Skip to contentSkip to conversation

Set SSL_OP_NO_TICKET flag in SSL context for OpenSSL builds, to disable TLS stateless session resumption.

PATCH In review James Yonan 7 participants 23 messages Mar 17, 2014 – Apr 1, 2014 1 message not in the archive

lore
Share & tools
lore

James Yonan

author
#0 lore raw

PATCH 1/1 Set SSL_OP_NO_TICKET flag in SSL context for OpenSSL builds, to disable TLS stateless session resumption.

OpenVPN doesn't want or need SSL session renegotiation or resumption, as it handles renegotiation on its own.

For this reason, OpenVPN always disables the SSL session cache:

SSL_CTX_set_session_cache_mode (ctx, SSL_SESS_CACHE_OFF)

However, even with the above code, stateless session resumption is still possible unless explicitly disabled with the SSL_OP_NO_TICKET flag. This patch does this.

---
 src/openvpn/ssl_openssl.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
src/openvpn/ssl_openssl.c +1 -1
index 0dc1e81..938e9d4 100644
@@ -208,7 +208,7 @@ tls_ctx_set_options (struct tls_root_ctx *ctx, unsigned int ssl_flags)
 
   /* process SSL options including minimum TLS version we will accept from peer */
   {
-    long sslopt = SSL_OP_SINGLE_DH_USE | SSL_OP_NO_SSLv2 | SSL_OP_NO_SSLv3;
+    long sslopt = SSL_OP_SINGLE_DH_USE | SSL_OP_NO_TICKET | SSL_OP_NO_SSLv2 | SSL_OP_NO_SSLv3;
     const int tls_version_min = (ssl_flags >> SSLF_TLS_VERSION_SHIFT) & SSLF_TLS_VERSION_MASK;
     if (tls_version_min > TLS_VER_UNSPEC)
       {
Signature · 1 line
1.8.5.3
#1 lore raw

Hi,

Quoted 10 lines · James Yonan

OpenVPN doesn't want or need SSL session renegotiation or resumption, as it handles renegotiation on its own.

For this reason, OpenVPN always disables the SSL session cache:

SSL_CTX_set_session_cache_mode (ctx, SSL_SESS_CACHE_OFF)

However, even with the above code, stateless session resumption is still possible unless explicitly disabled with the SSL_OP_NO_TICKET flag. This patch does this.

I assume this should go into all OpenVPN branches, that is, master, 2.3, and if we ever do another 2.2, into that one as well?

(not ACKing or NAKing the patch itself, this is not my field of expertise)

gert

Signature · 4 lines
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             [email protected]...
fax: +49-89-35655025                        [email protected]...
#2 lore raw

Hi,

Quoted 21 lines

-----Original Message-----
From: Gert Doering [mailto:[email protected]...]
Sent: maandag 17 maart 2014 9:34
Subject: Re: [Openvpn-devel] [PATCH] Set SSL_OP_NO_TICKET flag in SSL context for OpenSSL builds, to disable TLS stateless session resumption.

Hi,

Quoted 10 lines · James Yonan

OpenVPN doesn't want or need SSL session renegotiation or resumption, as it handles renegotiation on its own.

For this reason, OpenVPN always disables the SSL session cache:

SSL_CTX_set_session_cache_mode (ctx, SSL_SESS_CACHE_OFF)

However, even with the above code, stateless session resumption is still possible unless explicitly disabled with the SSL_OP_NO_TICKET flag. This patch does this.

Quoted 5 lines

I assume this should go into all OpenVPN branches, that is, master, 2.3, and if we ever do another 2.2, into that one as well?

(not ACKing or NAKing the patch itself, this is not my field of expertise)

I think this should go into all releases we'll do from now on.

Also, ACK on the patch. Together with SSL_SESS_CACHE_OFF, this seems to fully disable TLS session renegotiation and resumption.

-Steffan

#3 lore raw
Quoted 26 lines · Steffan Karger

Hi,

Quoted 24 lines

-----Original Message-----
From: Gert Doering [mailto:[email protected]...]
Sent: maandag 17 maart 2014 9:34
Subject: Re: [Openvpn-devel] [PATCH] Set SSL_OP_NO_TICKET flag in SSL context for OpenSSL builds, to disable TLS stateless session resumption.

Hi,

Quoted 10 lines · James Yonan

OpenVPN doesn't want or need SSL session renegotiation or resumption, as it handles renegotiation on its own.

For this reason, OpenVPN always disables the SSL session cache:

SSL_CTX_set_session_cache_mode (ctx, SSL_SESS_CACHE_OFF)

However, even with the above code, stateless session resumption is still possible unless explicitly disabled with the SSL_OP_NO_TICKET flag. This patch does this.

I assume this should go into all OpenVPN branches, that is, master, 2.3, and if we ever do another 2.2, into that one as well?

Quoted 7 lines
Quoted 2 lines

(not ACKing or NAKing the patch itself, this is not my field of expertise)

I think this should go into all releases we'll do from now on.

Also, ACK on the patch. Together with SSL_SESS_CACHE_OFF, this seems to fully disable TLS session renegotiation and resumption.

This patch only covers OpenSSL. Is there an equivalent for PolarSSL as well? Or isn't it needed at all on PolarSSL?

Signature · 3 lines
kind regards,

David Sommerseth
#4 lore raw

Hi,

Quoted 7 lines

-----Original Message-----
From: David Sommerseth [mailto:[email protected]...]
Sent: maandag 17 maart 2014 11:40
Subject: Re: [Openvpn-devel] [PATCH] Set SSL_OP_NO_TICKET flag in SSL context for OpenSSL builds, to disable TLS stateless session resumption.

Quoted 8 lines
Quoted 4 lines · Steffan Karger

I think this should go into all releases we'll do from now on.

Also, ACK on the patch. Together with SSL_SESS_CACHE_OFF, this seems to fully disable TLS session renegotiation and resumption.

This patch only covers OpenSSL. Is there an equivalent for PolarSSL as well? Or isn't it needed at all on PolarSSL?

PolarSSL disables session renegotiation by default (sane defaults ftw!), and OpenVPN does not enable it. So there's nothing to fix for PolarSSL.

-Steffan

#5 lore raw
In reply to James Yonan

Your patch has been applied to the master and release/2.3 branches.

commit 25f4d4b49bff342fd9dd54cd22f14c9de49e9f8b (master) commit 058e889d186c616c648802da2eadf0380086bd6e (release/2.3)

Author: James Yonan
Date:   Sun Mar 16 18:49:36 2014 -0600
     Set SSL_OP_NO_TICKET flag in SSL context for OpenSSL builds, to disable TLS stateless session resumption.

     Acked-by: Steffan Karger <[email protected]...>
     Message-Id: <[email protected]...>
     URL: http://article.gmane.org/gmane.network.openvpn.devel/8346
     Signed-off-by: Gert Doering <[email protected]...>

--
kind regards,

Gert Doering

#6 lore raw
In reply to James Yonan

Hi,

Quoted 3 lines · James Yonan

However, even with the above code, stateless session resumption is still possible unless explicitly disabled with the SSL_OP_NO_TICKET flag. This patch does this.

This actually raises an interesting question. My OpenSolaris buildslave fails compilation with

gcc -DHAVE_CONFIG_H -I. -I../.. -I../../include -I../../src/compat -g -O2 -MT ssl_openssl.o -MD -MP -MF .deps/ssl_openssl.Tpo -c -o ssl_openssl.o ssl_openssl.c ssl_openssl.c: In function `tls_ctx_set_options':
ssl_openssl.c:183: error: `SSL_OP_NO_TICKET' undeclared (first use in this function)

... while "configure" doesn't flag an error about the OpenSSL version installed (0.9.8a-fips). This is not ideal, if we know we're going to fail at compile time due to missing functionality, we should tell the user earlier.

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available
Fix 2: require a more recent OpenSSL version

Crypto guys, send me patches :-)

gert

Signature · 4 lines
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             [email protected]...
fax: +49-89-35655025                        [email protected]...
author
#7 lore raw
Quoted 19 lines · Gert Doering

Hi,

Quoted 3 lines · James Yonan

However, even with the above code, stateless session resumption is still possible unless explicitly disabled with the SSL_OP_NO_TICKET flag. This patch does this.

This actually raises an interesting question. My OpenSolaris buildslave fails compilation with

gcc -DHAVE_CONFIG_H -I. -I../.. -I../../include -I../../src/compat -g -O2 -MT ssl_openssl.o -MD -MP -MF .deps/ssl_openssl.Tpo -c -o ssl_openssl.o ssl_openssl.c ssl_openssl.c: In function `tls_ctx_set_options':
ssl_openssl.c:183: error: `SSL_OP_NO_TICKET' undeclared (first use in this function)

... while "configure" doesn't flag an error about the OpenSSL version installed (0.9.8a-fips). This is not ideal, if we know we're going to fail at compile time due to missing functionality, we should tell the user earlier.

Quoted 6 lines

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available
Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

James

#8 lore raw

Hi,

Quoted 11 lines · James Yonan
Quoted 6 lines · Gert Doering

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

SSL_OP_NO_TICKET was added in OpenSSL 0.9.8f / 1.0.0. The ECDH-patchset (for 2.4) already requires 0.9.8, so I would prefer to require 0.9.8f or newer for master/2.4, but just add #ifdef's for 2.3.

-Steffan

#9 lore raw
Quoted 5 lines · Steffan Karger

Hi,

Quoted 3 lines · James Yonan

On 17/03/2014 14:29, Gert Doering wrote:

Quoted 10 lines
Quoted 7 lines
Quoted 6 lines

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

SSL_OP_NO_TICKET was added in OpenSSL 0.9.8f / 1.0.0. The ECDH-patchset (for 2.4) already requires 0.9.8, so I would prefer to require 0.9.8f or newer for master/2.4, but just add #ifdef's for 2.3.

I disagree. It is not safe to assume that the #ifdef is bound to a particular version of Openssl; for example, on my Centos 6.5 box I have openssl 1.0.1e yet the define
#define SSL_OP_NO_TICKET 0x00004000L
is NOT present in the system ssl.h file.
Don't ask me why, don't question the sanity of CentOS/RedHat, don't assume that "openssl vX.Y.Z will have support for ...."
(RedHat also strips ECDH support from openssl due to claimed licensing restrictions).

Let's just check for the #ifdef regardless of the version of OpenSSL.

cheers,

JJK

A message that is not in the archive (sent off-list or not yet received): [email protected]...

#11 lore raw

Hi David,

Quoted 22 lines · David Sommerseth
Quoted 13 lines · Jan Just Keijser
Quoted 12 lines · Steffan Karger

Hi,

Quoted 10 lines · James Yonan
Quoted 6 lines · Gert Doering

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

Quoted 8 lines
Quoted 3 lines

SSL_OP_NO_TICKET was added in OpenSSL 0.9.8f / 1.0.0. The ECDH-patchset (for 2.4) already requires 0.9.8, so I would prefer to require 0.9.8f or newer for master/2.4, but just add #ifdef's for 2.3.

I disagree. It is not safe to assume that the #ifdef is bound to a particular version of Openssl; for example, on my Centos 6.5 box I have openssl 1.0.1e yet the define

    #define SSL_OP_NO_TICKET                0x00004000L

is NOT present in the system ssl.h file.

Quoted 6 lines

I just checked RHEL 6.5 and ScientificLinux 6.4 (openssl-1.0.1e-16.el6_5.4) ... they both have it this:

# grep SSL_OP_NO_TICKET /usr/include/openssl/*
/usr/include/openssl/ssl.h:#define SSL_OP_NO_TICKET	            0x00004000L

this is most odd - I just checked a few other machines (CentOS 6.5) and there the SSL_OP_NO_TICKET is present.
I then reinstalled openssl on the 'flawed' box and now it is present also.
So it seems I spoke too soon... sorry for the noise, although I must say that I'm still in favour of checking for the existence of an IFDEF instead of relying on a particular version...

cheers,

JJK

#12 lore raw

Hi,

Quoted 30 lines · Jan Just Keijser
Quoted 5 lines · David Sommerseth
Quoted 4 lines · Jan Just Keijser
Quoted 3 lines · Steffan Karger
Quoted 3 lines · James Yonan

On 17/03/2014 14:29, Gert Doering wrote:

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 6 lines

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

SSL_OP_NO_TICKET was added in OpenSSL 0.9.8f / 1.0.0. The ECDH-

patchset (for 2.4) already requires 0.9.8, so I would prefer to require 0.9.8f or newer for master/2.4, but just add #ifdef's for 2.3.

Quoted 7 lines
Quoted 7 lines
Quoted 2 lines

I disagree. It is not safe to assume that the #ifdef is bound to a particular version of Openssl; for example, on my Centos 6.5 box I have openssl 1.0.1e yet the define

    #define SSL_OP_NO_TICKET                0x00004000L

is NOT present in the system ssl.h file.

Quoted 5 lines

I just checked RHEL 6.5 and ScientificLinux 6.4 (openssl-1.0.1e-16.el6_5.4) ... they both have it this:

# grep SSL_OP_NO_TICKET /usr/include/openssl/*
/usr/include/openssl/ssl.h:#define SSL_OP_NO_TICKET

0x00004000L

Quoted 1 line
Quoted 7 lines

this is most odd - I just checked a few other machines (CentOS 6.5) and there the SSL_OP_NO_TICKET is present.
I then reinstalled openssl on the 'flawed' box and now it is present also.
So it seems I spoke too soon... sorry for the noise, although I must say that I'm still in favour of checking for the existence of an IFDEF instead of relying on a particular version...

Point taken, just checking the OpenSSL version does not suffice. Still, I'd like to prevent more #ifdef's in the code. We could check the #ifdef in configure.ac, and refuse to build when it's not present. Any objections against requiring SSL_OP_NO_TICKET to be present for OpenVPN 2.4+? We have to drop support for 'ancient stuff' at some point.

-Steffan

#13 lore raw
Quoted 41 lines · Steffan Karger

Hi,

Quoted 30 lines · Jan Just Keijser
Quoted 5 lines · David Sommerseth
Quoted 4 lines · Jan Just Keijser
Quoted 3 lines · Steffan Karger
Quoted 3 lines · James Yonan

On 17/03/2014 14:29, Gert Doering wrote:

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 6 lines

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Fix 1: only use SSL_OP_NO_TICKET if available Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

SSL_OP_NO_TICKET was added in OpenSSL 0.9.8f / 1.0.0. The ECDH-

patchset (for 2.4) already requires 0.9.8, so I would prefer to require 0.9.8f or newer for master/2.4, but just add #ifdef's for 2.3.

Quoted 7 lines
Quoted 7 lines
Quoted 2 lines

I disagree. It is not safe to assume that the #ifdef is bound to a particular version of Openssl; for example, on my Centos 6.5 box I have openssl 1.0.1e yet the define

    #define SSL_OP_NO_TICKET                0x00004000L

is NOT present in the system ssl.h file.

Quoted 5 lines

I just checked RHEL 6.5 and ScientificLinux 6.4 (openssl-1.0.1e-16.el6_5.4) ... they both have it this:

# grep SSL_OP_NO_TICKET /usr/include/openssl/*
/usr/include/openssl/ssl.h:#define SSL_OP_NO_TICKET

0x00004000L

Quoted 1 line
Quoted 7 lines

this is most odd - I just checked a few other machines (CentOS 6.5) and there the SSL_OP_NO_TICKET is present.
I then reinstalled openssl on the 'flawed' box and now it is present also.
So it seems I spoke too soon... sorry for the noise, although I must say that I'm still in favour of checking for the existence of an IFDEF instead of relying on a particular version...

Quoted 5 lines

Point taken, just checking the OpenSSL version does not suffice. Still, I'd like to prevent more #ifdef's in the code. We could check the #ifdef in configure.ac, and refuse to build when it's not present. Any objections against requiring SSL_OP_NO_TICKET to be present for OpenVPN 2.4+? We have to drop support for 'ancient stuff' at some point.

I don't disagree with you ... but we need to think about what our users may have installed. James was concerned about RHEL4 support some years ago, when we discussed if we should support OpenSSL 0.9.6 or not (which was available in RHEL4, iirc). At that time we agreed upon moving towards a 0.9.8 requirement when RHEL4 was out of the normal support cycle (which was February 29, 2012).

RHEL5 is fully supported until March 31, 2017.

I think it makes sense to follow RHEL's life cycle, as that is usually one of the distributions at an enterprise level which is kept up-to-date on critical issues throughout its life cycle.

Signature · 3 lines
kind regards,

David Sommerseth
#14 lore raw
In reply to Steffan Karger

Hi,

Quoted 10 lines · Steffan Karger
Quoted 3 lines

So it seems I spoke too soon... sorry for the noise, although I must say that I'm still in favour of checking for the existence of an IFDEF instead of relying on a particular version...

Point taken, just checking the OpenSSL version does not suffice. Still, I'd like to prevent more #ifdef's in the code. We could check the #ifdef in configure.ac, and refuse to build when it's not present. Any objections against requiring SSL_OP_NO_TICKET to be present for OpenVPN 2.4+? We have to drop support for 'ancient stuff' at some point.

If I understood this all right, this feature improves OpenVPN security against yet-unknown attacks using a feature of OpenSSL that we don't use anyway. Right?

So in that case, I'm fine with your proposal - do something on configure.ac that will check for SSL_OP_NO_TICKET and complain (with a useful error message :) ) if it's not there.

For 2.3, I'd propose to add code to ssl_openssl.h like this:

/* SSL_OP_NO_TICKET tells OpenSSL to disable "stateless session resumption",
* as this is something we do not want nor need, but could potentially be
* used for a future attack. For compatibility reasons, in the 2.3.x
* series, we keep building if the OpenSSL version is too old to support
* this. 2.4 requires it and will fail configure if not present. */
#ifndef SSL_OP_NO_TICKET

# define SSL_OP_NO_TICKET 0
#endif

ACK? Is the message correct?

gert

Signature · 4 lines
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             [email protected]...
fax: +49-89-35655025                        [email protected]...
#15 lore raw

Hi,

Quoted 12 lines · Gert Doering
Quoted 10 lines · Steffan Karger
Quoted 3 lines

So it seems I spoke too soon... sorry for the noise, although I must say that I'm still in favour of checking for the existence of an IFDEF instead of relying on a particular version...

Point taken, just checking the OpenSSL version does not suffice. Still, I'd like to prevent more #ifdef's in the code. We could check the #ifdef in configure.ac, and refuse to build when it's not present. Any objections against requiring SSL_OP_NO_TICKET to be present for OpenVPN 2.4+? We have to drop support for 'ancient stuff' at some point.

Quoted 4 lines

If I understood this all right, this feature improves OpenVPN security against yet-unknown attacks using a feature of OpenSSL that we don't use anyway. Right?

Right.

So in that case, I'm fine with your proposal - do something on configure.ac

Quoted 13 lines

that will check for SSL_OP_NO_TICKET and complain (with a useful error message :) ) if it's not there.

For 2.3, I'd propose to add code to ssl_openssl.h like this:

/* SSL_OP_NO_TICKET tells OpenSSL to disable "stateless session resumption",
* as this is something we do not want nor need, but could potentially be
* used for a future attack. For compatibility reasons, in the 2.3.x
* series, we keep building if the OpenSSL version is too old to support
* this. 2.4 requires it and will fail configure if not present. */
#ifndef SSL_OP_NO_TICKET

Quoted 6 lines
# define SSL_OP_NO_TICKET 0
#endif

ACK? Is the message correct?

ACK. Message looks correct and clear to me (but please remove the double space before 2.4).

-Steffan

#16 lore raw

Hi,

Quoted 1 line · Steffan Karger

ACK. Message looks correct and clear to me.

Thanks. Committed and pushed as 2cf9d4e3f06f4a61cb6d159728ac6c8a790d6849.

Can you send the needed patch for master/2.4?

gert

Signature · 4 lines
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             [email protected]...
fax: +49-89-35655025                        [email protected]...
#17 lore raw
Quoted 7 lines · Gert Doering
Quoted 1 line · Steffan Karger

ACK. Message looks correct and clear to me.

Thanks. Committed and pushed as 2cf9d4e3f06f4a61cb6d159728ac6c8a790d6849.

Can you send the needed patch for master/2.4?

See attachment :)

-Steffan

From 4f9d47bee3bef4102dfe8e13da21ab4bbe0a92a9 Mon Sep 17 00:00:00 2001
From: Steffan Karger <[email protected]>
Date: Sun, 23 Mar 2014 14:07:47 +0100
Subject: [PATCH] configure.ac: check for SSL_OP_NO_TICKET flag in OpenSSL

SSL_OP_NO_TICKET tells OpenSSL to disable "stateless session resumption". This is something we do not want nor need, but could potentially be used for a future attack. OpenVPN 2.4 requires the flag to be set and will fail configure if the flag is not present.

---
 configure.ac | 15 +++++++++++++++
 1 file changed, 15 insertions(+)
configure.ac +15 -0
index c622f33..2da6521 100644
@@ -793,6 +793,21 @@ if test "${have_openssl_crypto}" = "yes"; then
 	LIBS="${saved_LIBS}"
 fi
 
+if test "${have_openssl_ssl}" = "yes"; then
+    AC_MSG_CHECKING([for SSL_OP_NO_TICKET flag in OpenSSL])
+    AC_EGREP_CPP(have_ssl_op_no_ticket, [
+    #include <openssl/ssl.h>
+    #ifdef SSL_OP_NO_TICKET
+    have_ssl_op_no_ticket
+    #endif
+    ], [
+        AC_MSG_RESULT([yes])
+    ], [
+        AC_MSG_RESULT([no])
+        AC_ERROR([OpenVPN 2.4+ requires SSL_OP_NO_TICKET in OpenSSL])
+    ])
+fi
+
 AC_ARG_VAR([POLARSSL_CFLAGS], [C compiler flags for polarssl])
 AC_ARG_VAR([POLARSSL_LIBS], [linker flags for polarssl])
 have_polarssl_ssl="yes"
Signature · 1 line
1.8.3.2
#18 lore raw

[Openvpn-devel] [PATCH applied] Re: configure.ac: check for SSL_OP_NO_TICKET flag in OpenSSL

ACK.

Tested on my "normal" OpenSSL installation (OK) and by deliberately removing SSL_OP_NO_TICKET from openssl/ssl.h -> correct error message:

..
checking for SSL_OP_NO_TICKET flag in OpenSSL... no
configure: error: OpenVPN 2.4+ requires SSL_OP_NO_TICKET in OpenSSL

Your patch has been applied to the master branch.

commit e9b088b20847905ed2c2b85a12be58f457c10d06

Author: Steffan Karger
Date:   Sun Mar 23 14:07:47 2014 +0100
     configure.ac: check for SSL_OP_NO_TICKET flag in OpenSSL

     Acked-by: Gert Doering <[email protected]...>
     Message-Id: <[email protected]...>
     URL: http://article.gmane.org/gmane.network.openvpn.devel/8389
     Signed-off-by: Gert Doering <[email protected]...>

--
kind regards,

Gert Doering

#19 lore raw
In reply to Steffan Karger

configure.ac if test "${have_openssl_crypto}" = "yes"; then · line 799

Quoted 2 lines · On Sunday 23 March 2014 14:27:43 Steffan Karger wrote:
configure.ac +2 -0
+    AC_EGREP_CPP(have_ssl_op_no_ticket, [
+    #include <openssl/ssl.h>

We just found that this breaks if the openssl headers are in a non-standard place. The test above sets the -I option in CFLAGS, but not in CPPFLAGS. So, maybe we should generally set -I in CPPFLAGS instead.

Heiko

#20 lore raw

Hi,

Quoted 7 lines · Heiko Hund

Review

Quoted 2 lines · On Sunday 23 March 2014 14:27:43 Steffan Karger wrote:
(quoted diff) +2 -0
+    AC_EGREP_CPP(have_ssl_op_no_ticket, [
+    #include <openssl/ssl.h>

We just found that this breaks if the openssl headers are in a non-standard place. The test above sets the -I option in CFLAGS, but not in CPPFLAGS. So, maybe we should generally set -I in CPPFLAGS instead.

I'm not much of an autotools wizard, so I don't really know if there are downsides to using CPPFLAGS insteadof CFLAGS everywhere.

The attached patch fixes your problem, but if there is someone around with a better idea (and, preferrably, patch) to fix it, I'm all ears!

-Steffan

From ccebcab605325f98a0ff22edbbbc089194e2e0ad Mon Sep 17 00:00:00 2001
From: Steffan Karger <[email protected]...>
Date: Tue, 1 Apr 2014 00:33:55 +0200
Subject: [PATCH] configure.ac: use CPPFLAGS for SSL_OP_NO_TICKET check

AC_EGREP_CPP uses CPPFLAGS, not CFLAGS. Make sure the macro can find OpenSSL by temporarily adding OPENSSL_CRYPTO_FLAGS to CPPFLAGS.

Signed-off-by Steffan Karger <[email protected]...>
---
 configure.ac | 3 +++
 1 file changed, 3 insertions(+)
configure.ac +3 -0
index 7e94280..0c2abb9 100644
@@ -794,6 +794,8 @@ if test "${have_openssl_crypto}" = "yes"; then
 fi
 
 if test "${have_openssl_ssl}" = "yes"; then
+    saved_CPPFLAGS="${CPPFLAGS}"
+    CPPFLAGS="${CPPFLAGS} ${OPENSSL_CRYPTO_CFLAGS}"
     AC_MSG_CHECKING([for SSL_OP_NO_TICKET flag in OpenSSL])
     AC_EGREP_CPP(have_ssl_op_no_ticket, [
     #include <openssl/ssl.h>

@@ -806,6 +808,7 @@ if test "${have_openssl_ssl}" = "yes"; then
         AC_MSG_RESULT([no])
         AC_ERROR([OpenVPN 2.4+ requires SSL_OP_NO_TICKET in OpenSSL])
     ])
+    CPPFLAGS="${saved_CPPFLAGS}"
 fi
 
 AC_ARG_VAR([POLARSSL_CFLAGS], [C compiler flags for polarssl])
Signature · 1 line
1.8.3.2
#21 lore raw
Quoted 2 lines · On Tuesday 01 April 2014 00:45:16 Steffan Karger wrote:

The attached patch fixes your problem, but if there is someone around with a better idea (and, preferrably, patch) to fix it, I'm all ears!

ACK, fixes the problem indeed, without introducing a new set of flags even.

Thanks
Heiko

#22 lore raw

[Openvpn-devel] [PATCH applied] Re: configure.ac: use CPPFLAGS for SSL_OP_NO_TICKET check

In reply to Steffan Karger

Your patch has been applied to the master branch.

commit e38f554cd4b956f6eedd56c7a76a3ba7c84ad824

Author: Steffan Karger
Date:   Tue Apr 1 00:33:55 2014 +0200
     configure.ac: use CPPFLAGS for SSL_OP_NO_TICKET check

     Signed-off-by: Steffan Karger <[email protected]...>
     Message-Id: <[email protected]...>
     URL: http://article.gmane.org/gmane.network.openvpn.devel/8419
     Acked-by: Heiko Hund <[email protected]...>

--
kind regards,

Gert Doering

#23 lore raw
In reply to Steffan Karger
Quoted 9 lines · Steffan Karger

Hi,

Quoted 7 lines · James Yonan
Quoted 4 lines · Gert Doering

Right now, if I read configure.ac correct, we require 0.9.6 or later (and check this only if pkg-config is available) - but obviously, SSL_OP_NO_TICKET was added later on.

Quoted 8 lines
Quoted 4 lines
Quoted 2 lines

Fix 1: only use SSL_OP_NO_TICKET if available Fix 2: require a more recent OpenSSL version

I would think an #ifdef should be fine.

SSL_OP_NO_TICKET was added in OpenSSL 0.9.8f / 1.0.0. The ECDH-patchset (for 2.4) already requires 0.9.8, so I would prefer to require 0.9.8f or newer for master/2.4, but just add #ifdef's for 2.3.

There's JJK's comments, which makes sense. In addition, RHEL 5 (which we're supposed to support) only ships openssl-0.9.8e.

Signature · 3 lines
kind regards,

David Sommerseth