_certificate_get_openpgp_crt(gnutls_certificate_credentials_t res,
+                                   int index,
+                                   gnutls_openpgp_crt_t **crt_list,
+                                   int *crt_list_size);
+
+int
  gnutls_certificate_set_openpgp_key_file
     (gnutls_certificate_credentials_t res, const char *certfile,
      const char *keyfile, gnutls_openpgp_crt_fmt_t format);
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index 0e21496..3a2263f 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -1015,6 +1015,8 @@ GNUTLS_3_1_0 {
 	gnutls_certificate_get_verify_flags;
 	gnutls_certificate_get_x509_key;
 	gnutls_certificate_get_x509_crt;
+	gnutls_certificate_get_openpgp_key;
+	gnutls_certificate_get_openpgp_crt;
 	gnutls_credentials_get;
 	gnutls_x509_crl_iter_crt_serial;
 	gnutls_x509_crl_iter_deinit;
diff --git a/lib/openpgp/gnutls_openpgp.c b/lib/openpgp/gnutls_openpgp.c
index 7c05e1f..bbb101e 100644
--- a/lib/openpgp/gnutls_openpgp.c
+++ b/lib/openpgp/gnutls_openpgp.c
@@ -166,6 +166,104 @@ gnutls_certificate_set_openpgp_key(gnutls_certificate_credentials_t res,
 	return ret;
 }
 
+/**
+ * gnutls_certificate_get_openpgp_key:
+ * @res: is a #gnutls_certificate_credentials_t structure.
+ * @index: The index of the key to obtain.
+ * @key: Location to store the key.
+ *
+ * Obtains a OpenPGP private key that has been stored in @res with one of
+ * gnutls_certificate_set_openpgp_key(),
+ * gnutls_certificate_set_openpgp_key_file(),
+ * gnutls_certificate_set_openpgp_key_file2(),
+ * gnutls_certificate_set_openpgp_key_mem(), or
+ * gnutls_certificate_set_openpgp_key_mem2().
+ * The returned key must be deallocated with gnutls_openpgp_privkey_deinit()
+ * when no longer needed.
+ *
+ * If there is no key with the given index,
+ * %GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE is returned. If the key with the
+ * given index is not a X.509 key, %GNUTLS_E_INVALID_REQUEST is returned.
+ *
+ * Returns: %GNUTLS_E_SUCCESS (0) on success, or a negative error code.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_certificate_get_openpgp_key(gnutls_certificate_credentials_t res,
+                                   int index,
+                                   gnutls_openpgp_privkey_t *key)
+{
+	if (index >= res->ncerts) {
+		gnutls_assert();
+		return GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE;
+	}
+
+	return gnutls_privkey_export_openpgp(res->pkey[index], key);
+}
+
+/**
+ * gnutls_certificate_get_openpgp_crt:
+ * @res: is a #gnutls_certificate_credentials_t structure.
+ * @index: The index of the certificate list to obtain.
+ * @crt_list: Where to store the certificate list.
+ * @key: Will hold the number of certificates.
+ *
+ * Obtains a X.509 certificate list that has been stored in @res with one of
+ * gnutls_certificate_set_openpgp_key(),
+ * gnutls_certificate_set_openpgp_key_file(),
+ * gnutls_certificate_set_openpgp_key_file2(),
+ * gnutls_certificate_set_openpgp_key_mem(), or
+ * gnutls_certificate_set_openpgp_key_mem2().  Each certificate in the
+ * returned certificate list must be deallocated with
+ * gnutls_openpgp_crt_deinit(), and the list itself must be freed with
+ * gnutls_free().
+ *
+ * If there is no certificate with the given index,
+ * %GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE is returned. If the certificate
+ * with the given index is not a X.509 certificate, %GNUTLS_E_INVALID_REQUEST
+ * is returned.
+ *
+ * Returns: %GNUTLS_E_SUCCESS (0) on success, or a negative error code.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_certificate_get_openpgp_crt(gnutls_certificate_credentials_t res,
+                                   int index,
+                                   gnutls_openpgp_crt_t **crt_list,
+                                   int *crt_list_size)
+{
+	int ret, i;
+
+	if (index >= res->ncerts) {
+		gnutls_assert();
+		return GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE;
+	}
+
+	*crt_list_size = res->certs[index].cert_list_length;
+	*crt_list = gnutls_malloc(
+		res->certs[index].cert_list_length * sizeof (gnutls_openpgp_crt_t));
+	if (*crt_list == NULL) {
+		gnutls_assert();
+		return GNUTLS_E_MEMORY_ERROR;
+	}
+
+	for (i = 0; i < res->certs[index].cert_list_length; ++i) {
+		ret = gnutls_pcert_export_openpgp(&res->certs[index].cert_list[i], crt_list[i]);
+		if (ret < 0) {
+			while (i--)
+				gnutls_openpgp_crt_deinit(*crt_list[i]);
+			gnutls_free(*crt_list);
+			*crt_list = NULL;
+
+			return gnutls_assert_val(ret);
+		}
+	}
+
+	return 0;
+}
+
 /*-
  * gnutls_openpgp_get_key:
  * @key: the destination context to save the key.
-- 
2.1.0



From armin at arbur.net  Wed Sep 17 18:26:47 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 12:26:47 -0400
Subject: [gnutls-devel] [PATCH 3/9] Add API to retrieve a X.509 or OpenPGP
 certificate from a gnutls_pcert_t
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <05bb9ee6dbff87eecfe32565f451848fc2d5be38.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/gnutls_pcert.c             | 100 +++++++++++++++++++++++++++++++++++++++++
 lib/includes/gnutls/abstract.h |   6 +++
 lib/libgnutls.map              |   3 ++
 3 files changed, 109 insertions(+)

diff --git a/lib/gnutls_pcert.c b/lib/gnutls_pcert.c
index 83fbfda..a0510f5 100644
--- a/lib/gnutls_pcert.c
+++ b/lib/gnutls_pcert.c
@@ -339,6 +339,106 @@ int gnutls_pcert_import_openpgp_raw(gnutls_pcert_st * pcert,
 #endif
 
 /**
+ * gnutls_pcert_get_type:
+ * @pcert: The pcert structure.
+ *
+ * Returns the certificate type of @pcert, one of X.509 or OpenPGP.
+ *
+ * Returns: The certificate type.
+ *
+ * Since: 3.4.0
+ */
+gnutls_certificate_type_t
+gnutls_pcert_get_type(gnutls_pcert_st * pcert)
+{
+	return pcert->type;
+}
+
+/**
+ * gnutls_pcert_export_x509:
+ * @pcert: The pcert structure.
+ * @crt: An initialized #gnutls_x509_crt_t.
+ *
+ * Converts the given #gnutls_pcert_t structure into a #gnutls_x509_crt_t.
+ * This function only works if the type of @pcert is %GNUTLS_CRT_X509.
+ * When successful, the value written to @crt must be freed with
+ * gnutls_x509_crt_deinit() when no longer needed.
+ *
+ * Returns: On success, %GNUTLS_E_SUCCESS (0) is returned, otherwise a
+ * negative error value.
+ *
+ * Since: 3.4.0
+ */
+int gnutls_pcert_export_x509(gnutls_pcert_st * pcert,
+                             gnutls_x509_crt_t * crt)
+{
+	int ret;
+
+	if (pcert->type != GNUTLS_CRT_X509) {
+		gnutls_assert();
+		return GNUTLS_E_INVALID_REQUEST;
+	}
+
+	ret = gnutls_x509_crt_init(crt);
+	if (ret < 0)
+		return gnutls_assert_val(ret);
+
+	ret = gnutls_x509_crt_import(*crt, &pcert->cert, GNUTLS_X509_FMT_DER);
+	if (ret < 0) {
+		gnutls_x509_crt_deinit(*crt);
+		*crt = NULL;
+
+		return gnutls_assert_val(ret);
+	}
+
+	return 0;
+}
+
+#ifdef ENABLE_OPENPGP
+
+/**
+ * gnutls_pcert_export_x509:
+ * @pcert: The pcert structure.
+ * @crt: An initialized #gnutls_openpgp_crt_t.
+ *
+ * Converts the given #gnutls_pcert_t structure into a #gnutls_openpgp_crt_t.
+ * This function only works if the type of @pcert is %GNUTLS_CRT_OPENPGP.
+ * When successful, the value written to @crt must be freed with
+ * gnutls_openpgp_crt_deinit() when no longer needed.
+ *
+ * Returns: On success, %GNUTLS_E_SUCCESS (0) is returned, otherwise a
+ * negative error value.
+ *
+ * Since: 3.4.0
+ */
+int gnutls_pcert_export_openpgp(gnutls_pcert_st * pcert,
+                                gnutls_openpgp_crt_t * crt)
+{
+	int ret;
+
+	if (pcert->type != GNUTLS_CRT_OPENPGP) {
+		gnutls_assert();
+		return GNUTLS_E_INVALID_REQUEST;
+	}
+
+	ret = gnutls_openpgp_crt_init(crt);
+	if (ret < 0)
+		return gnutls_assert_val(ret);
+
+	ret = gnutls_openpgp_crt_import(*crt, &pcert->cert, GNUTLS_OPENPGP_FMT_RAW);
+	if (ret < 0) {
+		gnutls_openpgp_crt_deinit(*crt);
+		*crt = NULL;
+
+		return gnutls_assert_val(ret);
+	}
+
+	return 0;
+}
+
+#endif
+
+/**
  * gnutls_pcert_deinit:
  * @pcert: The structure to be deinitialized
  *
diff --git a/lib/includes/gnutls/abstract.h b/lib/includes/gnutls/abstract.h
index d9aa560..835fdb7 100644
--- a/lib/includes/gnutls/abstract.h
+++ b/lib/includes/gnutls/abstract.h
@@ -437,6 +437,9 @@ typedef struct gnutls_pcert_st {
 int gnutls_pcert_import_x509(gnutls_pcert_st * pcert,
 			     gnutls_x509_crt_t crt, unsigned int flags);
 
+int gnutls_pcert_export_x509(gnutls_pcert_st * pcert,
+                             gnutls_x509_crt_t * crt);
+
 int
 gnutls_pcert_list_import_x509_raw(gnutls_pcert_st * pcerts,
 				  unsigned int *pcert_max,
@@ -460,6 +463,9 @@ int gnutls_pcert_import_openpgp(gnutls_pcert_st * pcert,
 				gnutls_openpgp_crt_t crt,
 				unsigned int flags);
 
+int gnutls_pcert_export_openpgp(gnutls_pcert_st * pcert,
+                                gnutls_openpgp_crt_t * crt);
+
 void gnutls_pcert_deinit(gnutls_pcert_st * pcert);
 
 /* For certificate credentials */
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index eabf261..f3a0582 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -684,6 +684,9 @@ GNUTLS_3_0_0 {
 	gnutls_pcert_import_x509_raw;
 	gnutls_pcert_import_openpgp;
 	gnutls_pcert_import_openpgp_raw;
+	gnutls_pcert_get_type;
+	gnutls_pcert_export_x509;
+	gnutls_pcert_export_openpgp;
 	gnutls_pubkey_get_openpgp_key_id;
 	gnutls_certificate_set_retrieve_function2;
 	gnutls_x509_trust_list_get_issuer;
-- 
2.1.0



From armin at arbur.net  Wed Sep 17 23:33:40 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 17:33:40 -0400
Subject: [gnutls-devel] [PATCH 5/9] Add functions to export X.509 and
 OpenPGP private keys from the abstract type
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <8e7faab6a5727ca35887574daf1247a709ced168.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/gnutls_privkey.c           | 126 +++++++++++++++++++++++++++++++++++++++++
 lib/includes/gnutls/abstract.h |   5 ++
 lib/libgnutls.map              |   2 +
 3 files changed, 133 insertions(+)

diff --git a/lib/gnutls_privkey.c b/lib/gnutls_privkey.c
index 647777e..6249155 100644
--- a/lib/gnutls_privkey.c
+++ b/lib/gnutls_privkey.c
@@ -416,6 +416,51 @@ int gnutls_privkey_import_pkcs11_url(gnutls_privkey_t key, const char *url)
 	return ret;
 }
 
+/* This is currently disabled because there is no routine to copy a
+ * PKCS#11 private key. */
+#if 0
+/**
+ * gnutls_privkey_export_pkcs11:
+ * @pkey: The private key
+ * @key: Location for the key to be exported.
+ *
+ * Converts the given abstract private key to a #gnutls_pkcs11_privkey_t
+ * structure. The key must be of type %GNUTLS_PRIVKEY_PKCS11. The key
+ * returned in @key must be deinitialized with
+ * gnutls_pkcs11_privkey_deinit().
+ *
+ * Returns: On success, %GNUTLS_E_SUCCESS (0) is returned, otherwise a
+ *   negative error value.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_privkey_export_pkcs11(gnutls_privkey_t pkey,
+                             gnutls_pkcs11_privkey_t *key)
+{
+	int ret;
+
+	if (pkey->type != GNUTLS_PRIVKEY_PKCS11) {
+		gnutls_assert();
+		return GNUTLS_E_INVALID_REQUEST;
+	}
+
+	ret = gnutls_pkcs11_privkey_init(key);
+	if (ret < 0)
+		return gnutls_assert_val(ret);
+
+	ret = _gnutls_pkcs11_privkey_cpy(*key, pkey->key.openpgp); /* TODO */
+	if (ret < 0) {
+		gnutls_pkcs11_privkey_deinit(*key);
+		*key = NULL;
+
+		return gnutls_assert_val(ret);
+	}
+
+	return 0;
+}
+#endif
+
 #endif				/* ENABLE_PKCS11 */
 
 /**
@@ -563,6 +608,46 @@ gnutls_privkey_import_x509(gnutls_privkey_t pkey,
 }
 
 /**
+ * gnutls_privkey_export_x509:
+ * @pkey: The private key
+ * @key: Location for the key to be exported.
+ *
+ * Converts the given abstract private key to a #gnutls_x509_privkey_t
+ * structure. The key must be of type %GNUTLS_PRIVKEY_X509. The key returned
+ * in @key must be deinitialized with gnutls_x509_privkey_deinit().
+ *
+ * Returns: On success, %GNUTLS_E_SUCCESS (0) is returned, otherwise a
+ *   negative error value.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_privkey_export_x509(gnutls_privkey_t pkey,
+                           gnutls_x509_privkey_t *key)
+{
+	int ret;
+
+	if (pkey->type != GNUTLS_PRIVKEY_X509) {
+		gnutls_assert();
+		return GNUTLS_E_INVALID_REQUEST;
+	}
+
+	ret = gnutls_x509_privkey_init(key);
+	if (ret < 0)
+		return gnutls_assert_val(ret);
+
+	ret = gnutls_x509_privkey_cpy(*key, pkey->key.x509);
+	if (ret < 0) {
+		gnutls_x509_privkey_deinit(*key);
+		*key = NULL;
+
+		return gnutls_assert_val(ret);
+	}
+
+	return 0;
+}
+
+/**
  * gnutls_privkey_generate:
  * @pkey: The private key
  * @algo: is one of the algorithms in #gnutls_pk_algorithm_t.
@@ -736,6 +821,47 @@ int gnutls_privkey_import_openpgp_raw(gnutls_privkey_t pkey,
 
 	return ret;
 }
+
+/**
+ * gnutls_privkey_export_openpgp:
+ * @pkey: The private key
+ * @key: Location for the key to be exported.
+ *
+ * Converts the given abstract private key to a #gnutls_openpgp_privkey_t
+ * structure. The key must be of type %GNUTLS_PRIVKEY_OPENPGP. The key
+ * returned in @key must be deinitialized with
+ * gnutls_openpgp_privkey_deinit().
+ *
+ * Returns: On success, %GNUTLS_E_SUCCESS (0) is returned, otherwise a
+ *   negative error value.
+ *
+ * Since: 3.4.0
+ */
+int
+gnutls_privkey_export_openpgp(gnutls_privkey_t pkey,
+                              gnutls_openpgp_privkey_t *key)
+{
+	int ret;
+
+	if (pkey->type != GNUTLS_PRIVKEY_OPENPGP) {
+		gnutls_assert();
+		return GNUTLS_E_INVALID_REQUEST;
+	}
+
+	ret = gnutls_openpgp_privkey_init(key);
+	if (ret < 0)
+		return gnutls_assert_val(ret);
+
+	ret = _gnutls_openpgp_privkey_cpy(*key, pkey->key.openpgp);
+	if (ret < 0) {
+		gnutls_openpgp_privkey_deinit(*key);
+		*key = NULL;
+
+		return gnutls_assert_val(ret);
+	}
+
+	return 0;
+}
 #endif
 
 /**
diff --git a/lib/includes/gnutls/abstract.h b/lib/includes/gnutls/abstract.h
index 835fdb7..8ec1321 100644
--- a/lib/includes/gnutls/abstract.h
+++ b/lib/includes/gnutls/abstract.h
@@ -284,6 +284,11 @@ int gnutls_privkey_import_openpgp(gnutls_privkey_t pkey,
 				  gnutls_openpgp_privkey_t key,
 				  unsigned int flags);
 
+int gnutls_privkey_export_x509(gnutls_privkey_t pkey,
+                               gnutls_x509_privkey_t * key);
+int gnutls_privkey_export_openpgp(gnutls_privkey_t pkey,
+                                  gnutls_openpgp_privkey_t * key);
+
 int gnutls_privkey_import_openpgp_raw(gnutls_privkey_t pkey,
 				      const gnutls_datum_t * data,
 				      gnutls_openpgp_crt_fmt_t
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index b1121d0..27b12ba 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -925,6 +925,8 @@ GNUTLS_3_1_0 {
 	gnutls_x509_crl_get_raw_issuer_dn;
 	gnutls_certificate_get_crt_raw;
 	gnutls_privkey_generate;
+	gnutls_privkey_export_x509;
+	gnutls_privkey_export_openpgp;
 	gnutls_fips140_mode_enabled;
 	gnutls_record_check_corked;
 	gnutls_pkcs11_crt_is_known;
-- 
2.1.0



From armin at arbur.net  Thu Sep 18 19:12:08 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Thu, 18 Sep 2014 13:12:08 -0400
Subject: [gnutls-devel] [PATCH 0/9] Getter functions for gnutls_credentials_t
Message-ID: <cover.1411060328.git.armin@arbur.net>

Hi,

this is a series of patches that adds additional API to obtain
certificates, private keys and trusted CAs from a
gnutls_certificate_credentials_t structure. This could for example be
used to load the system CAs and then show those certificates in a user
interface. This was originally requested here:
https://savannah.gnu.org/support/index.php?108634

Most functions return copies of keys or certificates, so that the
implementation remains flexible with respect to future changes.

Cheers,
Armin

Armin Burgmeier (9):
  Fix a documentation typo
  Memory leak fix on certificate copy failure
  Add API to retrieve a X.509 or OpenPGP certificate from a
    gnutls_pcert_t
  Add a function to obtain the trust list of a
    gnutls_certificate_credentials_t
  Add functions to export X.509 and OpenPGP private keys from the
    abstract type
  Add gnutls_certificate_get_verify_flags
  Add functions to obtain X.509 keys and certificates from certificate
    credentials
  Add getter functions for openpgp keys and certificates
  Add an interface to iterate the trusted CA certificates in a trust
    list

 lib/gnutls_pcert.c              | 100 +++++++++++++++++++++++++++++++
 lib/gnutls_privkey.c            | 126 ++++++++++++++++++++++++++++++++++++++++
 lib/gnutls_ui.c                 |  19 +++++-
 lib/gnutls_x509.c               | 114 ++++++++++++++++++++++++++++++++++++
 lib/includes/gnutls/abstract.h  |  11 ++++
 lib/includes/gnutls/gnutls.h.in |  11 ++++
 lib/includes/gnutls/openpgp.h   |  10 ++++
 lib/includes/gnutls/x509.h      |  12 ++++
 lib/libgnutls.map               |  13 +++++
 lib/openpgp/gnutls_openpgp.c    |  98 +++++++++++++++++++++++++++++++
 lib/x509/verify-high.c          |  89 ++++++++++++++++++++++++++++
 11 files changed, 602 insertions(+), 1 deletion(-)

-- 
2.1.0



From armin at arbur.net  Thu Sep 18 00:54:09 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Wed, 17 Sep 2014 18:54:09 -0400
Subject: [gnutls-devel] [PATCH 6/9] Add gnutls_certificate_get_verify_flags
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <a28cb3459a8b4bb060e28fad05861b83ac3b8ed1.1411060328.git.armin@arbur.net>

Signed-off-by: Armin Burgmeier <armin at arbur.net>
---
 lib/gnutls_ui.c                 | 17 +++++++++++++++++
 lib/includes/gnutls/gnutls.h.in |  3 +++
 lib/libgnutls.map               |  1 +
 3 files changed, 21 insertions(+)

diff --git a/lib/gnutls_ui.c b/lib/gnutls_ui.c
index 03610e8..283fe97 100644
--- a/lib/gnutls_ui.c
+++ b/lib/gnutls_ui.c
@@ -631,6 +631,23 @@ gnutls_certificate_set_verify_flags(gnutls_certificate_credentials_t
 }
 
 /**
+ * gnutls_certificate_get_verify_flags:
+ * @res: is a gnutls_certificate_credentials_t structure
+ *
+ * Returns the verification flags set with
+ * gnutls_certificate_set_verify_flags().
+ *
+ * Returns: The certificate verification flags used by @res.
+ *
+ * Since: 3.4.0
+ */
+unsigned int
+gnutls_certificate_get_verify_flags(gnutls_certificate_credentials_t res)
+{
+	return res->verify_flags;
+}
+
+/**
  * gnutls_certificate_set_verify_limits:
  * @res: is a gnutls_certificate_credentials structure
  * @max_bits: is the number of bits of an acceptable certificate (default 8200)
diff --git a/lib/includes/gnutls/gnutls.h.in b/lib/includes/gnutls/gnutls.h.in
index bad59e0..17ff8fc 100644
--- a/lib/includes/gnutls/gnutls.h.in
+++ b/lib/includes/gnutls/gnutls.h.in
@@ -1349,6 +1349,9 @@ void gnutls_certificate_set_verify_limits(gnutls_certificate_credentials_t
 					  res, unsigned int max_bits,
 					  unsigned int max_depth);
 
+unsigned int
+gnutls_certificate_get_verify_flags(gnutls_certificate_credentials_t);
+
 int
 gnutls_certificate_set_x509_system_trust(gnutls_certificate_credentials_t
 					 cred);
diff --git a/lib/libgnutls.map b/lib/libgnutls.map
index 27b12ba..b64ff13 100644
--- a/lib/libgnutls.map
+++ b/lib/libgnutls.map
@@ -1012,6 +1012,7 @@ GNUTLS_3_1_0 {
 	gnutls_x509_crt_check_hostname2;
 	gnutls_openpgp_crt_check_hostname2;
 	gnutls_certificate_verify_peers;
+	gnutls_certificate_get_verify_flags;
 	gnutls_credentials_get;
 	gnutls_x509_crl_iter_crt_serial;
 	gnutls_x509_crl_iter_deinit;
-- 
2.1.0



From nmav at gnutls.org  Thu Sep 18 19:32:36 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 18 Sep 2014 19:32:36 +0200
Subject: [gnutls-devel] DANE: Verification failed for TLSA record with
	usage=0
In-Reply-To: <541A8E34.1090501@cesnet.cz>
References: <541A8E34.1090501@cesnet.cz>
Message-ID: <CAJU7zaKh+KRNMLJ6MkThc6sJT64um9ss5tPWy-Uoj4TxdzTTXg@mail.gmail.com>

On Thu, Sep 18, 2014 at 9:48 AM, Ond?ej Caletka
<Ondrej.Caletka at cesnet.cz> wrote:
> Hello,
> I've deployed TLSA records for www.cesnet.cz. I've used swede utility to
> constrain allowed CA and according to swede, the record is valid:
[...]
> However, ehen trying to validate this TLSA record using danetool from
> GnuTLS 3.3.7, I get validation error:
> $ danetool --check www.cesnet.cz --proto tcp --port 443
> Resolving 'www.cesnet.cz'...
> Obtaining certificate from '2001:718:1:101::4:443'...
> Querying DNS for www.cesnet.cz (tcp:443)...
> _443._tcp.www.cesnet.cz. IN TLSA ( 00 00 01
> 5c428b013b2e3f0d30abb5bebd92d066dc06dc223329eb0fc735609946cf8e1c )
> Certificate usage: CA (00)
> Certificate type:  X.509 (00)
> Contents:          SHA2-256 hash (01)
> Data:
> 5c428b013b2e3f0d30abb5bebd92d066dc06dc223329eb0fc735609946cf8e1c

Hi,
 Danetool requires that if the CA type is used, the CA should be the
immediate CA that signed your certificate. In your case the hash is
not of the CA that signed your certificate but from the CA that signed
your CA's certificate. That is an artificial danetool restriction that
would make sense to lift.

regards,
Nikos


From nmav at gnutls.org  Thu Sep 18 20:00:33 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 18 Sep 2014 20:00:33 +0200
Subject: [gnutls-devel] gnutls 3.3.8
In-Reply-To: <1411042209.22024.6.camel@nomad.lan>
References: <1411042209.22024.6.camel@nomad.lan>
Message-ID: <CAJU7zaLCay71zzJgPz2gbvsE-aDqS-2eHFe+u-F02s1bLOh4QQ@mail.gmail.com>

On Thu, Sep 18, 2014 at 2:10 PM, Nikos Mavrogiannopoulos
<nmav at gnutls.org> wrote:
> Hello,
>  I've just released gnutls 3.3.8. This is a bug-fix release on
> the next-stable branch. An important aspect of this release is that it
> completes support for the p11-kit trust module, allowing gnutls to
> utilize attached extensions in the system CA certificates (e.g. to
> restrict a CA certificate to certain domain names, or for a specific
> scope). I'll provide details on that feature on a follow up e-mail.

Ok. It seems I can now get back to that. P11-kit provides a trust
module that itself provides a PKCS #11 API to verify certificates
using some predefined trusted anchors. It provides a documented API
that gnutls uses (and as far as I understand an undocumented API that
NSS uses). In any case let's suppose you have p11-kit configured, say
with --with-trust-paths=/usr/share/pki/ca-trust-source and you have
the "normal" bundle of CAs there. Then a trust module needs to be
marked as such using a .conf file (e.g., in /etc/pkcs11/modules) with
the contents:
====
module: p11-kit-trust.so
priority: 1
trust-policy: yes
====

I'll give an example of the possibility of adding restrictions to the
a CAs in the bundle. E.g., let's change the scope of amazon.com's CA,
to prevent it from signing TLS certificates.
Before adding any restrictions let's ensure that connecting to our
target site works.
$ gnutls-cli --x509cafile "pkcs11:" www.amazon.com

if it doesn't there is something wrong with the p11-kit setup.

Then let's create /usr/share/pki/ca-trust-source/amazon.p11-kit
with the following contents:

[p11-kit-object-v1]
class: x-certificate-extension
label: "My label"
# The full URL encoded DER SubjectPublicKeyInfo SEQUENCE
public-key-info:
"%30%82%01%22%30%0d%06%09%2a%86%48%86%f7%0d%01%01%01%05%00%03%82%01%0f%00%30%82%01%0a%02%82%01%01%00%af%24%08%08%29%7a%35%9e%60%0c%aa%e7%4b%3b%4e%dc%7c%bc%3c%45%1c%bb%2b%e0%fe%29%02%f9%57%08%a3%64%85%15%27%f5%f1%ad%c8%31%89%5d%22%e8%2a%aa%a6%42%b3%8f%f8%b9%55%b7%b1%b7%4b%b3%fe%8f%7e%07%57%ec%ef%43%db%66%62%15%61%cf%60%0d%a4%d8%de%f8%e0%c3%62%08%3d%54%13%eb%49%ca%59%54%85%26%e5%2b%8f%1b%9f%eb%f5%a1%91%c2%33%49%d8%43%63%6a%52%4b%d2%8f%e8%70%51%4d%d1%89%69%7b%c7%70%f6%b3%dc%12%74%db%7b%5d%4b%56%d3%96%bf%15%77%a1%b0%f4%a2%25%f2%af%1c%92%67%18%e5%f4%06%04%ef%90%b9%e4%00%e4%dd%3a%b5%19%ff%02%ba%f4%3c%ee%e0%8b%eb%37%8b%ec%f4%d7%ac%f2%f6%f0%3d%af%dd%75%91%33%19%1d%1c%40%cb%74%24%19%21%93%d9%14%fe%ac%2a%52%c7%8f%d5%04%49%e4%8d%63%47%88%3c%69%83%cb%fe%47%bd%2b%7e%4f%c5%95%ae%0e%9d%d4%d1%43%c0%67%73%e3%14%08%7e%e5%3f%9f%73%b8%33%0a%cf%5d%3f%34%87%96%8a%ee%53%e8%25%15%02%03%01%00%01"
object-id: 2.5.29.37
value: "%30%16%06%03%55%1d%25%01%01%ff%04%0c%30%0a%06%08%2b%06%01%05%05%07%03%09"

That changes the purpose of the amazon CA to OCSP signing. The CA is
identified from the public-key-info (which is a HEX encoding of the
subjectPublicKeyInfo of that CA's key), and the object-id and value
are RFC5280 certificate extensions.

When that file is in place using the gnutls-cli command above should fail.


Other restrictions can be added, potentially all restrictions that can
be expressed with extensions used by gnutls, e.g., restricting the
host names a CA can sign for etc.
Unfortunately the tools required to make these p11-kit files pretty
much don't exist. You'll have to mess with some frob- examples or help
Stef Walter who maintains p11-kit to make them happen.

regards,
Nikos


From INVALID.NOREPLY at gnu.org  Fri Sep 19 03:17:55 2014
From: INVALID.NOREPLY at gnu.org (Ryan Schmidt)
Date: Fri, 19 Sep 2014 01:17:55 +0000
Subject: [gnutls-devel] [sr #108614] Reinstate support for C89
In-Reply-To: <20140727-123513.sv64386.57035@savannah.gnu.org>
References: <20140714-062418.sv64386.1640@savannah.gnu.org>
 <20140727-152108.sv707.62584@savannah.gnu.org>
 <20140727-123513.sv64386.57035@savannah.gnu.org>
Message-ID: <20140919-011754.sv64386.12112@savannah.gnu.org>

Follow-up Comment #3, sr #108614 (project gnutls):

It looks like you did apply this patch:

https://gitorious.org/gnutls/gnutls/commit/0345118a91ac776671cdf6c907164c75ac3559f1

Thanks! So this ticket can be closed.

While C99 may be 15 years old, C89 is the default mode of the gcc 4.2.1
compiler, which is the default compiler on OS X 10.6, which was released by
Apple just 5 years ago in August 2009 and was not replaced by OS X 10.7 until
July 2011. According to one company's survey just 6 weeks ago
(https://chitika.com/insights/2014/yosemite-preview-adoption) OS X 10.6
remains the second most popular version of OS X with 17.5% market share,
behind OS X 10.9 with 45.7%. So it is important to continue to support users
using OS X 10.6 and its older compiler.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108614>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From INVALID.NOREPLY at gnu.org  Fri Sep 19 03:24:55 2014
From: INVALID.NOREPLY at gnu.org (Ryan Schmidt)
Date: Fri, 19 Sep 2014 01:24:55 +0000
Subject: [gnutls-devel] [sr #108655] Issue tracker's git link outdated
Message-ID: <20140919-012454.sv64386.58648@savannah.gnu.org>

URL:
  <http://savannah.gnu.org/support/?108655>

                 Summary: Issue tracker's git link outdated
                 Project: GnuTLS
            Submitted by: ryandesign
            Submitted on: Fri 19 Sep 2014 01:24:54 AM GMT
                Category: None
                Priority: 5 - Normal
                Severity: 3 - Normal
                  Status: None
                 Privacy: Public
             Assigned to: None
        Originator Email: 
             Open/Closed: Open
         Discussion Lock: Any
        Operating System: None

    _______________________________________________________

Details:

In this issue tracker:

https://savannah.gnu.org/support/?func=additem&group=gnutls

Under the "Source Code" menu's "Browse Sources Repository" item, it takes you
to:

http://git.savannah.gnu.org/cgit/gnutls.git

This repository has not been updated in 2 years. It seems that development has
moved to this repository:

https://gitorious.org/gnutls/gnutls

The link in this issue tracker should be updated. The same update should be
made on this page:

https://savannah.gnu.org/git/?group=gnutls

Thanks.




    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108655>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From ametzler at bebt.de  Sat Sep 20 08:21:53 2014
From: ametzler at bebt.de (Andreas Metzler)
Date: Sat, 20 Sep 2014 08:21:53 +0200
Subject: [gnutls-devel] openconnect's vasprintf() license is (L)GPLv3
	incompatible
Message-ID: <20140920062152.GA14826@downhill.g.la>

Hello,

looking over the diff from 3.3.7 to 3.3.8 I have stumbled over the
newly added license header for lib/vasprintf.c (Thanks for it, BTW):

 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public License
 * version 2.1, as published by the Free Software Foundation.

This is a bit problematic, since LGPLv2.1 (without "or later")
is incompatible with (L)GPLv3+. Therefore (L)GPLv3+ software linked
against against GnuTLS could not be distributed. - It does not hit us
on Debian since we have vasprintf() in glibc but it is something of a
trap.

cu Andreas


-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'


From ametzler at bebt.de  Sat Sep 20 08:38:02 2014
From: ametzler at bebt.de (Andreas Metzler)
Date: Sat, 20 Sep 2014 08:38:02 +0200
Subject: [gnutls-devel] openconnect's vasprintf() license is (L)GPLv3
 incompatible
In-Reply-To: <20140920062152.GA14826@downhill.g.la>
References: <20140920062152.GA14826@downhill.g.la>
Message-ID: <20140920063802.GB14826@downhill.g.la>

On 2014-09-20 Andreas Metzler <ametzler at bebt.de> wrote:
> looking over the diff from 3.3.7 to 3.3.8 I have stumbled over the
> newly added license header for lib/vasprintf.c (Thanks for it, BTW):

>  * This program is free software; you can redistribute it and/or
>  * modify it under the terms of the GNU Lesser General Public License
>  * version 2.1, as published by the Free Software Foundation.

> This is a bit problematic, since LGPLv2.1 (without "or later")
> is incompatible with (L)GPLv3+. Therefore (L)GPLv3+ software linked
> against against GnuTLS could not be distributed. - It does not hit us
> on Debian since we have vasprintf() in glibc but it is something of a
> trap.

Please ignore. Looking at the compat matrix
http://www.gnu.org/licenses/gpl-faq.html shows that there is actually
no problem.

cu Andreas

-- 
`What a good friend you are to him, Dr. Maturin. His other friends are
so grateful to you.'
`I sew his ears on from time to time, sure'


From nmav at gnutls.org  Sun Sep 21 00:55:52 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Sun, 21 Sep 2014 00:55:52 +0200
Subject: [gnutls-devel] [PATCH 0/9] Getter functions for
 gnutls_credentials_t
In-Reply-To: <cover.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
Message-ID: <1411253752.2517.3.camel@nomad.lan>

On Thu, 2014-09-18 at 13:12 -0400, Armin Burgmeier wrote:
> Hi,
> 
> this is a series of patches that adds additional API to obtain
> certificates, private keys and trusted CAs from a
> gnutls_certificate_credentials_t structure. This could for example be
> used to load the system CAs and then show those certificates in a user
> interface. This was originally requested here:
> https://savannah.gnu.org/support/index.php?108634
> 
> Most functions return copies of keys or certificates, so that the
> implementation remains flexible with respect to future changes.

>   Fix a documentation typo
>   Memory leak fix on certificate copy failure
>   Add API to retrieve a X.509 or OpenPGP certificate from a
>     gnutls_pcert_t
>   Add a function to obtain the trust list of a
>     gnutls_certificate_credentials_t
>  Add gnutls_certificate_get_verify_flags

Thank you. I've checked and applied these so far. 

>   Add functions to export X.509 and OpenPGP private keys from the
>     abstract type
>   Add functions to obtain X.509 keys and certificates from certificate
>     credentials
>   Add getter functions for openpgp keys and certificates
>   Add an interface to iterate the trusted CA certificates in a trust
>     list

Would it be possible to modify some test programs, or a add new ones to
test these functions? That would ensure that the functionality offered
will not break on any internal change.

I'll try to add the missing part for the pkcs11 private key copy.

regards,
Nikos




From INVALID.NOREPLY at gnu.org  Sun Sep 21 01:21:27 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sat, 20 Sep 2014 23:21:27 +0000
Subject: [gnutls-devel] [sr #108655] Issue tracker's git link outdated
In-Reply-To: <20140919-012454.sv64386.58648@savannah.gnu.org>
References: <20140919-012454.sv64386.58648@savannah.gnu.org>
Message-ID: <20140921-022127.sv707.66691@savannah.gnu.org>

Follow-up Comment #1, sr #108655 (project gnutls):

Thanks but unfortunately these links are hard coded by savannah. As gitorious
doesn't offer a decent issue tracker we use that one here. I've modified the
project's description here to reflex that.

    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108655>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From INVALID.NOREPLY at gnu.org  Sun Sep 21 01:21:56 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sat, 20 Sep 2014 23:21:56 +0000
Subject: [gnutls-devel] [sr #108623] gnutls for Windows port.
In-Reply-To: <20140731-132008.sv707.28574@savannah.gnu.org>
References: <20140730-131401.sv79827.20795@savannah.gnu.org>
 <20140731-132008.sv707.28574@savannah.gnu.org>
Message-ID: <20140921-022156.sv707.65820@savannah.gnu.org>

Update of sr #108623 (project gnutls):

                  Status:                    None => Done                   
             Open/Closed:                    Open => Closed                 


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108623>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From INVALID.NOREPLY at gnu.org  Sun Sep 21 01:22:11 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sat, 20 Sep 2014 23:22:11 +0000
Subject: [gnutls-devel] [sr #108614] Reinstate support for C89
In-Reply-To: <20140919-011754.sv64386.12112@savannah.gnu.org>
References: <20140714-062418.sv64386.1640@savannah.gnu.org>
 <20140727-152108.sv707.62584@savannah.gnu.org>
 <20140727-123513.sv64386.57035@savannah.gnu.org>
 <20140919-011754.sv64386.12112@savannah.gnu.org>
Message-ID: <20140921-022211.sv707.33608@savannah.gnu.org>

Update of sr #108614 (project gnutls):

             Open/Closed:                    Open => Closed                 


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108614>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From INVALID.NOREPLY at gnu.org  Sun Sep 21 01:22:48 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sat, 20 Sep 2014 23:22:48 +0000
Subject: [gnutls-devel] [sr #108612] Both verify_ca() and verify_ee() abort
	DANE processing with DANE_E_UNKNOWN_DANE_DATA for unrecognised types
In-Reply-To: <20140707-001314.sv707.39125@savannah.gnu.org>
References: <20140706-193905.sv0.41002@savannah.gnu.org>
 <20140707-001314.sv707.39125@savannah.gnu.org>
Message-ID: <20140921-022248.sv707.30825@savannah.gnu.org>

Update of sr #108612 (project gnutls):

                  Status:          Ready For Test => Done                   
             Open/Closed:                    Open => Closed                 


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108612>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From INVALID.NOREPLY at gnu.org  Sun Sep 21 01:23:16 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sat, 20 Sep 2014 23:23:16 +0000
Subject: [gnutls-devel] [sr #108611] verify_ca() bypasses DANE checking if
	there are fewer than 2 certificates
In-Reply-To: <20140707-000504.sv707.63200@savannah.gnu.org>
References: <20140706-193407.sv0.89922@savannah.gnu.org>
 <20140707-000504.sv707.63200@savannah.gnu.org>
Message-ID: <20140921-022316.sv707.46729@savannah.gnu.org>

Update of sr #108611 (project gnutls):

                  Status:                    None => Done                   
             Open/Closed:                    Open => Closed                 

    _______________________________________________________

Follow-up Comment #2:

This was fixed in the repository.


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108611>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From nmav at gnutls.org  Sun Sep 21 08:52:28 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Sun, 21 Sep 2014 08:52:28 +0200
Subject: [gnutls-devel] certtool --crq-info should show signing algorithm
In-Reply-To: <5418BA4E.7090602@fifthhorseman.net>
References: <5418BA4E.7090602@fifthhorseman.net>
Message-ID: <1411282348.2576.1.camel@nomad.lan>

On Tue, 2014-09-16 at 18:31 -0400, Daniel Kahn Gillmor wrote:
> when looking at a human-readable representation of a certificate signing
> request (CSR), certtool --crq-info currently doesn't display the
> algorithm used for the self-signature of the CSR.  (i.e. it doesn't
> indicate if it was RSAwithSHA1 or RSAwithSHA256, etc)
> 
> It would occasionally be useful for people examining CSRs to see this
> information (i was just asked for it privately, and i think it would be
> useful too, e.g. when reviewing a CSR before submitting it to a CA).

This was added in master (i.e., what to become 3.4.0).

regards,
Nikos




From nmav at gnutls.org  Mon Sep 22 09:03:51 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 22 Sep 2014 09:03:51 +0200
Subject: [gnutls-devel] [PATCH 3/9] Add API to retrieve a X.509 or
 OpenPGP certificate from a gnutls_pcert_t
In-Reply-To: <05bb9ee6dbff87eecfe32565f451848fc2d5be38.1411060328.git.armin@arbur.net>
References: <cover.1411060328.git.armin@arbur.net>
 <05bb9ee6dbff87eecfe32565f451848fc2d5be38.1411060328.git.armin@arbur.net>
Message-ID: <CAJU7zaLnDA3tuwktGDwnGEXN9jQN_4z_AGxidjSUmsuPTg-tDw@mail.gmail.com>

On Wed, Sep 17, 2014 at 6:26 PM, Armin Burgmeier <armin at arbur.net> wrote:
>  /**
> + * gnutls_pcert_get_type:
> + * @pcert: The pcert structure.
> + *
> + * Returns the certificate type of @pcert, one of X.509 or OpenPGP.
> + *
> + * Returns: The certificate type.
> + *
> + * Since: 3.4.0
> + */
> +gnutls_certificate_type_t
> +gnutls_pcert_get_type(gnutls_pcert_st * pcert)

Is this function necessary? As the structure is exported I'm wondering
about the value of it...


From ludo at gnu.org  Mon Sep 22 15:51:16 2014
From: ludo at gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=)
Date: Mon, 22 Sep 2014 15:51:16 +0200
Subject: [gnutls-devel] configure should check for <idna.h>
Message-ID: <871tr3pxuz.fsf@gnu.org>

When GNU libidn isn?t installed, compilation of ?master? fails with:

--8<---------------cut here---------------start------------->8---
make[4]: Entering directory '/home/ludo/src/gnutls/+build/src/crywrap'
  CC       crywrap.o
../../../src/crywrap/crywrap.c:35:18: fatal error: idna.h: No such file or directory
 #include <idna.h>
                  ^
compilation terminated.
--8<---------------cut here---------------end--------------->8---

I think configure should check for this header, and possibly bail out if
this is a mandatory feature.  WDYT?

Ludo?.


From ludo at gnu.org  Mon Sep 22 16:25:14 2014
From: ludo at gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=)
Date: Mon, 22 Sep 2014 16:25:14 +0200
Subject: [gnutls-devel] Guile patches committed
Message-ID: <87wq8vohpx.fsf@gnu.org>

Hi,

Just to let you know I?ve committed two patches for the Guile bindings:
e9fc746 adds bindings for ?gnutls_server_name_set?, and abcb456 restores
a cross-reference in the manual.

Nikos: could you port them to the stable branch (I?d rather let you do
it than mess things up)?

Make sure to regenerate doc/core.c.texi and doc/gnutls-guile.info for
the next release.

Thanks,
Ludo?.


From nmav at gnutls.org  Mon Sep 22 18:57:11 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 22 Sep 2014 18:57:11 +0200
Subject: [gnutls-devel] configure should check for <idna.h>
In-Reply-To: <871tr3pxuz.fsf@gnu.org>
References: <871tr3pxuz.fsf@gnu.org>
Message-ID: <1411405031.3469.1.camel@nomad.lan>

On Mon, 2014-09-22 at 15:51 +0200, Ludovic Court?s wrote:
> When GNU libidn isn?t installed, compilation of ?master? fails with:
> 
> --8<---------------cut here---------------start------------->8---
> make[4]: Entering directory '/home/ludo/src/gnutls/+build/src/crywrap'
>   CC       crywrap.o
> ../../../src/crywrap/crywrap.c:35:18: fatal error: idna.h: No such file or directory
>  #include <idna.h>
>                   ^
> compilation terminated.
> I think configure should check for this header, and possibly bail out if
> this is a mandatory feature.  WDYT?

Thanks. It is an optional requirement. I've modified the configure
script to disable crywrap if libidn isn't there.

regards,
Nikos




From nmav at gnutls.org  Mon Sep 22 19:04:24 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 22 Sep 2014 19:04:24 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <87wq8vohpx.fsf@gnu.org>
References: <87wq8vohpx.fsf@gnu.org>
Message-ID: <1411405464.3469.6.camel@nomad.lan>

On Mon, 2014-09-22 at 16:25 +0200, Ludovic Court?s wrote:
> Hi,
> 
> Just to let you know I?ve committed two patches for the Guile bindings:
> e9fc746 adds bindings for ?gnutls_server_name_set?, and abcb456 restores
> a cross-reference in the manual.
> 
> Nikos: could you port them to the stable branch (I?d rather let you do
> it than mess things up)?

Hi,
 I've done the cherry pick. Please add or suggest some NEWS entry.

> Make sure to regenerate doc/core.c.texi and doc/gnutls-guile.info for
> the next release.

I realized that this manual isn't available through the document.wml.
I'll try to add it before next release.

regards,
Nikos




From ludo at gnu.org  Mon Sep 22 19:21:08 2014
From: ludo at gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=)
Date: Mon, 22 Sep 2014 19:21:08 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <1411405464.3469.6.camel@nomad.lan> (Nikos Mavrogiannopoulos's
 message of "Mon, 22 Sep 2014 19:04:24 +0200")
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
Message-ID: <87k34vmv0b.fsf@gnu.org>

Nikos Mavrogiannopoulos <nmav at gnutls.org> skribis:

> On Mon, 2014-09-22 at 16:25 +0200, Ludovic Court?s wrote:
>> Hi,
>> 
>> Just to let you know I?ve committed two patches for the Guile bindings:
>> e9fc746 adds bindings for ?gnutls_server_name_set?, and abcb456 restores
>> a cross-reference in the manual.
>> 
>> Nikos: could you port them to the stable branch (I?d rather let you do
>> it than mess things up)?
>
> Hi,
>  I've done the cherry pick. Please add or suggest some NEWS entry.

Cool.  What about this:

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/x-patch
Size: 308 bytes
Desc: not available
URL: </pipermail/attachments/20140922/69eed84b/attachment.bin>
-------------- next part --------------

>> Make sure to regenerate doc/core.c.texi and doc/gnutls-guile.info for
>> the next release.
>
> I realized that this manual isn't available through the document.wml.

What is document.wml?

> I'll try to add it before next release.

Great, thank you.

Ludo?.

From armin at arbur.net  Mon Sep 22 21:56:46 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Mon, 22 Sep 2014 15:56:46 -0400
Subject: [gnutls-devel] [PATCH 0/9] Getter functions for
 gnutls_credentials_t
In-Reply-To: <1411253752.2517.3.camel@nomad.lan>
References: <cover.1411060328.git.armin@arbur.net>
 <1411253752.2517.3.camel@nomad.lan>
Message-ID: <1411415806.1335.6.camel@waverley>

> >   Add functions to export X.509 and OpenPGP private keys from the
> >     abstract type
> >   Add functions to obtain X.509 keys and certificates from certificate
> >     credentials
> >   Add getter functions for openpgp keys and certificates
> >   Add an interface to iterate the trusted CA certificates in a trust
> >     list
> 
> Would it be possible to modify some test programs, or a add new ones to
> test these functions? That would ensure that the functionality offered
> will not break on any internal change.

Sounds reasonable, I'll take care of it.

Cheers,
Armin



From armin at arbur.net  Mon Sep 22 21:58:53 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Mon, 22 Sep 2014 15:58:53 -0400
Subject: [gnutls-devel] [PATCH 3/9] Add API to retrieve a X.509 or
 OpenPGP certificate from a gnutls_pcert_t
In-Reply-To: <CAJU7zaLnDA3tuwktGDwnGEXN9jQN_4z_AGxidjSUmsuPTg-tDw@mail.gmail.com>
References: <cover.1411060328.git.armin@arbur.net>
 <05bb9ee6dbff87eecfe32565f451848fc2d5be38.1411060328.git.armin@arbur.net>
 <CAJU7zaLnDA3tuwktGDwnGEXN9jQN_4z_AGxidjSUmsuPTg-tDw@mail.gmail.com>
Message-ID: <1411415933.1335.8.camel@waverley>

On Mon, 2014-09-22 at 09:03 +0200, Nikos Mavrogiannopoulos wrote:
> On Wed, Sep 17, 2014 at 6:26 PM, Armin Burgmeier <armin at arbur.net> wrote:
> >  /**
> > + * gnutls_pcert_get_type:
> > + * @pcert: The pcert structure.
> > + *
> > + * Returns the certificate type of @pcert, one of X.509 or OpenPGP.
> > + *
> > + * Returns: The certificate type.
> > + *
> > + * Since: 3.4.0
> > + */
> > +gnutls_certificate_type_t
> > +gnutls_pcert_get_type(gnutls_pcert_st * pcert)
> 
> Is this function necessary? As the structure is exported I'm wondering
> about the value of it...

Oh, I was under the impression the structure is private. If the fields
in the structure are part of the stable API then indeed, then indeed
there is no additional value. Sorry, I missed that.

Cheers,
Armin



From nmav at gnutls.org  Mon Sep 22 22:13:02 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 22 Sep 2014 22:13:02 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <87k34vmv0b.fsf@gnu.org>
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org>
Message-ID: <1411416782.5630.1.camel@nomad.lan>

On Mon, 2014-09-22 at 19:21 +0200, Ludovic Court?s wrote:
> Nikos Mavrogiannopoulos <nmav at gnutls.org> skribis:
> 
> > On Mon, 2014-09-22 at 16:25 +0200, Ludovic Court?s wrote:
> >> Hi,
> >> 
> >> Just to let you know I?ve committed two patches for the Guile bindings:
> >> e9fc746 adds bindings for ?gnutls_server_name_set?, and abcb456 restores
> >> a cross-reference in the manual.
> >> 
> >> Nikos: could you port them to the stable branch (I?d rather let you do
> >> it than mess things up)?
> >
> > Hi,
> >  I've done the cherry pick. Please add or suggest some NEWS entry.
> 
> Cool.  What about this:
> 
> >> Make sure to regenerate doc/core.c.texi and doc/gnutls-guile.info for
> >> the next release.
> >
> > I realized that this manual isn't available through the document.wml.
> 
> What is document.wml?

If you checkout the web-pages branch you'll see the source code of the
web pages.

regards,
Nikos




From nmav at gnutls.org  Mon Sep 22 22:39:45 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Mon, 22 Sep 2014 22:39:45 +0200
Subject: [gnutls-devel] [PATCH 3/9] Add API to retrieve a X.509 or
 OpenPGP certificate from a gnutls_pcert_t
In-Reply-To: <1411415933.1335.8.camel@waverley>
References: <cover.1411060328.git.armin@arbur.net>
 <05bb9ee6dbff87eecfe32565f451848fc2d5be38.1411060328.git.armin@arbur.net>
 <CAJU7zaLnDA3tuwktGDwnGEXN9jQN_4z_AGxidjSUmsuPTg-tDw@mail.gmail.com>
 <1411415933.1335.8.camel@waverley>
Message-ID: <1411418385.5630.3.camel@nomad.lan>

On Mon, 2014-09-22 at 15:58 -0400, Armin Burgmeier wrote:

> > >  /**
> > > + * gnutls_pcert_get_type:
> > > + * @pcert: The pcert structure.
> > > + *
> > > + * Returns the certificate type of @pcert, one of X.509 or OpenPGP.
> > > + *
> > > + * Returns: The certificate type.
> > > + *
> > > + * Since: 3.4.0
> > > + */
> > > +gnutls_certificate_type_t
> > > +gnutls_pcert_get_type(gnutls_pcert_st * pcert)
> > 
> > Is this function necessary? As the structure is exported I'm wondering
> > about the value of it...
> 
> Oh, I was under the impression the structure is private. If the fields
> in the structure are part of the stable API then indeed, then indeed
> there is no additional value. Sorry, I missed that.

No problem. I've removed this function.

regards,
Nikos




From nmav at gnutls.org  Tue Sep 23 11:00:38 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Tue, 23 Sep 2014 11:00:38 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <87k34vmv0b.fsf@gnu.org>
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org>
Message-ID: <CAJU7zaLzaHn=1qmtkUVHOr45XFB672zODwCpkx39AiOTgzhz2g@mail.gmail.com>

On Mon, Sep 22, 2014 at 7:21 PM, Ludovic Court?s <ludo at gnu.org> wrote:
>>> Just to let you know I've committed two patches for the Guile bindings:
>>> e9fc746 adds bindings for 'gnutls_server_name_set', and abcb456 restores
>>> a cross-reference in the manual.
>>> Nikos: could you port them to the stable branch (I'd rather let you do
>>> it than mess things up)?
>> Hi,
>>  I've done the cherry pick. Please add or suggest some NEWS entry.
> Cool.  What about this:

I think there is some text missing here...


From ludo at gnu.org  Tue Sep 23 11:46:21 2014
From: ludo at gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=)
Date: Tue, 23 Sep 2014 11:46:21 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <CAJU7zaLzaHn=1qmtkUVHOr45XFB672zODwCpkx39AiOTgzhz2g@mail.gmail.com>
 (Nikos Mavrogiannopoulos's message of "Tue, 23 Sep 2014 11:00:38
 +0200")
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org>
 <CAJU7zaLzaHn=1qmtkUVHOr45XFB672zODwCpkx39AiOTgzhz2g@mail.gmail.com>
Message-ID: <877g0uoej6.fsf@gnu.org>

Nikos Mavrogiannopoulos <nmav at gnutls.org> skribis:

> On Mon, Sep 22, 2014 at 7:21 PM, Ludovic Court?s <ludo at gnu.org> wrote:
>>>> Just to let you know I've committed two patches for the Guile bindings:
>>>> e9fc746 adds bindings for 'gnutls_server_name_set', and abcb456 restores
>>>> a cross-reference in the manual.
>>>> Nikos: could you port them to the stable branch (I'd rather let you do
>>>> it than mess things up)?
>>> Hi,
>>>  I've done the cherry pick. Please add or suggest some NEWS entry.
>> Cool.  What about this:
>
> I think there is some text missing here...

What do you have in mind?  The function is documented in the manual, so
I didn?t see the need to describe it any further.

Ludo?.


From armin at arbur.net  Tue Sep 23 22:12:38 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Tue, 23 Sep 2014 16:12:38 -0400
Subject: [gnutls-devel] [PATCH] Check the credentials getter functions as
 part of the unit tests
Message-ID: <ebe7aab56bdabc81e8d08a1624fb319846d452be.1411503220.git.armin@arbur.net>

---
 tests/openpgp-auth.c |  47 ++++++++++++++++++++++
 tests/x509cert.c     | 107 ++++++++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 149 insertions(+), 5 deletions(-)

diff --git a/tests/openpgp-auth.c b/tests/openpgp-auth.c
index 67faf73..82e3712 100644
--- a/tests/openpgp-auth.c
+++ b/tests/openpgp-auth.c
@@ -64,6 +64,49 @@ int key_recv_func(gnutls_session_t session, const unsigned char *keyfpr,
 	return 0;
 }
 
+void check_loaded_key(gnutls_certificate_credentials_t cred)
+{
+	int err;
+	gnutls_openpgp_privkey_t key;
+	gnutls_openpgp_crt_t *crts;
+	int n_crts;
+	gnutls_datum_t datum;
+	gnutls_openpgp_keyid_t keyid;
+	int i;
+
+	/* check that the getter functions for openpgp keys of
+	 * gnutls_certificate_credentials_t work and deliver the
+	 * expected key ID. */
+
+	err = gnutls_certificate_get_openpgp_key(cred, 0, &key);
+	if (err != 0)
+		fail("get openpgp key %s\n",
+		     gnutls_strerror(err));
+
+	gnutls_openpgp_privkey_get_subkey_id(key, 0, keyid);
+	if (keyid[0] != 0xf3 || keyid[1] != 0x0f || keyid[2] != 0xd4 || keyid[3] != 0x23 ||
+	    keyid[4] != 0xc1 || keyid[5] != 0x43 || keyid[6] != 0xe7 || keyid[7] != 0xba)
+		fail("incorrect key id (privkey)\n");
+
+	err = gnutls_certificate_get_openpgp_crt(cred, 0, &crts, &n_crts);
+	if (err != 0)
+		fail("get openpgp crts %s\n",
+		     gnutls_strerror(err));
+
+	if (n_crts != 1)
+		fail("openpgp n_crts != 1\n");
+
+	gnutls_openpgp_crt_get_subkey_id(crts[0], 0, keyid);
+	if (keyid[0] != 0xf3 || keyid[1] != 0x0f || keyid[2] != 0xd4 || keyid[3] != 0x23 ||
+	    keyid[4] != 0xc1 || keyid[5] != 0x43 || keyid[6] != 0xe7 || keyid[7] != 0xba)
+		fail("incorrect key id (pubkey)\n");
+
+	for (i = 0; i < n_crts; ++i)
+		gnutls_openpgp_crt_deinit(crts[i]);
+	gnutls_free(crts);
+	gnutls_openpgp_privkey_deinit(key);
+}
+
 void doit()
 {
 	int err, i;
@@ -151,6 +194,8 @@ void doit()
 				fail("client openpgp keys %s\n",
 				     gnutls_strerror(err));
 
+			check_loaded_key(cred);
+
 			err =
 			    gnutls_credentials_set(session,
 						   GNUTLS_CRD_CERTIFICATE,
@@ -230,6 +275,8 @@ void doit()
 				fail("server openpgp keys %s\n",
 				     gnutls_strerror(err));
 
+			check_loaded_key(cred);
+
 			err = gnutls_dh_params_init(&dh_params);
 			if (err)
 				fail("server DH params init %d\n", err);
diff --git a/tests/x509cert.c b/tests/x509cert.c
index 853e7e7..4e07ae5 100644
--- a/tests/x509cert.c
+++ b/tests/x509cert.c
@@ -64,7 +64,7 @@ static unsigned char ca_pem[] =
     "njuu7kHq5peUgYn8Jd9zNzExBOEp1VOipGsf6G66oQAhDFp2o8zkz7ZH71zR4HEW\n"
     "KoX6n5Emn6DvcEH/9pAhnGxNHJAoS7czTKv/JDZJhkqHxyrE1fuLsg5Qv25DTw7+\n"
     "PfqUpIhz5Bbm7J4=\n" "-----END CERTIFICATE-----\n";
-const gnutls_datum_t ca = { ca_pem, sizeof(ca_pem) };
+const gnutls_datum_t ca = { ca_pem, sizeof(ca_pem) - 1};
 
 static unsigned char cert_pem[] =
     "-----BEGIN CERTIFICATE-----\n"
@@ -92,7 +92,7 @@ static unsigned char cert_pem[] =
     "njuu7kHq5peUgYn8Jd9zNzExBOEp1VOipGsf6G66oQAhDFp2o8zkz7ZH71zR4HEW\n"
     "KoX6n5Emn6DvcEH/9pAhnGxNHJAoS7czTKv/JDZJhkqHxyrE1fuLsg5Qv25DTw7+\n"
     "PfqUpIhz5Bbm7J4=\n" "-----END CERTIFICATE-----\n";
-const gnutls_datum_t cert = { cert_pem, sizeof(cert_pem) };
+const gnutls_datum_t cert = { cert_pem, sizeof(cert_pem) - 1};
 
 static unsigned char key_pem[] =
     "-----BEGIN RSA PRIVATE KEY-----\n"
@@ -110,7 +110,7 @@ static unsigned char key_pem[] =
     "/iVX2cmMTSh3w3z8MaECQEp0XJWDVKOwcTW6Ajp9SowtmiZ3YDYo1LF9igb4iaLv\n"
     "sWZGfbnU3ryjvkb6YuFjgtzbZDZHWQCo8/cOtOBmPdk=\n"
     "-----END RSA PRIVATE KEY-----\n";
-const gnutls_datum_t key = { key_pem, sizeof(key_pem) };
+const gnutls_datum_t key = { key_pem, sizeof(key_pem) - 1};
 
 static unsigned char server_cert_pem[] =
     "-----BEGIN CERTIFICATE-----\n"
@@ -129,7 +129,7 @@ static unsigned char server_cert_pem[] =
     "rOKLUQRWJ0K3HyXRMhbqjdLIaQiCvQLuizo=\n" "-----END CERTIFICATE-----\n";
 
 const gnutls_datum_t server_cert = { server_cert_pem,
-	sizeof(server_cert_pem)
+	sizeof(server_cert_pem) - 1
 };
 
 static unsigned char server_key_pem[] =
@@ -150,7 +150,7 @@ static unsigned char server_key_pem[] =
     "-----END RSA PRIVATE KEY-----\n";
 
 const gnutls_datum_t server_key = { server_key_pem,
-	sizeof(server_key_pem)
+	sizeof(server_key_pem) - 1
 };
 
 #define LIST_SIZE 3
@@ -165,6 +165,15 @@ void doit(void)
 	size_t dn_size;
 	unsigned int list_size;
 
+	gnutls_x509_privkey_t get_key;
+	gnutls_x509_crt_t *get_crts;
+	int n_get_crts;
+	gnutls_datum_t get_datum;
+	gnutls_x509_trust_list_t trust_list;
+	gnutls_x509_trust_list_iter_t trust_iter;
+	gnutls_x509_crt_t get_ca_crt;
+	int n_get_ca_crts;
+
 	/* this must be called once in the program
 	 */
 	global_init();
@@ -203,6 +212,94 @@ void doit(void)
 
 	if (debug)
 		fprintf(stderr, "Issuer's DN: %s\n", dn);
+
+	/* test the getter functions of gnutls_certificate_credentials_t */
+
+	ret =
+	    gnutls_certificate_get_x509_key(x509_cred, 0, &get_key);
+	if (ret < 0)
+		fail("gnutls_certificate_get_x509_key");
+
+	ret =
+	    gnutls_x509_privkey_export2(get_key,
+	                                GNUTLS_X509_FMT_PEM,
+	                                &get_datum);
+	if (ret < 0)
+		fail("gnutls_x509_privkey_export2");
+
+	if (get_datum.size != server_key.size ||
+	    memcmp(get_datum.data, server_key.data, get_datum.size) != 0) {
+		fail(
+		    "exported key %u vs. %u\n\n%s\n\nvs.\n\n%s",
+		    get_datum.size, server_key.size,
+		    get_datum.data, server_key.data);
+	}
+
+	gnutls_free(get_datum.data);
+
+	ret =
+	    gnutls_certificate_get_x509_crt(x509_cred, 0, &get_crts, &n_get_crts);
+	if (ret < 0)
+		fail("gnutls_certificate_get_x509_crt");
+	if (n_get_crts != 1)
+		fail("gnutls_certificate_get_x509_crt: n_crts != 1");
+
+	ret =
+	    gnutls_x509_crt_export2(get_crts[0],
+	                            GNUTLS_X509_FMT_PEM,
+	                            &get_datum);
+	if (ret < 0)
+		fail("gnutls_x509_crt_export2");
+
+	if (get_datum.size != server_cert.size ||
+	    memcmp(get_datum.data, server_cert.data, get_datum.size) != 0) {
+		fail(
+		    "exported certificate %u vs. %u\n\n%s\n\nvs.\n\n%s",
+		    get_datum.size, server_cert.size,
+		    get_datum.data, server_cert.data);
+	}
+
+	gnutls_free(get_datum.data);
+
+	gnutls_certificate_get_trust_list(x509_cred, &trust_list);
+
+	n_get_ca_crts = 0;
+	trust_iter = NULL;
+	while (gnutls_x509_trust_list_iter_get_ca(trust_list,
+	                                          &trust_iter,
+	                                          &get_ca_crt) !=
+	       GNUTLS_E_REQUESTED_DATA_NOT_AVAILABLE) {
+		ret =
+		    gnutls_x509_crt_export2(get_ca_crt,
+		                            GNUTLS_X509_FMT_PEM,
+		                            &get_datum);
+		if (ret < 0)
+			fail("gnutls_x509_crt_export2");
+
+		if (get_datum.size != ca.size ||
+		    memcmp(get_datum.data, ca.data, get_datum.size) != 0) {
+			fail(
+			    "exported CA certificate %u vs. %u\n\n%s\n\nvs.\n\n%s",
+			    get_datum.size, ca.size,
+			    get_datum.data, ca.data);
+		}
+
+		gnutls_x509_crt_deinit(get_ca_crt);
+		gnutls_free(get_datum.data);
+
+		++n_get_ca_crts;
+	}
+
+	if (n_get_ca_crts != 1)
+		fail("gnutls_x509_trust_list_iter_get_ca: n_cas != 1");
+	if (trust_iter != NULL)
+		fail("gnutls_x509_trust_list_iter_get_ca: iterator not NULL after iteration");
+
+	gnutls_x509_privkey_deinit(get_key);
+	for (i = 0; i < n_get_crts; i++)
+		gnutls_x509_crt_deinit(get_crts[i]);
+	gnutls_free(get_crts);
+
 	for (i = 0; i < list_size; i++)
 		gnutls_x509_crt_deinit(list[i]);
 	gnutls_certificate_free_credentials(x509_cred);
-- 
2.1.0



From nmav at gnutls.org  Wed Sep 24 10:37:55 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Wed, 24 Sep 2014 10:37:55 +0200
Subject: [gnutls-devel] [PATCH] Check the credentials getter functions
 as part of the unit tests
In-Reply-To: <ebe7aab56bdabc81e8d08a1624fb319846d452be.1411503220.git.armin@arbur.net>
References: <ebe7aab56bdabc81e8d08a1624fb319846d452be.1411503220.git.armin@arbur.net>
Message-ID: <1411547875.2495.3.camel@nomad.lan>

On Tue, 2014-09-23 at 16:12 -0400, Armin Burgmeier wrote:
> ---
>  tests/openpgp-auth.c |  47 ++++++++++++++++++++++
>  tests/x509cert.c     | 107 ++++++++++++++++++++++++++++++++++++++++++++++++---
>  2 files changed, 149 insertions(+), 5 deletions(-)

Thank you. I've applied the complete patch set.

regards,
Nikos




From citypw at gmail.com  Wed Sep 24 16:30:34 2014
From: citypw at gmail.com (Shawn)
Date: Wed, 24 Sep 2014 22:30:34 +0800
Subject: [gnutls-devel] [gnutls-help] gnutls 3.2.18
In-Reply-To: <1411041962.22024.2.camel@nomad.lan>
References: <1411041962.22024.2.camel@nomad.lan>
Message-ID: <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>

Can't download it from:
ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz

Plz check it out...Thanks!


On Thu, Sep 18, 2014 at 8:06 PM, Nikos Mavrogiannopoulos
<nmav at gnutls.org> wrote:
> Hello,
>  I've just released gnutls 3.2.18. This is a bugfix release on the
> current stable branch.
>
> * Version 3.2.18 (released 2014-09-18)
>
> ** libgnutls: Fixes in gnutls_x509_crt_set_dn() and friends to properly handle
> strings with embedded spaces and escaped commas.
>
> ** libgnutls: Corrected gnutls_x509_crl_verify() which would always report
> a CRL signature as invalid. Reported by Armin Burgmeier.
>
> ** libgnutls: Fixed issue with certificates being sanitized by gnutls prior
> to signature verification. That resulted to certain non-DER compliant modifications
> of valid certificates, being corrected by libtasn1's parser and restructured as
> the original. Issue found and reported by Antti Karjalainen and Matti Kamunen from
> Codenomicon.
>
> ** API and ABI modifications:
> No changes since last version.
>
>
> Getting the Software
> ====================
>
> GnuTLS may be downloaded directly from
> <ftp://ftp.gnutls.org/gcrypt/gnutls/>.  A list of GnuTLS mirrors can be
> found at <http://www.gnutls.org/download.html>.
>
> Here are the XZ and LZIP compressed sources:
>
>   ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
>   ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.lz
>
> Here are OpenPGP detached signatures signed using key 0x96865171:
>
>   ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz.sig
>   ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.lz.sig
>
> Note that it has been signed with my openpgp key:
> pub   3104R/96865171 2008-05-04 [expires: 2028-04-29]
> uid                  Nikos Mavrogiannopoulos <nmav <at> gnutls.org>
> uid                  Nikos Mavrogiannopoulos <n.mavrogiannopoulos <at>
> gmail.com>
> sub   2048R/9013B842 2008-05-04 [expires: 2018-05-02]
> sub   2048R/1404A91D 2008-05-04 [expires: 2018-05-02]
>
> regards,
> Nikos
>
>
>
>
> _______________________________________________
> Gnutls-help mailing list
> Gnutls-help at lists.gnutls.org
> http://lists.gnupg.org/mailman/listinfo/gnutls-help



-- 
GNU powered it...
GPL protect it...
God blessing it...

regards
Shawn


From citypw at gmail.com  Wed Sep 24 19:14:11 2014
From: citypw at gmail.com (Shawn)
Date: Thu, 25 Sep 2014 01:14:11 +0800
Subject: [gnutls-devel] [gnutls-help] gnutls 3.2.18
In-Reply-To: <CAPZ8mV63gBUz2iG2mW6nSveNLZh33M-iQAWkDF4BVcZaML_9rg@mail.gmail.com>
References: <1411041962.22024.2.camel@nomad.lan>
 <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>
 <CAPZ8mV63gBUz2iG2mW6nSveNLZh33M-iQAWkDF4BVcZaML_9rg@mail.gmail.com>
Message-ID: <CABniQZNug5MEWmWuO99VYtcjvFfTZDPjt6Ge-kP3oAY4tEJUSQ@mail.gmail.com>

hey Mark,

On Thu, Sep 25, 2014 at 1:03 AM, Mark Doliner <mark at kingant.net> wrote:
> On Wed, Sep 24, 2014 at 7:30 AM, Shawn <citypw at gmail.com> wrote:
>> Can't download it from:
>> ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
>>
>> Plz check it out...Thanks!
>
> Hi Shawn. I was able to download from that URL just now. I suppose
> it's possible someone fixed the problem since you sent your email. Are
> you still unable to download? If so, can you describe the problem
> you're having?
>
I'm still not able to download from above URL for now. wget would have
some speed then totally down to the zero:
------------------------------------------------------------------------------------------------------------------------
--2014-09-25 01:12:10--
ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
           => ?gnutls-3.2.18.tar.xz.1?
Resolving ftp.gnutls.org (ftp.gnutls.org)... 217.69.76.55
Connecting to ftp.gnutls.org (ftp.gnutls.org)|217.69.76.55|:21... connected.
Logging in as anonymous ... Logged in!
==> SYST ... done.    ==> PWD ... done.
==> TYPE I ... done.  ==> CWD (1) /gcrypt/gnutls/v3.2 ... done.
==> SIZE gnutls-3.2.18.tar.xz ... 5154476
==> PASV ... done.    ==> RETR gnutls-3.2.18.tar.xz ... done.
Length: 5154476 (4.9M) (unauthoritative)

 1% [>
              ] 71,000      --.-K/s  eta 22m 6s
------------------------------------------------------------------------------------------------------------------------

I tried two different IPs but still got the same issue....Some mirrors
are working fine, like this one:

http://artfiles.org/gnupg.org/gnutls/v3.2/gnutls-3.2.18.tar.xz



-- 
GNU powered it...
GPL protect it...
God blessing it...

regards
Shawn


From mark at kingant.net  Wed Sep 24 19:03:23 2014
From: mark at kingant.net (Mark Doliner)
Date: Wed, 24 Sep 2014 10:03:23 -0700
Subject: [gnutls-devel] [gnutls-help] gnutls 3.2.18
In-Reply-To: <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>
References: <1411041962.22024.2.camel@nomad.lan>
 <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>
Message-ID: <CAPZ8mV63gBUz2iG2mW6nSveNLZh33M-iQAWkDF4BVcZaML_9rg@mail.gmail.com>

On Wed, Sep 24, 2014 at 7:30 AM, Shawn <citypw at gmail.com> wrote:
> Can't download it from:
> ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
>
> Plz check it out...Thanks!

Hi Shawn. I was able to download from that URL just now. I suppose
it's possible someone fixed the problem since you sent your email. Are
you still unable to download? If so, can you describe the problem
you're having?


From nmav at gnutls.org  Thu Sep 25 11:31:56 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 25 Sep 2014 11:31:56 +0200
Subject: [gnutls-devel] [gnutls-help] gnutls 3.2.18
In-Reply-To: <CABniQZNug5MEWmWuO99VYtcjvFfTZDPjt6Ge-kP3oAY4tEJUSQ@mail.gmail.com>
References: <1411041962.22024.2.camel@nomad.lan>
 <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>
 <CAPZ8mV63gBUz2iG2mW6nSveNLZh33M-iQAWkDF4BVcZaML_9rg@mail.gmail.com>
 <CABniQZNug5MEWmWuO99VYtcjvFfTZDPjt6Ge-kP3oAY4tEJUSQ@mail.gmail.com>
Message-ID: <1411637516.2494.11.camel@nomad.lan>

On Thu, 2014-09-25 at 01:14 +0800, Shawn wrote:
> hey Mark,
> 
> On Thu, Sep 25, 2014 at 1:03 AM, Mark Doliner <mark at kingant.net> wrote:
> > On Wed, Sep 24, 2014 at 7:30 AM, Shawn <citypw at gmail.com> wrote:
> >> Can't download it from:
> >> ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
> >>
> >> Plz check it out...Thanks!
> >
> > Hi Shawn. I was able to download from that URL just now. I suppose
> > it's possible someone fixed the problem since you sent your email. Are
> > you still unable to download? If so, can you describe the problem
> > you're having?
> >
> I'm still not able to download from above URL for now. wget would have
> some speed then totally down to the zero:

That could be some firewall or routing issue between you and the gnupg
server. You can use the available mirror list for alternative servers.

regards,
Nikos




From nmav at gnutls.org  Thu Sep 25 14:14:13 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Thu, 25 Sep 2014 14:14:13 +0200
Subject: [gnutls-devel] Symbol versioning in gnutls broken -> crashes
In-Reply-To: <CAJU7zaLvs1S3YyxDVKMut5BpdKdaE+CYQDnxWLihfxkF8pxajg@mail.gmail.com>
References: <20140524065807.GA2250@downhill.g.la>
 <1410468974.3121.3.camel@nomad.lan> <20140912162227.GA5709@downhill.g.la>
 <1410546977.11023.8.camel@nomad.lan> <20140913060441.GA2429@downhill.g.la>
 <CAJU7zaKYOuRAy8V2ns=eoj0KJ+YfMuUBQ33GR1vfqqH0ZOdiNw@mail.gmail.com>
 <CAJU7zaLvs1S3YyxDVKMut5BpdKdaE+CYQDnxWLihfxkF8pxajg@mail.gmail.com>
Message-ID: <1411647253.24226.3.camel@nomad.lan>

On Mon, 2014-09-15 at 15:49 +0200, Nikos Mavrogiannopoulos wrote:
> On Mon, Sep 15, 2014 at 9:46 AM, Nikos Mavrogiannopoulos
> <nmav at gnutls.org> wrote:
> > I'm wondering whether there is some way to transition to a new ABI
> > without breaking the old one. I'll try to see whether the ld script
> > allows such move.
> Ok, it seems that adding a new VERSION section with all the symbols in
> the beginning of libgnutls.map, makes the default symbols to be the
> new, but also exposes the old ones. So we can have a slow migration to
> a new ABI, in the 3.4.0 release. However, That is not really
> documented in [0], and I wonder whether we can rely on that.

Unfortunately from what I could get from a brief thread in binutils, the
only way to set a default symbol is through an explicit tag of the
function. Thus my understanding is that we can't rely on that.

Nevertheless, that would allow a program compiled in debian to work on
other systems too. That doesn't sound ideal, but I think it is better
than having completely incompatible versions around. What do you think?

regards,
Nikos




From dev at cor0.com  Thu Sep 25 14:51:46 2014
From: dev at cor0.com (dev)
Date: Thu, 25 Sep 2014 08:51:46 -0400 (EDT)
Subject: [gnutls-devel] [gnutls-help] gnutls 3.2.18
In-Reply-To: <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>
References: <1411041962.22024.2.camel@nomad.lan>
 <CABniQZM2DDuz4s9aBt46aAYcZiPEw3H4yHREo+Kojt9F+eNLmA@mail.gmail.com>
Message-ID: <1505977686.21326.1411649506698.JavaMail.vpopmail@webmail2.networksolutionsemail.com>



On September 24, 2014 at 10:30 AM Shawn <citypw at gmail.com> wrote:
> Can't download it from:
> ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
>
> Plz check it out...Thanks!


wfm :

node000 $ wget
ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
--2014-09-25 12:50:08--
 ftp://ftp.gnutls.org/gcrypt/gnutls/v3.2/gnutls-3.2.18.tar.xz
           => 'gnutls-3.2.18.tar.xz'
Resolving ftp.gnutls.org... 217.69.76.55
Connecting to ftp.gnutls.org|217.69.76.55|:21... connected.
Logging in as anonymous ... Logged in!
==> SYST ... done.    ==> PWD ... done.
==> TYPE I ... done.  ==> CWD (1) /gcrypt/gnutls/v3.2 ... done.
==> SIZE gnutls-3.2.18.tar.xz ... 5154476
==> PASV ... done.    ==> RETR gnutls-3.2.18.tar.xz ... done.
Length: 5154476 (4.9M) (unauthoritative)

     0K ....                                                  100%
 349K=14s

2014-09-25 12:50:25 (349 KB/s) - 'gnutls-3.2.18.tar.xz' saved [5154476]


node000 $ /usr/local/ssl/bin/openssl dgst -sha256 gnutls-3.2.18.tar.xz
SHA256(gnutls-3.2.18.tar.xz)=
4762afab5e1b9e829c5f53d2b00cd5e41d43fa6d035efcf239e3fe0459134d45


From armin at arbur.net  Thu Sep 25 21:53:02 2014
From: armin at arbur.net (Armin Burgmeier)
Date: Thu, 25 Sep 2014 15:53:02 -0400
Subject: [gnutls-devel] [PATCH] Check the credentials getter functions
 as part of the unit tests
In-Reply-To: <1411547875.2495.3.camel@nomad.lan>
References: <ebe7aab56bdabc81e8d08a1624fb319846d452be.1411503220.git.armin@arbur.net>
 <1411547875.2495.3.camel@nomad.lan>
Message-ID: <1411674782.1339.13.camel@waverley>

On Wed, 2014-09-24 at 10:37 +0200, Nikos Mavrogiannopoulos wrote:
> On Tue, 2014-09-23 at 16:12 -0400, Armin Burgmeier wrote:
> > ---
> >  tests/openpgp-auth.c |  47 ++++++++++++++++++++++
> >  tests/x509cert.c     | 107 ++++++++++++++++++++++++++++++++++++++++++++++++---
> >  2 files changed, 149 insertions(+), 5 deletions(-)
> 
> Thank you. I've applied the complete patch set.

Thanks! I suppose the ticket on savannah could then be closed as well.

Armin



From INVALID.NOREPLY at gnu.org  Sat Sep 27 09:06:50 2014
From: INVALID.NOREPLY at gnu.org (Nikos Mavrogiannopoulos)
Date: Sat, 27 Sep 2014 07:06:50 +0000
Subject: [gnutls-devel] [sr #108634] Getter functions for
	gnutls_certificate_credentials_t
In-Reply-To: <20140907-200409.sv96363.42296@savannah.gnu.org>
References: <20140818-164743.sv96363.52612@savannah.gnu.org>
 <20140825-203439.sv707.32318@savannah.gnu.org>
 <20140907-132239.sv707.3610@savannah.gnu.org>
 <20140907-200409.sv96363.42296@savannah.gnu.org>
Message-ID: <20140927-100650.sv707.22938@savannah.gnu.org>

Update of sr #108634 (project gnutls):

                  Status:                    None => Done                   
             Open/Closed:                    Open => Closed                 


    _______________________________________________________

Reply to this item at:

  <http://savannah.gnu.org/support/?108634>

_______________________________________________
  Message sent via/by Savannah
  http://savannah.gnu.org/



From ludo at gnu.org  Tue Sep 30 13:24:13 2014
From: ludo at gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=)
Date: Tue, 30 Sep 2014 13:24:13 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <1411416782.5630.1.camel@nomad.lan> (Nikos Mavrogiannopoulos's
 message of "Mon, 22 Sep 2014 22:13:02 +0200")
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org> <1411416782.5630.1.camel@nomad.lan>
Message-ID: <87lhp1bbc2.fsf@gnu.org>

Nikos Mavrogiannopoulos <nmav at gnutls.org> skribis:

> If you checkout the web-pages branch you'll see the source code of the
> web pages.

I?ve now added the GnuTLS-Guile manual.  Could you upload it?

Thank you!

Ludo?.


From ludo at gnu.org  Tue Sep 30 13:25:06 2014
From: ludo at gnu.org (Ludovic =?utf-8?Q?Court=C3=A8s?=)
Date: Tue, 30 Sep 2014 13:25:06 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <CAJU7zaLzaHn=1qmtkUVHOr45XFB672zODwCpkx39AiOTgzhz2g@mail.gmail.com>
 (Nikos Mavrogiannopoulos's message of "Tue, 23 Sep 2014 11:00:38
 +0200")
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org>
 <CAJU7zaLzaHn=1qmtkUVHOr45XFB672zODwCpkx39AiOTgzhz2g@mail.gmail.com>
Message-ID: <87h9zpbbal.fsf@gnu.org>

Nikos Mavrogiannopoulos <nmav at gnutls.org> skribis:

> On Mon, Sep 22, 2014 at 7:21 PM, Ludovic Court?s <ludo at gnu.org> wrote:
>>>> Just to let you know I've committed two patches for the Guile bindings:
>>>> e9fc746 adds bindings for 'gnutls_server_name_set', and abcb456 restores
>>>> a cross-reference in the manual.
>>>> Nikos: could you port them to the stable branch (I'd rather let you do
>>>> it than mess things up)?
>>> Hi,
>>>  I've done the cherry pick. Please add or suggest some NEWS entry.
>> Cool.  What about this:
>
> I think there is some text missing here...

Commit e9941db adds a NEWS entry.  Could you cherry-pick it to the
relevant branch(es)?

Thanks,
Ludo?.


From nmav at gnutls.org  Tue Sep 30 20:54:37 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Tue, 30 Sep 2014 20:54:37 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <87lhp1bbc2.fsf@gnu.org>
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org> <1411416782.5630.1.camel@nomad.lan>
 <87lhp1bbc2.fsf@gnu.org>
Message-ID: <1412103277.2908.1.camel@nomad.lan>

On Tue, 2014-09-30 at 13:24 +0200, Ludovic Court?s wrote:
> Nikos Mavrogiannopoulos <nmav at gnutls.org> skribis:
> 
> > If you checkout the web-pages branch you'll see the source code of the
> > web pages.
> 
> I?ve now added the GnuTLS-Guile manual.  Could you upload it?

Hi,
 I've regenerated the html and uploaded (it's automatically done every
few hours) it but there is something wrong with the CSS (you can see the
style difference between the two manuals). I think using
--css-include=texinfo.css (the one in doc) would suffice.

regards,
Nikos




From nmav at gnutls.org  Tue Sep 30 20:57:21 2014
From: nmav at gnutls.org (Nikos Mavrogiannopoulos)
Date: Tue, 30 Sep 2014 20:57:21 +0200
Subject: [gnutls-devel] Guile patches committed
In-Reply-To: <87h9zpbbal.fsf@gnu.org>
References: <87wq8vohpx.fsf@gnu.org> <1411405464.3469.6.camel@nomad.lan>
 <87k34vmv0b.fsf@gnu.org>
 <CAJU7zaLzaHn=1qmtkUVHOr45XFB672zODwCpkx39AiOTgzhz2g@mail.gmail.com>
 <87h9zpbbal.fsf@gnu.org>
Message-ID: <1412103441.2908.2.camel@nomad.lan>

On Tue, 2014-09-30 at 13:25 +0200, Ludovic Court?s wrote:

> > I think there is some text missing here...
> 
> Commit e9941db adds a NEWS entry.  Could you cherry-pick it to the
> relevant branch(es)?

Done, thanks.