[Buildroot] [PATCH 1/3] knock: new package

[Buildroot] [PATCH 1/3] knock: new package

[Michael Rommel]

A port knocking implementation with daemon and user application.

Signed-off-by: Michael Rommel <rommel@layer-7.net>
---
 package/Config.in       |    1 +
 package/knock/Config.in |   13 +++++++++++++
 package/knock/knock.mk  |   13 +++++++++++++
 3 files changed, 27 insertions(+)
 create mode 100644 package/knock/Config.in
 create mode 100644 package/knock/knock.mk

diff --git a/package/Config.in b/package/Config.in
index 6306947..fd98da0 100644
--- a/package/Config.in
+++ b/package/Config.in
@@ -763,6 +763,7 @@ source "package/ipset/Config.in"
 source "package/iptables/Config.in"
 source "package/iw/Config.in"
 source "package/kismet/Config.in"
+source "package/knock/Config.in"
 source "package/lighttpd/Config.in"
 source "package/linknx/Config.in"
 source "package/links/Config.in"
diff --git a/package/knock/Config.in b/package/knock/Config.in
new file mode 100644
index 0000000..8d80274
--- /dev/null
+++ b/package/knock/Config.in
@@ -0,0 +1,13 @@
+config BR2_PACKAGE_KNOCK
+	bool "knock"
+	select BR2_PACKAGE_LIBPCAP
+	help
+	  A port knocking implementation.
+	  Provides a daemon and a user application. Port knocking can be
+	  used to run an arbitrary application, once the daemon detects
+	  a predefined sequence of incoming TCP/UDP packets on a network
+	  interface. This can be used, e.g. to open up ports in a
+	  firewall.
+
+	  http://www.zeroflux.org/projects/knock
+
diff --git a/package/knock/knock.mk b/package/knock/knock.mk
new file mode 100644
index 0000000..85dfe62
--- /dev/null
+++ b/package/knock/knock.mk
@@ -0,0 +1,13 @@
+################################################################################
+#
+# knock
+#
+################################################################################
+
+KNOCK_VERSION = 7666f2e86e
+KNOCK_SITE = https://github.com/jvinet/knock/tarball/master
+KNOCK_LICENSE = GPLv2+
+KNOCK_LICENSE_FILES = COPYING
+KNOCK_AUTORECONF = YES
+
+$(eval $(autotools-package))
-- 
1.7.9.5

[Buildroot] [PATCH 2/3] aiccu: new package

[Michael Rommel]

Automatic IPv6 Connectivity Configuration Utility for users of a
IPv6 tunnel broker, developed by sixxs.net

Signed-off-by: Michael Rommel <rommel@layer-7.net>
---
 package/Config.in                                  |    1 +
 package/aiccu/Config.in                            |   24 +++++
 .../aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch |  104 ++++++++++++++++++++
 package/aiccu/aiccu.mk                             |   25 +++++
 4 files changed, 154 insertions(+)
 create mode 100644 package/aiccu/Config.in
 create mode 100644 package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch
 create mode 100644 package/aiccu/aiccu.mk

diff --git a/package/Config.in b/package/Config.in
index fd98da0..ed7a509 100644
--- a/package/Config.in
+++ b/package/Config.in
@@ -714,6 +714,7 @@ source "package/sound-theme-freedesktop/Config.in"
 endmenu
 
 menu "Networking applications"
+source "package/aiccu/Config.in"
 source "package/aircrack-ng/Config.in"
 source "package/argus/Config.in"
 source "package/arptables/Config.in"
diff --git a/package/aiccu/Config.in b/package/aiccu/Config.in
new file mode 100644
index 0000000..008ada7
--- /dev/null
+++ b/package/aiccu/Config.in
@@ -0,0 +1,24 @@
+config BR2_PACKAGE_AICCU
+	bool "aiccu"
+	depends on BR2_INET_IPV6
+	depends on BR2_USE_WCHAR
+	select BR2_PACKAGE_GNUTLS
+	help
+	  SixXS Automatic IPv6 Connectivity Client Utility
+
+	  AICCU (Automatic IPv6 Connectivity Client Utility) makes it
+	  easy for users to get IPv6 connectivity. After having
+	  requested an account, tunnel and optionally a subnet, AICCU
+	  can be used to automatically configure the tunnel. AICCU
+	  supports TIC (Tunnel Information & Control protocol), which it
+	  uses for retrieving the tunnel configuration information,
+	  AYIYA, which allows tunnels to be created even behind
+	  firewalls and NAT's.
+
+	  http://www.sixxs.net/tools/aiccu/
+
+comment "aiccu requires a toolchain with IPv6 support"
+	depends on !BR2_INET_IPV6
+
+comment "aiccu requires a toolchain with WCHAR support"
+	depends on !BR2_USE_WCHAR
diff --git a/package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch b/package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch
new file mode 100644
index 0000000..d316f8f
--- /dev/null
+++ b/package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch
@@ -0,0 +1,104 @@
+aiccu.h, common.c, common.h: fixes for deprecated GNUTLS functions and types
+resolver.c: fixes for selection of wrong resolver function under uclibc
+
+Signed-off-by: Michael Rommel <rommel@layer-7.net>
+
+diff -purN aiccu_20070115.orig/common/aiccu.h aiccu_20070115/common/aiccu.h
+--- aiccu_20070115.orig/common/aiccu.h	2007-01-15 13:01:43.000000000 +0100
++++ aiccu_20070115/common/aiccu.h	2013-08-31 23:50:53.651936146 +0200
+@@ -111,7 +111,7 @@ struct AICCU_conf
+ #endif
+ 
+ #ifdef AICCU_GNUTLS
+-	gnutls_certificate_credentials	tls_cred;	/* GNUTLS credentials */
++	gnutls_certificate_credentials_t tls_cred;	/* GNUTLS credentials */
+ #endif
+ 
+ 	bool		daemonize;		/* Daemonize? */
+diff -purN aiccu_20070115.orig/common/common.c aiccu_20070115/common/common.c
+--- aiccu_20070115.orig/common/common.c	2006-12-21 15:08:50.000000000 +0100
++++ aiccu_20070115/common/common.c	2013-09-01 01:21:36.031396740 +0200
+@@ -271,9 +271,8 @@ TLSSOCKET sock_alloc(void);
+ TLSSOCKET sock_alloc(void)
+ {
+ #ifdef AICCU_GNUTLS
+-	/* Allow connections to servers that have OpenPGP keys as well */
+-	const int	cert_type_priority[3] = { GNUTLS_CRT_X509, GNUTLS_CRT_OPENPGP, 0 };
+ 	int		ret;
++	const char	*err;
+ #endif /* AICCU_GNUTLS*/
+ 
+ 	TLSSOCKET	sock;	
+@@ -297,11 +296,16 @@ TLSSOCKET sock_alloc(void)
+ 	}
+ 
+ 	/* Use default priorities */
+-	gnutls_set_default_priority(sock->session);
+-	/* XXX: Return value is not documented in GNUTLS documentation! */
+-
+-	gnutls_certificate_type_set_priority(sock->session, cert_type_priority);
+-	/* XXX: Return value is not documented in GNUTLS documentation! */
++	ret = gnutls_priority_set_direct(sock->session, "NORMAL", &err); 
++	if (ret < 0)
++	{
++		if (ret == GNUTLS_E_INVALID_REQUEST)
++		{
++			dolog( LOG_ERR, "TLS set priority failed, syntax error at: %s\n", err);
++		}
++		free(sock);
++		return NULL;
++	}
+ 
+ 	/* Configure the x509 credentials for the current session */
+ 	gnutls_credentials_set(sock->session, GNUTLS_CRD_CERTIFICATE, g_aiccu->tls_cred);
+@@ -474,7 +478,7 @@ bool sock_gotls(TLSSOCKET sock)
+ 	}
+ 
+ 	/* Set the transport */
+-	gnutls_transport_set_ptr(sock->session, (gnutls_transport_ptr)sock->socket);
++	gnutls_transport_set_ptr(sock->session, (gnutls_transport_ptr_t) sock->socket);
+ 
+ 	/* Perform the TLS handshake */
+ 	ret = gnutls_handshake(sock->session);
+diff -purN aiccu_20070115.orig/common/common.h aiccu_20070115/common/common.h
+--- aiccu_20070115.orig/common/common.h	2007-01-11 15:50:51.000000000 +0100
++++ aiccu_20070115/common/common.h	2013-08-31 23:26:13.683659455 +0200
+@@ -381,7 +381,7 @@ struct tlssocket
+ 	SOCKET			socket;
+ #ifdef AICCU_GNUTLS
+ 	bool			tls_active;	/* TLS active? */
+-	gnutls_session		session;	/* The GnuTLS sesision */
++	gnutls_session_t	session;	/* The GnuTLS sesision */
+ #endif
+ };
+ 
+diff -purN aiccu_20070115.orig/common/resolver.c aiccu_20070115/common/resolver.c
+--- aiccu_20070115.orig/common/resolver.c	2006-07-23 16:55:14.000000000 +0200
++++ aiccu_20070115/common/resolver.c	2013-08-31 23:44:31.574866862 +0200
+@@ -26,7 +26,7 @@
+ 
+ int getrrs(const char *label, int rrtype, void gotrec(unsigned int num, int type, const char *record))
+ {
+-#ifdef _LINUX
++#if defined(_LINUX) && !defined(__UCLIBC__)
+ 	struct __res_state	res;
+ #endif
+ 	unsigned char		answer[8192];
+@@ -38,7 +38,7 @@ int getrrs(const char *label, int rrtype
+ 	uint16_t		type = 0, class = 0;
+ 	uint32_t		ttl = 0;
+ 
+-#ifdef _LINUX
++#if defined(_LINUX) && !defined(__UCLIBC__)
+ 	memset(&res, 0, sizeof(res));
+ 	res.options = RES_DEBUG;
+ 	res_ninit(&res);
+@@ -47,7 +47,7 @@ int getrrs(const char *label, int rrtype
+ #endif
+ 
+ 	memset(answer, 0, sizeof(answer));
+-#ifdef _LINUX
++#if defined(_LINUX) && !defined(__UCLIBC__)
+ 	ret = res_nquery(&res, label, C_IN, rrtype, answer, sizeof(answer));
+ #else
+ 	ret = res_query(label, C_IN, rrtype, answer, sizeof(answer));
diff --git a/package/aiccu/aiccu.mk b/package/aiccu/aiccu.mk
new file mode 100644
index 0000000..da3cc3b
--- /dev/null
+++ b/package/aiccu/aiccu.mk
@@ -0,0 +1,25 @@
+################################################################################
+#
+# aiccu
+#
+################################################################################
+
+AICCU_VERSION = 20070115
+AICCU_SOURCE = aiccu_$(AICCU_VERSION).tar.gz
+AICCU_SITE = http://www.sixxs.net/archive/sixxs/aiccu/unix/
+AICCU_LICENSE = SixXS License, concise redistribution license
+AICCU_LICENSE_FILES = doc/LICENSE
+AICCU_DEPENDENCIES = gnutls
+
+define AICCU_BUILD_CMDS
+	$(MAKE) CC="$(TARGET_CC)" LD="$(TARGET_LD)" -C $(@D) all
+endef
+
+define AICCU_INSTALL_TARGET_CMDS
+	$(INSTALL) -D -m 0755 $(@D)/unix-console/aiccu $(TARGET_DIR)/usr/sbin
+	$(INSTALL) -D -m 0644 $(@D)/doc/aiccu.conf $(TARGET_DIR)/etc
+	$(INSTALL) -D -m 0755 $(@D)/doc/aiccu.init \
+		$(TARGET_DIR)/usr/sbin/S90aiccu
+endef
+
+$(eval $(generic-package))
-- 
1.7.9.5

[Buildroot] [PATCH 3/3] python-pyzmq: new package

[Michael Rommel]

The python language bindings for ZeroMQ.

Signed-off-by: Michael Rommel <rommel@layer-7.net>
---
 package/Config.in                                  |    1 +
 package/python-pyzmq/Config.in                     |   31 ++++++++++++++
 ...e-buildroot-zmq-version-instead-of-detect.patch |   43 ++++++++++++++++++++
 package/python-pyzmq/python-pyzmq.mk               |   43 ++++++++++++++++++++
 4 files changed, 118 insertions(+)
 create mode 100644 package/python-pyzmq/Config.in
 create mode 100644 package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
 create mode 100644 package/python-pyzmq/python-pyzmq.mk

diff --git a/package/Config.in b/package/Config.in
index ed7a509..d5e3b52 100644
--- a/package/Config.in
+++ b/package/Config.in
@@ -371,6 +371,7 @@ source "package/python-protobuf/Config.in"
 source "package/python-pygame/Config.in"
 source "package/python-pyparsing/Config.in"
 source "package/python-pyro/Config.in"
+source "package/python-pyzmq/Config.in"
 source "package/python-serial/Config.in"
 source "package/python-setuptools/Config.in"
 source "package/python-thrift/Config.in"
diff --git a/package/python-pyzmq/Config.in b/package/python-pyzmq/Config.in
new file mode 100644
index 0000000..350f995
--- /dev/null
+++ b/package/python-pyzmq/Config.in
@@ -0,0 +1,31 @@
+config BR2_PACKAGE_PYTHON_PYZMQ
+	bool "pyzmq"
+	depends on BR2_INSTALL_LIBSTDCPP # zeromq
+	depends on BR2_INET_IPV6 # zeromq
+	depends on BR2_TOOLCHAIN_HAS_THREADS # zeromq
+	depends on BR2_LARGEFILE # zeromq
+	depends on BR2_USE_WCHAR # zeromq
+	select BR2_PACKAGE_ZEROMQ
+	select BR2_PACKAGE_PYTHON_HASHLIB
+	help
+	  This package contains the python language binding for zeromq.
+	  Due to issues with cross-compiling this package is hardcoded
+	  to the ZeroMQ version of this buildroot; the ZeroMQ version 
+	  must be at least v3.2.2
+
+	  http://zeromq.org/bindings:python
+
+comment "pyzmq requires a toolchain with C++ support enabled"
+	depends on !BR2_INSTALL_LIBSTDCPP
+
+comment "pyzmq requires a toolchain with IPv6 support"
+	depends on !BR2_INET_IPV6
+
+comment "pyzmq requires a toolchain with thread support"
+	depends on !BR2_TOOLCHAIN_HAS_THREADS
+
+comment "pyzmq requires a toolchain with LARGEFILE support"
+	depends on !BR2_LARGEFILE
+
+comment "pyzmq requires a toolchain with WCHAR support"
+	depends on !BR2_USE_WCHAR
diff --git a/package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch b/package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
new file mode 100644
index 0000000..75a72c4
--- /dev/null
+++ b/package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
@@ -0,0 +1,43 @@
+detect.py: fix the ZMQ version check to the ZMQ version of the buildroot
+
+The setup.py script tries to compile a test C program and runs it, to
+retrieve a version string for the installed ZMQ library, but if the cross
+compiler links it together, the result cannot be run on the host, due to
+different architectures and libraries.
+
+And if the host compiler would compile/link it, it would not link with the
+library version inside buildroot but with the library from the host, possibly returning a wrong version number.
+
+Instead of trying to run the compiled test program to get the version
+dynamically, return the version of the buildroot environment.
+
+Signed-off-by: Michael Rommel <rommel@layer-7.net>
+
+--- pyzmq-13.0.2/buildutils/detect.py.orig	2013-08-10 00:49:28.242557978 +0200
++++ pyzmq-13.0.2/buildutils/detect.py	2013-08-10 00:44:35.197572704 +0200
+@@ -118,15 +118,17 @@ def detect_zmq(basedir, compiler=None, *
+             
+     efile = test_compilation(cfile, compiler=compiler, **compiler_attrs)
+     
+-    result = Popen(efile, stdout=PIPE, stderr=PIPE)
+-    so, se = result.communicate()
++    # result = Popen(efile, stdout=PIPE, stderr=PIPE)
++    # so, se = result.communicate()
+     # for py3k:
+-    so = so.decode()
+-    se = se.decode()
+-    if result.returncode:
+-        msg = "Error running version detection script:\n%s\n%s" % (so,se)
+-        logging.error(msg)
+-        raise IOError(msg)
++    #so = so.decode()
++    #se = se.decode()
++    #if result.returncode:
++    #    msg = "Error running version detection script:\n%s\n%s" % (so,se)
++    #    logging.error(msg)
++    #    raise IOError(msg)
++
++    so = "vers: ##ZEROMQ_VERSION##"
+ 
+     handlers = {'vers':  lambda val: tuple(int(v) for v in val.split('.'))}
+ 
diff --git a/package/python-pyzmq/python-pyzmq.mk b/package/python-pyzmq/python-pyzmq.mk
new file mode 100644
index 0000000..9061f84
--- /dev/null
+++ b/package/python-pyzmq/python-pyzmq.mk
@@ -0,0 +1,43 @@
+################################################################################
+#
+# pyzmq
+#
+################################################################################
+
+PYTHON_PYZMQ_VERSION = 13.1.0
+PYTHON_PYZMQ_SOURCE = pyzmq-$(PYTHON_PYZMQ_VERSION).tar.gz
+PYTHON_PYZMQ_SITE = https://pypi.python.org/packages/source/p/pyzmq/
+PYTHON_PYZMQ_LICENSE = LGPLv3+ BSD-3c Apache License Version 2.0 
+# Apache license only online: http://www.apache.org/licenses/LICENSE-2.0
+PYTHON_PYZMQ_LICENSE_FILES = COPYING.LESSER COPYING.BSD
+PYTHON_PYZMQ_DEPENDENCIES = zeromq python
+
+define PYTHON_PYZMQ_POST_PATCH_FIXUP
+	$(SED) 's/##ZEROMQ_VERSION##/$(ZEROMQ_VERSION)/' \
+		$(@D)/buildutils/detect.py
+endef
+
+PYTHON_PYZMQ_POST_PATCH_HOOKS += PYTHON_PYZMQ_POST_PATCH_FIXUP
+
+PYTHON_PYZMQ_PARAMS = CC="$(TARGET_CC)" \
+		CFLAGS="$(TARGET_CFLAGS)" \
+		LDSHARED="$(TARGET_CC) -shared" \
+		CROSS_COMPILING=yes \
+		_python_sysroot=$(STAGING_DIR) \
+		_python_srcdir=$(BUILD_DIR)/python$(PYTHON_VERSION) \
+		_python_prefix=/usr \
+		_python_exec_prefix=/usr
+
+define PYTHON_PYZMQ_CONFIGURE_CMDS
+	(cd $(@D); $(PYTHON_PYZMQ_PARAMS) \
+		$(HOST_DIR)/usr/bin/python setup.py configure \
+		--zmq=$(STAGING_DIR)/usr)
+endef
+
+define PYTHON_PYZMQ_INSTALL_TARGET_CMDS
+	(cd $(@D); $(PYTHON_PYZMQ_PARAMS) \
+		$(HOST_DIR)/usr/bin/python setup.py install \
+		--prefix=$(TARGET_DIR)/usr)
+endef
+
+$(eval $(generic-package))
-- 
1.7.9.5

[Buildroot] [PATCH 3/3] python-pyzmq: new package

[Thomas De Schampheleire]

Op 1-sep.-2013 23:35 schreef "Michael Rommel" <rommel@layer-7.net> het
volgende:
>
> The python language bindings for ZeroMQ.
>
> Signed-off-by: Michael Rommel <rommel@layer-7.net>
> ---
>  package/Config.in                                  |    1 +
>  package/python-pyzmq/Config.in                     |   31 ++++++++++++++
>  ...e-buildroot-zmq-version-instead-of-detect.patch |   43
++++++++++++++++++++
>  package/python-pyzmq/python-pyzmq.mk               |   43
++++++++++++++++++++
>  4 files changed, 118 insertions(+)
>  create mode 100644 package/python-pyzmq/Config.in
>  create mode 100644
package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
>  create mode 100644 package/python-pyzmq/python-pyzmq.mk
>
> diff --git a/package/Config.in b/package/Config.in
> index ed7a509..d5e3b52 100644
> --- a/package/Config.in
> +++ b/package/Config.in
> @@ -371,6 +371,7 @@ source "package/python-protobuf/Config.in"
>  source "package/python-pygame/Config.in"
>  source "package/python-pyparsing/Config.in"
>  source "package/python-pyro/Config.in"
> +source "package/python-pyzmq/Config.in"
>  source "package/python-serial/Config.in"
>  source "package/python-setuptools/Config.in"
>  source "package/python-thrift/Config.in"
> diff --git a/package/python-pyzmq/Config.in
b/package/python-pyzmq/Config.in
> new file mode 100644
> index 0000000..350f995
> --- /dev/null
> +++ b/package/python-pyzmq/Config.in
> @@ -0,0 +1,31 @@
> +config BR2_PACKAGE_PYTHON_PYZMQ
> +       bool "pyzmq"
> +       depends on BR2_INSTALL_LIBSTDCPP # zeromq
> +       depends on BR2_INET_IPV6 # zeromq
> +       depends on BR2_TOOLCHAIN_HAS_THREADS # zeromq
> +       depends on BR2_LARGEFILE # zeromq
> +       depends on BR2_USE_WCHAR # zeromq
> +       select BR2_PACKAGE_ZEROMQ
> +       select BR2_PACKAGE_PYTHON_HASHLIB
> +       help
> +         This package contains the python language binding for zeromq.
> +         Due to issues with cross-compiling this package is hardcoded
> +         to the ZeroMQ version of this buildroot; the ZeroMQ version
> +         must be at least v3.2.2

Is it relevant to mention this, since the zeromq version is not
configurable in buildroot?

> +
> +         http://zeromq.org/bindings:python
> +
> +comment "pyzmq requires a toolchain with C++ support enabled"
> +       depends on !BR2_INSTALL_LIBSTDCPP
> +
> +comment "pyzmq requires a toolchain with IPv6 support"
> +       depends on !BR2_INET_IPV6
> +
> +comment "pyzmq requires a toolchain with thread support"
> +       depends on !BR2_TOOLCHAIN_HAS_THREADS
> +
> +comment "pyzmq requires a toolchain with LARGEFILE support"
> +       depends on !BR2_LARGEFILE
> +
> +comment "pyzmq requires a toolchain with WCHAR support"
> +       depends on !BR2_USE_WCHAR
> diff --git
a/package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
b/package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
> new file mode 100644
> index 0000000..75a72c4
> --- /dev/null
> +++
b/package/python-pyzmq/python-pyzmq-0001-use-buildroot-zmq-version-instead-of-detect.patch
> @@ -0,0 +1,43 @@
> +detect.py: fix the ZMQ version check to the ZMQ version of the buildroot
> +
> +The setup.py script tries to compile a test C program and runs it, to
> +retrieve a version string for the installed ZMQ library, but if the cross
> +compiler links it together, the result cannot be run on the host, due to
> +different architectures and libraries.
> +
> +And if the host compiler would compile/link it, it would not link with
the
> +library version inside buildroot but with the library from the host,
possibly returning a wrong version number.
> +
> +Instead of trying to run the compiled test program to get the version
> +dynamically, return the version of the buildroot environment.
> +
> +Signed-off-by: Michael Rommel <rommel@layer-7.net>
> +
> +--- pyzmq-13.0.2/buildutils/detect.py.orig     2013-08-10
00:49:28.242557978 +0200
> ++++ pyzmq-13.0.2/buildutils/detect.py  2013-08-10 00:44:35.197572704
+0200
> +@@ -118,15 +118,17 @@ def detect_zmq(basedir, compiler=None, *
> +
> +     efile = test_compilation(cfile, compiler=compiler, **compiler_attrs)
> +
> +-    result = Popen(efile, stdout=PIPE, stderr=PIPE)
> +-    so, se = result.communicate()
> ++    # result = Popen(efile, stdout=PIPE, stderr=PIPE)
> ++    # so, se = result.communicate()
> +     # for py3k:
> +-    so = so.decode()
> +-    se = se.decode()
> +-    if result.returncode:
> +-        msg = "Error running version detection script:\n%s\n%s" %
(so,se)
> +-        logging.error(msg)
> +-        raise IOError(msg)
> ++    #so = so.decode()
> ++    #se = se.decode()
> ++    #if result.returncode:
> ++    #    msg = "Error running version detection script:\n%s\n%s" %
(so,se)
> ++    #    logging.error(msg)
> ++    #    raise IOError(msg)
> ++
> ++    so = "vers: ##ZEROMQ_VERSION##"
> +
> +     handlers = {'vers':  lambda val: tuple(int(v) for v in
val.split('.'))}
> +
> diff --git a/package/python-pyzmq/python-pyzmq.mk b/package/python-pyzmq/
python-pyzmq.mk
> new file mode 100644
> index 0000000..9061f84
> --- /dev/null
> +++ b/package/python-pyzmq/python-pyzmq.mk
> @@ -0,0 +1,43 @@
>
+################################################################################
> +#
> +# pyzmq
> +#
>
+################################################################################
> +
> +PYTHON_PYZMQ_VERSION = 13.1.0
> +PYTHON_PYZMQ_SOURCE = pyzmq-$(PYTHON_PYZMQ_VERSION).tar.gz
> +PYTHON_PYZMQ_SITE = https://pypi.python.org/packages/source/p/pyzmq/
> +PYTHON_PYZMQ_LICENSE = LGPLv3+ BSD-3c Apache License Version 2.0
> +# Apache license only online: http://www.apache.org/licenses/LICENSE-2.0
> +PYTHON_PYZMQ_LICENSE_FILES = COPYING.LESSER COPYING.BSD
> +PYTHON_PYZMQ_DEPENDENCIES = zeromq python
> +
> +define PYTHON_PYZMQ_POST_PATCH_FIXUP

We generally name hooks after what they do, like
PYTHON_PYZMQ_PATCH_ZEROMQ_VERSION for example, instead of such a generic
name. If there are several distinct actions to take, multiple hooks are
created.

> +       $(SED) 's/##ZEROMQ_VERSION##/$(ZEROMQ_VERSION)/' \
> +               $(@D)/buildutils/detect.py
> +endef
> +
> +PYTHON_PYZMQ_POST_PATCH_HOOKS += PYTHON_PYZMQ_POST_PATCH_FIXUP
> +
> +PYTHON_PYZMQ_PARAMS = CC="$(TARGET_CC)" \
> +               CFLAGS="$(TARGET_CFLAGS)" \
> +               LDSHARED="$(TARGET_CC) -shared" \
> +               CROSS_COMPILING=yes \
> +               _python_sysroot=$(STAGING_DIR) \
> +               _python_srcdir=$(BUILD_DIR)/python$(PYTHON_VERSION) \

$(PYTHON_DIR) should work fine here.


> +               _python_prefix=/usr \
> +               _python_exec_prefix=/usr
> +
> +define PYTHON_PYZMQ_CONFIGURE_CMDS
> +       (cd $(@D); $(PYTHON_PYZMQ_PARAMS) \
> +               $(HOST_DIR)/usr/bin/python setup.py configure \
> +               --zmq=$(STAGING_DIR)/usr)
> +endef
> +
> +define PYTHON_PYZMQ_INSTALL_TARGET_CMDS
> +       (cd $(@D); $(PYTHON_PYZMQ_PARAMS) \
> +               $(HOST_DIR)/usr/bin/python setup.py install \
> +               --prefix=$(TARGET_DIR)/usr)
> +endef

Since you're using $(HOST_DIR)/usr/bin/python here, there is a dependency
on host-python which is not explicitly expressed. Since python itself
already depends on host-python (see package/python/python.mk), the
dependency is implicitly satisfied.
I check and there are some python packages that do explicitly depend on
host-python, while others do not. It seems there is no clear strategy for
this yet. Ideally, these things should be lined up. My view would be to add
the explicit dependency, if the .mk file explicitly references the host
python binary, but this may not be the view of others.


> +
> +$(eval $(generic-package))
> --
> 1.7.9.5

Best regards,

Thomas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.busybox.net/pipermail/buildroot/attachments/20130902/9354919b/attachment-0001.html>

[Buildroot] [PATCH 2/3] aiccu: new package

[Thomas De Schampheleire]

On Sun, Sep 1, 2013 at 11:33 PM, Michael Rommel <rommel@layer-7.net> wrote:
> Automatic IPv6 Connectivity Configuration Utility for users of a
> IPv6 tunnel broker, developed by sixxs.net
>
> Signed-off-by: Michael Rommel <rommel@layer-7.net>
> ---
>  package/Config.in                                  |    1 +
>  package/aiccu/Config.in                            |   24 +++++
>  .../aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch |  104 ++++++++++++++++++++
>  package/aiccu/aiccu.mk                             |   25 +++++
>  4 files changed, 154 insertions(+)
>  create mode 100644 package/aiccu/Config.in
>  create mode 100644 package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch
>  create mode 100644 package/aiccu/aiccu.mk
>
> diff --git a/package/Config.in b/package/Config.in
> index fd98da0..ed7a509 100644
> --- a/package/Config.in
> +++ b/package/Config.in
> @@ -714,6 +714,7 @@ source "package/sound-theme-freedesktop/Config.in"
>  endmenu
>
>  menu "Networking applications"
> +source "package/aiccu/Config.in"
>  source "package/aircrack-ng/Config.in"
>  source "package/argus/Config.in"
>  source "package/arptables/Config.in"
> diff --git a/package/aiccu/Config.in b/package/aiccu/Config.in
> new file mode 100644
> index 0000000..008ada7
> --- /dev/null
> +++ b/package/aiccu/Config.in
> @@ -0,0 +1,24 @@
> +config BR2_PACKAGE_AICCU
> +       bool "aiccu"
> +       depends on BR2_INET_IPV6
> +       depends on BR2_USE_WCHAR
> +       select BR2_PACKAGE_GNUTLS
> +       help
> +         SixXS Automatic IPv6 Connectivity Client Utility
> +
> +         AICCU (Automatic IPv6 Connectivity Client Utility) makes it
> +         easy for users to get IPv6 connectivity. After having
> +         requested an account, tunnel and optionally a subnet, AICCU
> +         can be used to automatically configure the tunnel. AICCU
> +         supports TIC (Tunnel Information & Control protocol), which it
> +         uses for retrieving the tunnel configuration information,
> +         AYIYA, which allows tunnels to be created even behind
> +         firewalls and NAT's.
> +
> +         http://www.sixxs.net/tools/aiccu/
> +
> +comment "aiccu requires a toolchain with IPv6 support"
> +       depends on !BR2_INET_IPV6
> +
> +comment "aiccu requires a toolchain with WCHAR support"
> +       depends on !BR2_USE_WCHAR
> diff --git a/package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch b/package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch
> new file mode 100644
> index 0000000..d316f8f
> --- /dev/null
> +++ b/package/aiccu/aiccu-0001-gnutls-and-uclibc-fixes.patch
> @@ -0,0 +1,104 @@
> +aiccu.h, common.c, common.h: fixes for deprecated GNUTLS functions and types
> +resolver.c: fixes for selection of wrong resolver function under uclibc
> +
> +Signed-off-by: Michael Rommel <rommel@layer-7.net>
> +
> +diff -purN aiccu_20070115.orig/common/aiccu.h aiccu_20070115/common/aiccu.h
> +--- aiccu_20070115.orig/common/aiccu.h 2007-01-15 13:01:43.000000000 +0100
> ++++ aiccu_20070115/common/aiccu.h      2013-08-31 23:50:53.651936146 +0200
> +@@ -111,7 +111,7 @@ struct AICCU_conf
> + #endif
> +
> + #ifdef AICCU_GNUTLS
> +-      gnutls_certificate_credentials  tls_cred;       /* GNUTLS credentials */
> ++      gnutls_certificate_credentials_t tls_cred;      /* GNUTLS credentials */
> + #endif
> +
> +       bool            daemonize;              /* Daemonize? */
> +diff -purN aiccu_20070115.orig/common/common.c aiccu_20070115/common/common.c
> +--- aiccu_20070115.orig/common/common.c        2006-12-21 15:08:50.000000000 +0100
> ++++ aiccu_20070115/common/common.c     2013-09-01 01:21:36.031396740 +0200
> +@@ -271,9 +271,8 @@ TLSSOCKET sock_alloc(void);
> + TLSSOCKET sock_alloc(void)
> + {
> + #ifdef AICCU_GNUTLS
> +-      /* Allow connections to servers that have OpenPGP keys as well */
> +-      const int       cert_type_priority[3] = { GNUTLS_CRT_X509, GNUTLS_CRT_OPENPGP, 0 };
> +       int             ret;
> ++      const char      *err;
> + #endif /* AICCU_GNUTLS*/
> +
> +       TLSSOCKET       sock;
> +@@ -297,11 +296,16 @@ TLSSOCKET sock_alloc(void)
> +       }
> +
> +       /* Use default priorities */
> +-      gnutls_set_default_priority(sock->session);
> +-      /* XXX: Return value is not documented in GNUTLS documentation! */
> +-
> +-      gnutls_certificate_type_set_priority(sock->session, cert_type_priority);
> +-      /* XXX: Return value is not documented in GNUTLS documentation! */
> ++      ret = gnutls_priority_set_direct(sock->session, "NORMAL", &err);
> ++      if (ret < 0)
> ++      {
> ++              if (ret == GNUTLS_E_INVALID_REQUEST)
> ++              {
> ++                      dolog( LOG_ERR, "TLS set priority failed, syntax error at: %s\n", err);
> ++              }
> ++              free(sock);
> ++              return NULL;
> ++      }
> +
> +       /* Configure the x509 credentials for the current session */
> +       gnutls_credentials_set(sock->session, GNUTLS_CRD_CERTIFICATE, g_aiccu->tls_cred);
> +@@ -474,7 +478,7 @@ bool sock_gotls(TLSSOCKET sock)
> +       }
> +
> +       /* Set the transport */
> +-      gnutls_transport_set_ptr(sock->session, (gnutls_transport_ptr)sock->socket);
> ++      gnutls_transport_set_ptr(sock->session, (gnutls_transport_ptr_t) sock->socket);
> +
> +       /* Perform the TLS handshake */
> +       ret = gnutls_handshake(sock->session);
> +diff -purN aiccu_20070115.orig/common/common.h aiccu_20070115/common/common.h
> +--- aiccu_20070115.orig/common/common.h        2007-01-11 15:50:51.000000000 +0100
> ++++ aiccu_20070115/common/common.h     2013-08-31 23:26:13.683659455 +0200
> +@@ -381,7 +381,7 @@ struct tlssocket
> +       SOCKET                  socket;
> + #ifdef AICCU_GNUTLS
> +       bool                    tls_active;     /* TLS active? */
> +-      gnutls_session          session;        /* The GnuTLS sesision */
> ++      gnutls_session_t        session;        /* The GnuTLS sesision */
> + #endif
> + };
> +
> +diff -purN aiccu_20070115.orig/common/resolver.c aiccu_20070115/common/resolver.c
> +--- aiccu_20070115.orig/common/resolver.c      2006-07-23 16:55:14.000000000 +0200
> ++++ aiccu_20070115/common/resolver.c   2013-08-31 23:44:31.574866862 +0200
> +@@ -26,7 +26,7 @@
> +
> + int getrrs(const char *label, int rrtype, void gotrec(unsigned int num, int type, const char *record))
> + {
> +-#ifdef _LINUX
> ++#if defined(_LINUX) && !defined(__UCLIBC__)
> +       struct __res_state      res;
> + #endif
> +       unsigned char           answer[8192];
> +@@ -38,7 +38,7 @@ int getrrs(const char *label, int rrtype
> +       uint16_t                type = 0, class = 0;
> +       uint32_t                ttl = 0;
> +
> +-#ifdef _LINUX
> ++#if defined(_LINUX) && !defined(__UCLIBC__)
> +       memset(&res, 0, sizeof(res));
> +       res.options = RES_DEBUG;
> +       res_ninit(&res);
> +@@ -47,7 +47,7 @@ int getrrs(const char *label, int rrtype
> + #endif
> +
> +       memset(answer, 0, sizeof(answer));
> +-#ifdef _LINUX
> ++#if defined(_LINUX) && !defined(__UCLIBC__)
> +       ret = res_nquery(&res, label, C_IN, rrtype, answer, sizeof(answer));
> + #else
> +       ret = res_query(label, C_IN, rrtype, answer, sizeof(answer));
> diff --git a/package/aiccu/aiccu.mk b/package/aiccu/aiccu.mk
> new file mode 100644
> index 0000000..da3cc3b
> --- /dev/null
> +++ b/package/aiccu/aiccu.mk
> @@ -0,0 +1,25 @@
> +################################################################################
> +#
> +# aiccu
> +#
> +################################################################################
> +
> +AICCU_VERSION = 20070115
> +AICCU_SOURCE = aiccu_$(AICCU_VERSION).tar.gz
> +AICCU_SITE = http://www.sixxs.net/archive/sixxs/aiccu/unix/
> +AICCU_LICENSE = SixXS License, concise redistribution license
> +AICCU_LICENSE_FILES = doc/LICENSE
> +AICCU_DEPENDENCIES = gnutls
> +
> +define AICCU_BUILD_CMDS
> +       $(MAKE) CC="$(TARGET_CC)" LD="$(TARGET_LD)" -C $(@D) all
> +endef
> +
> +define AICCU_INSTALL_TARGET_CMDS
> +       $(INSTALL) -D -m 0755 $(@D)/unix-console/aiccu $(TARGET_DIR)/usr/sbin
> +       $(INSTALL) -D -m 0644 $(@D)/doc/aiccu.conf $(TARGET_DIR)/etc
> +       $(INSTALL) -D -m 0755 $(@D)/doc/aiccu.init \
> +               $(TARGET_DIR)/usr/sbin/S90aiccu
> +endef

I'll ask the same question here: is there a specific reason why the
standard 'install' recipe is not suitable?

Best regards,
Thomas

[Buildroot] [PATCH 2/3] aiccu: new package

[Thomas Petazzoni]

Dear Michael Rommel,

Thanks a lot for your continuous effort with those patches. I now see
that you managed to use 'git rebase -i' and generate a very nice patch
series!

On Sun,  1 Sep 2013 23:33:35 +0200, Michael Rommel wrote:

> diff --git a/package/aiccu/Config.in b/package/aiccu/Config.in
> new file mode 100644
> index 0000000..008ada7
> --- /dev/null
> +++ b/package/aiccu/Config.in
> @@ -0,0 +1,24 @@
> +config BR2_PACKAGE_AICCU
> +	bool "aiccu"
> +	depends on BR2_INET_IPV6
> +	depends on BR2_USE_WCHAR
> +	select BR2_PACKAGE_GNUTLS
> +	help
> +	  SixXS Automatic IPv6 Connectivity Client Utility
> +
> +	  AICCU (Automatic IPv6 Connectivity Client Utility) makes it
> +	  easy for users to get IPv6 connectivity. After having
> +	  requested an account, tunnel and optionally a subnet, AICCU
> +	  can be used to automatically configure the tunnel. AICCU
> +	  supports TIC (Tunnel Information & Control protocol), which it
> +	  uses for retrieving the tunnel configuration information,
> +	  AYIYA, which allows tunnels to be created even behind
> +	  firewalls and NAT's.
> +
> +	  http://www.sixxs.net/tools/aiccu/
> +
> +comment "aiccu requires a toolchain with IPv6 support"
> +	depends on !BR2_INET_IPV6
> +
> +comment "aiccu requires a toolchain with WCHAR support"
> +	depends on !BR2_USE_WCHAR

We generally have only one comment for both. See examples in other
packages.

> diff --git a/package/aiccu/aiccu.mk b/package/aiccu/aiccu.mk
> new file mode 100644
> index 0000000..da3cc3b
> --- /dev/null
> +++ b/package/aiccu/aiccu.mk
> @@ -0,0 +1,25 @@
> +################################################################################
> +#
> +# aiccu
> +#
> +################################################################################
> +
> +AICCU_VERSION = 20070115
> +AICCU_SOURCE = aiccu_$(AICCU_VERSION).tar.gz
> +AICCU_SITE = http://www.sixxs.net/archive/sixxs/aiccu/unix/
> +AICCU_LICENSE = SixXS License, concise redistribution license
> +AICCU_LICENSE_FILES = doc/LICENSE
> +AICCU_DEPENDENCIES = gnutls
> +
> +define AICCU_BUILD_CMDS
> +	$(MAKE) CC="$(TARGET_CC)" LD="$(TARGET_LD)" -C $(@D) all

You could use:

	$(MAKE) $(TARGET_CONFIGURE_OPTS) -C $(@D) all

TARGET_CONFIGURE_OPTS contains the definitions of CC, LD, CFLAGS,
LDFLAGS, and more.

> +define AICCU_INSTALL_TARGET_CMDS
> +	$(INSTALL) -D -m 0755 $(@D)/unix-console/aiccu $(TARGET_DIR)/usr/sbin
> +	$(INSTALL) -D -m 0644 $(@D)/doc/aiccu.conf $(TARGET_DIR)/etc
> +	$(INSTALL) -D -m 0755 $(@D)/doc/aiccu.init \
> +		$(TARGET_DIR)/usr/sbin/S90aiccu
> +endef

Either use 'make install' as Thomas suggested, or if it's really not
possible, you need to make the second argument the complete filename,
not only the destination directory.

Moreover, the init script must be installed in /etc/init.d, not
in /usr/sbin. And finally, it should not be installed manually, but
instead should use the INIT_SYSV mechanism. See the Buildroot manual,
which states:

  LIBFOO_INSTALL_INIT_SYSV and LIBFOO_INSTALL_INIT_SYSTEMD list the
  actions to install init scripts either for the systemV-like init
  systems (busybox, sysvinit, etc.) or for the systemd units. These
  commands will be run only when the relevant init system is installed
  (i.e. if systemd is selected as the init system in the configuration,
  only LIBFOO_INSTALL_INIT_SYSTEMD will be run). 

Thanks!

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com

[Buildroot] [PATCH 3/3] python-pyzmq: new package

[Arnout Vandecappelle]

On 09/02/13 10:24, Thomas De Schampheleire wrote:
[snip]
>  > +define PYTHON_PYZMQ_INSTALL_TARGET_CMDS
>  > +       (cd $(@D); $(PYTHON_PYZMQ_PARAMS) \
>  > +               $(HOST_DIR)/usr/bin/python setup.py install \
>  > +               --prefix=$(TARGET_DIR)/usr)
>  > +endef
>
> Since you're using $(HOST_DIR)/usr/bin/python here, there is a dependency
> on host-python which is not explicitly expressed. Since python itself
> already depends on host-python (see package/python/python.mk
> <http://python.mk>), the dependency is implicitly satisfied.
> I check and there are some python packages that do explicitly depend on
> host-python, while others do not. It seems there is no clear strategy for
> this yet. Ideally, these things should be lined up. My view would be to
> add the explicit dependency, if the .mk file explicitly references the
> host python binary, but this may not be the view of others.

  I agree that dependencies that are specific to the package itself 
should be explicit, even if they are also implicit. Not just for things 
explicitly mentioned in the .mk file, but also e.g. for packages that are 
checked for in the configure script.


  Regards,
  Arnout

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F

[Buildroot] [PATCH 2/3] aiccu: new package

[Michael Rommel]

Dear all,

On Sep 2, 2013, at 10:57 , Thomas Petazzoni <thomas.petazzoni@free-electrons.com> wrote:

>> +AICCU_LICENSE_FILES = doc/LICENSE
>> +AICCU_DEPENDENCIES = gnutls
>> +
>> +define AICCU_BUILD_CMDS
>> +	$(MAKE) CC="$(TARGET_CC)" LD="$(TARGET_LD)" -C $(@D) all
> 
> You could use:
> 
> 	$(MAKE) $(TARGET_CONFIGURE_OPTS) -C $(@D) all
> 
> TARGET_CONFIGURE_OPTS contains the definitions of CC, LD, CFLAGS,
> LDFLAGS, and more.

I tried this, but it aborted with a lot of errors. I couldn't find the description for this
in the manual - where should I look about these variables, it seems there are
a lot more, e.g. $(SED) that might be handy...


>> +define AICCU_INSTALL_TARGET_CMDS
>> +	$(INSTALL) -D -m 0755 $(@D)/unix-console/aiccu $(TARGET_DIR)/usr/sbin
>> +	$(INSTALL) -D -m 0644 $(@D)/doc/aiccu.conf $(TARGET_DIR)/etc
>> +	$(INSTALL) -D -m 0755 $(@D)/doc/aiccu.init \
>> +		$(TARGET_DIR)/usr/sbin/S90aiccu
>> +endef
> 
> Either use 'make install' as Thomas suggested, or if it's really not
> possible, you need to make the second argument the complete filename,
> not only the destination directory.

I have now changed that in the latest submission. But I am not sure it is an 
improvement. I had to patch the Makefile, so that the correct, working 
version of the init script gets installed and had to manually correct the 
permissions in the .mk file. All in all, that makes the whole patch a lot 
larger.


> Moreover, the init script must be installed in /etc/init.d, not
> in /usr/sbin.

Yes, of course, my mistake!

> And finally, it should not be installed manually, but
> instead should use the INIT_SYSV mechanism. See the Buildroot manual,
> which states:
> 
>  LIBFOO_INSTALL_INIT_SYSV and LIBFOO_INSTALL_INIT_SYSTEMD list the
>  actions to install init scripts either for the systemV-like init
>  systems (busybox, sysvinit, etc.) or for the systemd units. These
>  commands will be run only when the relevant init system is installed
>  (i.e. if systemd is selected as the init system in the configuration,
>  only LIBFOO_INSTALL_INIT_SYSTEMD will be run). 

Actually I do not understand this well enough and the manual is 
not very explicit.

Now, either I shall use the "make install" approach, then I get what
the makefile does, which is "aiccu" in "/etc/init.d".

Or I can do it explicitly, which would then be more like:

+define AICCU_INSTALL_INIT_SYSV
+	$(INSTALL) -D -m 0755 $(@D)/doc/aiccu.init \
+		$(TARGET_DIR)/etc/init.d/S90aiccu
+endef

The Makefile does not support any other sensible targets, than
all and install. So I might be back to the original approach (of
course with the wrong dir corrected).

I'll send the submission and you can have a look what it looks
like now, after I incorporated your suggestions!

Thanks,

  Michael.

-- 
Michael Rommel, Erlangen, Germany

[Buildroot] [PATCH 3/3] python-pyzmq: new package

[Michael Rommel]

On Sep 2, 2013, at 17:55 , Arnout Vandecappelle <arnout@mind.be> wrote:
> On 09/02/13 10:24, Thomas De Schampheleire wrote:

Hi Arnout and Thomas,

I incorporated all your suggestions here and some additional changes I saw myself:
 - added host-python dependency
 - removed mentioning of minimum zeromq version needed
 - combined comments for dependencies on IPv6 and WCHAR
 - fixed naming of the patch hook
 - used $(PYTHON_DIR) - although for the heck of me I could not find out, where this is defined :-(
 - changed depends on PYTHON from select (actually, I got it wrong, using the select gave the cyclic dependency :-()
 - named the package, bool and filenames consistently python-pyzmq
 
Patch will come in some minutes.

Cheers,

   Michael.

-- 
Michael Rommel, Erlangen, Germany

[Buildroot] [PATCH 3/3] python-pyzmq: new package

[Arnout Vandecappelle]

On 09/03/13 20:04, Michael Rommel wrote:
> On Sep 2, 2013, at 17:55 , Arnout Vandecappelle <arnout@mind.be> wrote:
>> On 09/02/13 10:24, Thomas De Schampheleire wrote:
>
> Hi Arnout and Thomas,
>
> I incorporated all your suggestions here and some additional changes I saw myself:
>   - added host-python dependency
>   - removed mentioning of minimum zeromq version needed
>   - combined comments for dependencies on IPv6 and WCHAR
>   - fixed naming of the patch hook
>   - used $(PYTHON_DIR) - although for the heck of me I could not find out, where this is defined :-(

  This variable is defined as part of the $(pkg-generic) infrastructure.

  You can use 'make printvars | grep ^PYTHON_DIR=' to get the expansion. 
Hm, looks like this feature isn't mentioned in the manual...


  Regards,
  Arnout


>   - changed depends on PYTHON from select (actually, I got it wrong, using the select gave the cyclic dependency :-()
>   - named the package, bool and filenames consistently python-pyzmq
>
> Patch will come in some minutes.
>
> Cheers,
>
>     Michael.
>


-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F

[Buildroot] [PATCH 1/3] knock: new package

[Thomas De Schampheleire]

Hi Michael,

On Sun, Sep 1, 2013 at 11:33 PM, Michael Rommel <rommel@layer-7.net> wrote:
> A port knocking implementation with daemon and user application.
>
> Signed-off-by: Michael Rommel <rommel@layer-7.net>
> ---
>  package/Config.in       |    1 +
>  package/knock/Config.in |   13 +++++++++++++
>  package/knock/knock.mk  |   13 +++++++++++++
>  3 files changed, 27 insertions(+)
>  create mode 100644 package/knock/Config.in
>  create mode 100644 package/knock/knock.mk
>
> diff --git a/package/Config.in b/package/Config.in
> index 6306947..fd98da0 100644
> --- a/package/Config.in
> +++ b/package/Config.in
> @@ -763,6 +763,7 @@ source "package/ipset/Config.in"
>  source "package/iptables/Config.in"
>  source "package/iw/Config.in"
>  source "package/kismet/Config.in"
> +source "package/knock/Config.in"
>  source "package/lighttpd/Config.in"
>  source "package/linknx/Config.in"
>  source "package/links/Config.in"
> diff --git a/package/knock/Config.in b/package/knock/Config.in
> new file mode 100644
> index 0000000..8d80274
> --- /dev/null
> +++ b/package/knock/Config.in
> @@ -0,0 +1,13 @@
> +config BR2_PACKAGE_KNOCK
> +       bool "knock"
> +       select BR2_PACKAGE_LIBPCAP
> +       help
> +         A port knocking implementation.
> +         Provides a daemon and a user application. Port knocking can be
> +         used to run an arbitrary application, once the daemon detects
> +         a predefined sequence of incoming TCP/UDP packets on a network
> +         interface. This can be used, e.g. to open up ports in a
> +         firewall.
> +
> +         http://www.zeroflux.org/projects/knock
> +
> diff --git a/package/knock/knock.mk b/package/knock/knock.mk
> new file mode 100644
> index 0000000..85dfe62
> --- /dev/null
> +++ b/package/knock/knock.mk
> @@ -0,0 +1,13 @@
> +################################################################################
> +#
> +# knock
> +#
> +################################################################################
> +
> +KNOCK_VERSION = 7666f2e86e
> +KNOCK_SITE = https://github.com/jvinet/knock/tarball/master
> +KNOCK_LICENSE = GPLv2+
> +KNOCK_LICENSE_FILES = COPYING
> +KNOCK_AUTORECONF = YES
> +
> +$(eval $(autotools-package))

The Config.in file mentions a dependency on libpcap, but this is not
properly expressed in the .mk file.
You need:
KNOCK_DEPENDENCIES = libpcap
Without this, the compilation on a clean system fails.

Best regards,
Thomas

[Buildroot] [PATCH 1/3] knock: new package

[Thomas De Schampheleire]

Hi Michael,

On Sun, Sep 1, 2013 at 11:33 PM, Michael Rommel <rommel@layer-7.net> wrote:
> A port knocking implementation with daemon and user application.
>
> Signed-off-by: Michael Rommel <rommel@layer-7.net>
> ---
>  package/Config.in       |    1 +
>  package/knock/Config.in |   13 +++++++++++++
>  package/knock/knock.mk  |   13 +++++++++++++
>  3 files changed, 27 insertions(+)
>  create mode 100644 package/knock/Config.in
>  create mode 100644 package/knock/knock.mk
>
> diff --git a/package/Config.in b/package/Config.in
> index 6306947..fd98da0 100644
> --- a/package/Config.in
> +++ b/package/Config.in
> @@ -763,6 +763,7 @@ source "package/ipset/Config.in"
>  source "package/iptables/Config.in"
>  source "package/iw/Config.in"
>  source "package/kismet/Config.in"
> +source "package/knock/Config.in"
>  source "package/lighttpd/Config.in"
>  source "package/linknx/Config.in"
>  source "package/links/Config.in"
> diff --git a/package/knock/Config.in b/package/knock/Config.in
> new file mode 100644
> index 0000000..8d80274
> --- /dev/null
> +++ b/package/knock/Config.in
> @@ -0,0 +1,13 @@
> +config BR2_PACKAGE_KNOCK
> +       bool "knock"
> +       select BR2_PACKAGE_LIBPCAP
> +       help
> +         A port knocking implementation.
> +         Provides a daemon and a user application. Port knocking can be
> +         used to run an arbitrary application, once the daemon detects
> +         a predefined sequence of incoming TCP/UDP packets on a network
> +         interface. This can be used, e.g. to open up ports in a
> +         firewall.
> +
> +         http://www.zeroflux.org/projects/knock
> +
> diff --git a/package/knock/knock.mk b/package/knock/knock.mk
> new file mode 100644
> index 0000000..85dfe62
> --- /dev/null
> +++ b/package/knock/knock.mk
> @@ -0,0 +1,13 @@
> +################################################################################
> +#
> +# knock
> +#
> +################################################################################
> +
> +KNOCK_VERSION = 7666f2e86e
> +KNOCK_SITE = https://github.com/jvinet/knock/tarball/master
> +KNOCK_LICENSE = GPLv2+
> +KNOCK_LICENSE_FILES = COPYING
> +KNOCK_AUTORECONF = YES
> +
> +$(eval $(autotools-package))
> --

The Config.in file mentions a dependency on libpcap, but this is not
properly expressed in the .mk file.
You need:
KNOCK_DEPENDENCIES = libpcap
Without this, the compilation on a clean system fails.

Best regards,
Thomas

[Buildroot] [PATCH 1/3] knock: new package

[Michael Rommel]

Hi Thomas,

I got it. I could reproduce the issue with the missing dependency in the .mk file and have set up
a standard tree just for testing the build process. And with the new dependency in the .mk file, 
the build succeeds.

After the test build my tree looks like this:

rommel at pelican:/var/nfs/software/initrds/uclibc/buildroot-patchtest$ ls -la build/
total 108
drwxr-xr-x 24 rommel users  4096 Sep  8 16:04 .
drwxr-xr-x  7 rommel users  4096 Sep  8 16:02 ..
drwxr-xr-x  4 rommel users  4096 Sep  8 15:42 buildroot-config
drwxr-xr-x 34 rommel users  4096 Sep  8 16:00 busybox-1.21.1
drwxr-xr-x  9 rommel users  4096 Sep  8 15:47 host-autoconf-2.69
drwxr-xr-x  7 rommel users  4096 Sep  8 15:47 host-automake-1.11.6
drwxr-xr-x 18 rommel users  4096 Sep  8 15:45 host-binutils-2.21.1
drwxr-xr-x 14 rommel users  4096 Sep  8 16:03 host-bison-2.7.1
drwxr-xr-x  9 rommel users  4096 Sep  8 16:04 host-fakeroot-1.18.2
drwxr-xr-x 10 rommel users  4096 Sep  8 16:03 host-flex-2.5.37
drwxr-xr-x 31 rommel users  4096 Sep  8 16:00 host-gcc-final-4.7.3
drwxr-xr-x 31 rommel users  4096 Sep  8 15:49 host-gcc-initial-4.7.3
drwxr-xr-x 31 rommel users  4096 Sep  8 15:57 host-gcc-intermediate-4.7.3
drwxr-xr-x 16 rommel users  4096 Sep  8 15:46 host-gmp-5.1.2
drwxr-xr-x  5 rommel users  4096 Sep  8 15:47 host-libtool-2.4.2
drwxr-xr-x 10 rommel users  4096 Sep  8 15:46 host-m4-1.4.16
drwxr-xr-x  2 rommel users  4096 Sep  8 16:04 host-makedevs-undefined
drwxr-xr-x  7 rommel users  4096 Sep  8 15:47 host-mpc-1.0.1
drwxr-xr-x  9 rommel users  4096 Sep  8 15:46 host-mpfr-3.1.2
drwxr-xr-x  6 rommel users  4096 Sep  8 16:03 knock-7666f2e86e
drwxr-xr-x 13 rommel users 12288 Sep  8 16:03 libpcap-1.4.0
drwxr-xr-x 23 rommel users  4096 Sep  8 15:55 linux-headers-3.11
-rw-r--r--  1 rommel users     0 Sep  8 15:39 .root
drwxr-xr-x 19 rommel users  4096 Sep  8 15:58 uclibc-0.9.33.2
-rw-r--r--  1 rommel users     7 Sep  8 16:04 _users_table.txt
drwxr-xr-x 14 rommel users  4096 Sep  8 16:02 zlib-1.2.8

I will submit the patch later again.

Thanks,

  Michael.


On Sep 8, 2013, at 15:00 , Michael Rommel <rommel@layer-7.net> wrote:

> Hi Thomas,
> 
> thanks - will fix this in the next iteration. Is there a description, how one can go ahead with discovering things like
> this, i.e. an explanation of a "clean system"? I only have i586 infrastructure ready, so the error on the aiccu package
> I could not have found (but I will try to follow your suggestions and see, what I can come up with), but I certainly
> can set up a separate buildroot just for testing packages. I assume that a clean system is one, where I just get
> a clone of the git, apply the patch and just select this package - am I right or do I have to consider something else?
> 
> Thanks!
> 
>   Michael.
> 
> 
> On Sep 8, 2013, at 14:08 , Thomas De Schampheleire <patrickdepinguin@gmail.com> wrote:
> 
>> Hi Michael,
>> 
>>> +KNOCK_SITE = https://github.com/jvinet/knock/tarball/master
>>> +KNOCK_LICENSE = GPLv2+
>>> +KNOCK_LICENSE_FILES = COPYING
>>> +KNOCK_AUTORECONF = YES
>>> +
>>> +$(eval $(autotools-package))
>>> --
>> 
>> The Config.in file mentions a dependency on libpcap, but this is not
>> properly expressed in the .mk file.
>> You need:
>> KNOCK_DEPENDENCIES = libpcap
>> Without this, the compilation on a clean system fails.

-- 
Michael Rommel, Erlangen, Germany

[Buildroot] [PATCH 1/3] knock: new package

[Thomas De Schampheleire]

Hi Michael,

On Sun, Sep 8, 2013 at 4:09 PM, Michael Rommel <rommel@layer-7.net> wrote:
>
> Hi Thomas,
>
> I got it. I could reproduce the issue with the missing dependency in the .mk file and have set up
> a standard tree just for testing the build process. And with the new dependency in the .mk file,
> the build succeeds.

Great.
To discover such mandatory dependencies, you can use the following idiom:
make clean toolchain knock

This will first clean the buildroot tree, build/install the toolchain,
and then build only the knock package and its dependencies. We should
probably add this to the documentation...

Another tip: for these type of dependency problems, you could also use
an external toolchain, like the Sourcery ones, to get much quicker
results. With such external toolchains, it also becomes very easy to
test some other architectures like powerpc, arm, ...

Best regards,
Thomas

[Buildroot] [PATCH 1/3] knock: new package

[Arnout Vandecappelle]

On 08/09/13 16:32, Thomas De Schampheleire wrote:
> Another tip: for these type of dependency problems, you could also use
> an external toolchain, like the Sourcery ones, to get much quicker
> results. With such external toolchains, it also becomes very easy to
> test some other architectures like powerpc, arm, ...

  And you can also use the toolchains used by the autobuilders, to test 
with uClibc-based toolchains.

  Regards,
  Arnout

-- 
Arnout Vandecappelle                          arnout at mind be
Senior Embedded Software Architect            +32-16-286500
Essensium/Mind                                http://www.mind.be
G.Geenslaan 9, 3001 Leuven, Belgium           BE 872 984 063 RPR Leuven
LinkedIn profile: http://www.linkedin.com/in/arnoutvandecappelle
GPG fingerprint:  7CB5 E4CC 6C2E EFD4 6E3D A754 F963 ECAB 2450 2F1F

[Buildroot] [PATCH 1/3] knock: new package

[Thomas Petazzoni]

Hello,

On Tue, 10 Sep 2013 22:58:53 +0200, Arnout Vandecappelle wrote:

>   And you can also use the toolchains used by the autobuilders, to test 
> with uClibc-based toolchains.

Correct. As a reminder, I put all the toolchain defconfigs tested by
the autobuilders in
http://autobuild.buildroot.org/toolchains/configs/free-electrons/. Some
of them point to well-known external toolchains (Linaro, Sourcery
CodeBench, Blackfin, etc.), and some other point to external toolchains
that I have built specifically for the autobuilders, but the defconfig
contains an URL that is available for everyone.

Whenever I need to test some package, I often start with:

wget -O .config http://autobuild.buildroot.org/toolchains/configs/free-electrons/br-arm-basic.config

which will create me a minimal .config file that points to a very
simple uClibc toolchain.

Best regards,

Thomas
-- 
Thomas Petazzoni, Free Electrons
Kernel, drivers, real-time and embedded Linux
development, consulting, training and support.
http://free-electrons.com