[Buildroot] [PATCH 0/4] package/xenomai: Fixes, updates, enhancements

[Buildroot] [PATCH 0/4] package/xenomai: Fixes, updates, enhancements

[Jan Kiszka]

Just a couple of improvements I came across while building me a virtual
ppc32 test target.

Jan

PS: @CC, sorry, had to send twice as the list is not open.

Jan Kiszka (4):
  package/xenomai: Smokey's not a skin
  package/xenomai: Pull from git
  package/xenomai: Bump version to 3.0.10
  package/xenomai: Clean up wrappers cleanup

 package/xenomai/Config.in    |  5 -----
 package/xenomai/xenomai.hash |  2 +-
 package/xenomai/xenomai.mk   | 39 +++++++++++++++++++++------------------
 3 files changed, 22 insertions(+), 24 deletions(-)

-- 
2.16.4

[Buildroot] [PATCH 1/4] package/xenomai: Smokey's not a skin

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

That's a testsuite tool.

Rename XENOMAI_REMOVE_SKIN_LIST at this chance as it's cleaning
libraries.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 package/xenomai/Config.in  |  5 -----
 package/xenomai/xenomai.mk | 23 ++++++++++-------------
 2 files changed, 10 insertions(+), 18 deletions(-)

diff --git a/package/xenomai/Config.in b/package/xenomai/Config.in
index 83ba8ca162..0cb0b343de 100644
--- a/package/xenomai/Config.in
+++ b/package/xenomai/Config.in
@@ -170,11 +170,6 @@ config BR2_PACKAGE_XENOMAI_RTAI_SKIN
 	help
 	  This option allows to install the RTAI Xenomai skin
 
-config BR2_PACKAGE_XENOMAI_SMOKEY_SKIN
-	bool "Smokey skin"
-	help
-	  This option allows to install the Smokey Xenomai skin
-
 config BR2_PACKAGE_XENOMAI_UITRON_SKIN
 	bool "uITRON skin"
 	help
diff --git a/package/xenomai/xenomai.mk b/package/xenomai/xenomai.mk
index 16c486eb34..f25adf747b 100644
--- a/package/xenomai/xenomai.mk
+++ b/package/xenomai/xenomai.mk
@@ -73,7 +73,7 @@ XENOMAI_POST_INSTALL_TARGET_HOOKS += XENOMAI_REMOVE_UNNEEDED_FILES
 ifeq ($(BR2_PACKAGE_XENOMAI_TESTSUITE),)
 define XENOMAI_REMOVE_TESTSUITE
 	rm -rf $(TARGET_DIR)/usr/share/xenomai/
-	for i in clocktest gpiotest latency spitest switchtest \
+	for i in clocktest gpiotest latency smokey spitest switchtest \
 		xeno-test-run-wrapper dohell xeno-test-run xeno-test ; do \
 		rm -f $(TARGET_DIR)/usr/bin/$$i ; \
 	done
@@ -109,24 +109,21 @@ endef
 XENOMAI_POST_INSTALL_TARGET_HOOKS += XENOMAI_REMOVE_ANALOGY
 endif
 
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_NATIVE_SKIN),,alchemy)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_POSIX_SKIN),,posix)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_PSOS_SKIN),,psos)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_RTAI_SKIN),,rtai)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_SMOKEY_SKIN),,smokey)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_UITRON_SKIN),,uitron)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_VXWORKS_SKIN),,vxworks)
-XENOMAI_REMOVE_SKIN_LIST += $(if $(BR2_PACKAGE_XENOMAI_VRTX_SKIN),,vrtx)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_NATIVE_SKIN),,alchemy)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_POSIX_SKIN),,posix)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_PSOS_SKIN),,psos)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_RTAI_SKIN),,rtai)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_TESTSUITE),,smokey)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_UITRON_SKIN),,uitron)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_VXWORKS_SKIN),,vxworks)
+XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_VRTX_SKIN),,vrtx)
 
 define XENOMAI_REMOVE_SKINS
-	for i in $(XENOMAI_REMOVE_SKIN_LIST) ; do \
+	for i in $(XENOMAI_REMOVE_LIBS_LIST) ; do \
 		rm -f $(TARGET_DIR)/usr/lib/lib$$i.* ; \
 		if [ $$i == "posix" ] ; then \
 			rm -f $(TARGET_DIR)/usr/lib/posix.wrappers ; \
 		fi ; \
-		if [ $$i == "smokey" ] ; then \
-			rm -f $(TARGET_DIR)/usr/bin/smokey* ; \
-		fi ; \
 	done
 endef
 
-- 
2.16.4

[Buildroot] [PATCH 1/4] package/xenomai: Smokey's not a skin

[Thomas Petazzoni]

On Thu, 12 Dec 2019 19:44:37 +0100
Jan Kiszka <jan.kiszka@siemens.com> wrote:

> From: Jan Kiszka <jan.kiszka@siemens.com>
> 
> That's a testsuite tool.
> 
> Rename XENOMAI_REMOVE_SKIN_LIST at this chance as it's cleaning
> libraries.

I am confused, aren't psos, rtai, uitron, vxworks and al. called
"skins" in the Xenomai speak ?

Ideally, this variable rename should be done in a separate patch, since
it's independent from the smokey option removal.

> Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
> ---
>  package/xenomai/Config.in  |  5 -----
>  package/xenomai/xenomai.mk | 23 ++++++++++-------------
>  2 files changed, 10 insertions(+), 18 deletions(-)
> 
> diff --git a/package/xenomai/Config.in b/package/xenomai/Config.in
> index 83ba8ca162..0cb0b343de 100644
> --- a/package/xenomai/Config.in
> +++ b/package/xenomai/Config.in
> @@ -170,11 +170,6 @@ config BR2_PACKAGE_XENOMAI_RTAI_SKIN
>  	help
>  	  This option allows to install the RTAI Xenomai skin
>  
> -config BR2_PACKAGE_XENOMAI_SMOKEY_SKIN
> -	bool "Smokey skin"
> -	help
> -	  This option allows to install the Smokey Xenomai skin

We normally want to have Config.in.legacy handling when removing
options, but in this case, the options was in fact not really doing
anything useful, so I believe it's OK to just drop it like you did.

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [PATCH 1/4] package/xenomai: Smokey's not a skin

[Jan Kiszka]

On 12.12.19 21:48, Thomas Petazzoni wrote:
> On Thu, 12 Dec 2019 19:44:37 +0100
> Jan Kiszka <jan.kiszka@siemens.com> wrote:
> 
>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>
>> That's a testsuite tool.
>>
>> Rename XENOMAI_REMOVE_SKIN_LIST at this chance as it's cleaning
>> libraries.
> 
> I am confused, aren't psos, rtai, uitron, vxworks and al. called
> "skins" in the Xenomai speak ?

Yes, they are and remain so.

> 
> Ideally, this variable rename should be done in a separate patch, since
> it's independent from the smokey option removal.

We still use XENOMAI_REMOVE_LIBS_LIST to remove the smokey *lib* when
the testsuite is disabled. Therefore, XENOMAI_REMOVE_SKIN_LIST ws not an
appropriate name. Same context, therefore same patch.

Jan

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux

[Buildroot] [PATCH 2/4] package/xenomai: Pull from git

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

This allows to pick also git revisions via XENOMAI_VERSION.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 package/xenomai/xenomai.hash |  2 +-
 package/xenomai/xenomai.mk   | 10 ++++++++--
 2 files changed, 9 insertions(+), 3 deletions(-)

diff --git a/package/xenomai/xenomai.hash b/package/xenomai/xenomai.hash
index 0fc560f978..83994384eb 100644
--- a/package/xenomai/xenomai.hash
+++ b/package/xenomai/xenomai.hash
@@ -1,2 +1,2 @@
 # Locally computed;
-sha256 9e4b879ee48e6423f76cba5882f41d129164fe977119080ff41ef531ee50dc4e  xenomai-3.0.9.tar.bz2
+sha256 be7a7a453a8d1f0763ed5feb3bee159fe60b8a42326b0ff564e94184bd4d0734  xenomai-v3.0.9.tar.bz2
diff --git a/package/xenomai/xenomai.mk b/package/xenomai/xenomai.mk
index f25adf747b..1d38cffa30 100644
--- a/package/xenomai/xenomai.mk
+++ b/package/xenomai/xenomai.mk
@@ -6,12 +6,12 @@
 
 XENOMAI_VERSION = $(call qstrip,$(BR2_PACKAGE_XENOMAI_VERSION))
 ifeq ($(XENOMAI_VERSION),)
-XENOMAI_VERSION = 3.0.9
+XENOMAI_VERSION = v3.0.9
 else
 BR_NO_CHECK_HASH_FOR += $(XENOMAI_SOURCE)
 endif
 
-XENOMAI_SITE = http://xenomai.org/downloads/xenomai/stable
+XENOMAI_SITE = https://gitlab.denx.de/Xenomai/xenomai/-/archive/$(XENOMAI_VERSION)
 XENOMAI_SOURCE = xenomai-$(XENOMAI_VERSION).tar.bz2
 XENOMAI_LICENSE = GPL-2.0+ with exception (headers), LGPL-2.1+ (libraries), GPL-2.0+ (kernel), GFDL-1.2+ (docs), GPL-2.0 (ipipe patch, can driver)
 # GFDL is not included but refers to gnu.org
@@ -28,6 +28,12 @@ XENOMAI_INSTALL_STAGING = YES
 XENOMAI_INSTALL_TARGET_OPTS = DESTDIR=$(TARGET_DIR) install-user
 XENOMAI_INSTALL_STAGING_OPTS = DESTDIR=$(STAGING_DIR) install-user
 
+define XENOMAI_BOOTSTRAP
+	(cd $(@D); ./scripts/bootstrap)
+endef
+
+XENOMAI_PRE_CONFIGURE_HOOKS += XENOMAI_BOOTSTRAP
+
 XENOMAI_CONF_OPTS += --includedir=/usr/include/xenomai/ --disable-doc-install
 
 ifeq ($(BR2_PACKAGE_XENOMAI_MERCURY),y)
-- 
2.16.4

[Buildroot] [PATCH 2/4] package/xenomai: Pull from git

[Thomas Petazzoni]

Hello Jan,

On Thu, 12 Dec 2019 19:44:38 +0100
Jan Kiszka <jan.kiszka@siemens.com> wrote:

> From: Jan Kiszka <jan.kiszka@siemens.com>
> 
> This allows to pick also git revisions via XENOMAI_VERSION.

You don't have tarballs for new releases? We generally prefer to use
tarballs when available.

>  XENOMAI_VERSION = $(call qstrip,$(BR2_PACKAGE_XENOMAI_VERSION))
>  ifeq ($(XENOMAI_VERSION),)
> -XENOMAI_VERSION = 3.0.9
> +XENOMAI_VERSION = v3.0.9

The version needs to stay 3.0.9, i.e without the "v" prefix.

>  else
>  BR_NO_CHECK_HASH_FOR += $(XENOMAI_SOURCE)
>  endif
>  
> -XENOMAI_SITE = http://xenomai.org/downloads/xenomai/stable
> +XENOMAI_SITE = https://gitlab.denx.de/Xenomai/xenomai/-/archive/$(XENOMAI_VERSION)

.. which can be done by changing this to:

XENOMAI_SITE = https://gitlab.denx.de/Xenomai/xenomai/-/archive/v$(XENOMAI_VERSION)

We want this so that our XENOMAI_VERSION value matches the
release-monitoring.org entry for Xenomai:
https://release-monitoring.org/project/18451/. BTW, it seems like they
don't know about version 3.0.10, as they are checking
https://xenomai.org/downloads/xenomai/stable/ and there is no tarball
for 3.0.10. I guess we need to change the Xenomai entry in
release-monitoring.org to make it use the Git repository.

> +define XENOMAI_BOOTSTRAP
> +	(cd $(@D); ./scripts/bootstrap)
> +endef
> +
> +XENOMAI_PRE_CONFIGURE_HOOKS += XENOMAI_BOOTSTRAP

This is not good, because nothing guarantees you autoconf/automake are
available. Our preferred way of handling this is to use:

XENOMAI_AUTORECONF = YES

which simply calls autoreconf on the project. However, I can imagine
that Xenomai is a bit special, so if AUTORECONF = YES doesn't work, you
can keep your PRE_CONFIGURE_HOOKS, but you need to add host-autoconf
and host-automake to XENOMAI_DEPENDENCIES, and possibly host-libtool if
libtool is used to build the libraries.

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com

[Buildroot] [PATCH 2/4] package/xenomai: Pull from git

[Jan Kiszka]

On 12.12.19 21:52, Thomas Petazzoni wrote:
> Hello Jan,
> 
> On Thu, 12 Dec 2019 19:44:38 +0100
> Jan Kiszka <jan.kiszka@siemens.com> wrote:
> 
>> From: Jan Kiszka <jan.kiszka@siemens.com>
>>
>> This allows to pick also git revisions via XENOMAI_VERSION.
> 
> You don't have tarballs for new releases? We generally prefer to use
> tarballs when available.

We do, but pulling those from git is more flexible as it allows to
specify also non-release states.

> 
>>  XENOMAI_VERSION = $(call qstrip,$(BR2_PACKAGE_XENOMAI_VERSION))
>>  ifeq ($(XENOMAI_VERSION),)
>> -XENOMAI_VERSION = 3.0.9
>> +XENOMAI_VERSION = v3.0.9
> 
> The version needs to stay 3.0.9, i.e without the "v" prefix.

If that is a general requirement on <pkg>_VERSION variables, I need to
model pulling git SHAs differently, likely as the kernel recipe does.

> 
>>  else
>>  BR_NO_CHECK_HASH_FOR += $(XENOMAI_SOURCE)
>>  endif
>>  
>> -XENOMAI_SITE = http://xenomai.org/downloads/xenomai/stable
>> +XENOMAI_SITE = https://gitlab.denx.de/Xenomai/xenomai/-/archive/$(XENOMAI_VERSION)
> 
> .. which can be done by changing this to:
> 
> XENOMAI_SITE = https://gitlab.denx.de/Xenomai/xenomai/-/archive/v$(XENOMAI_VERSION)

That breaks XENOMAI_VERSION=<branch> or <sha>, see above.

> 
> We want this so that our XENOMAI_VERSION value matches the
> release-monitoring.org entry for Xenomai:
> https://release-monitoring.org/project/18451/. BTW, it seems like they
> don't know about version 3.0.10, as they are checking
> https://xenomai.org/downloads/xenomai/stable/ and there is no tarball
> for 3.0.10. I guess we need to change the Xenomai entry in
> release-monitoring.org to make it use the Git repository.

We have some delay in updating the regular download area. There is also
currently none of our 3.1-rc releases available under testing/.

> 
>> +define XENOMAI_BOOTSTRAP
>> +	(cd $(@D); ./scripts/bootstrap)
>> +endef
>> +
>> +XENOMAI_PRE_CONFIGURE_HOOKS += XENOMAI_BOOTSTRAP
> 
> This is not good, because nothing guarantees you autoconf/automake are
> available. Our preferred way of handling this is to use:
> 
> XENOMAI_AUTORECONF = YES
> 
> which simply calls autoreconf on the project. However, I can imagine
> that Xenomai is a bit special, so if AUTORECONF = YES doesn't work, you
> can keep your PRE_CONFIGURE_HOOKS, but you need to add host-autoconf
> and host-automake to XENOMAI_DEPENDENCIES, and possibly host-libtool if
> libtool is used to build the libraries.

$ cat scripts/bootstrap
#!/bin/sh

autoreconf -fi


Looks unspectacular. Will try AUTORECONF = YES.

Thanks,
Jan

-- 
Siemens AG, Corporate Technology, CT RDA IOT SES-DE
Corporate Competence Center Embedded Linux

[Buildroot] [PATCH 3/4] package/xenomai: Bump version to 3.0.10

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 package/xenomai/xenomai.hash | 2 +-
 package/xenomai/xenomai.mk   | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/package/xenomai/xenomai.hash b/package/xenomai/xenomai.hash
index 83994384eb..31a925f2c4 100644
--- a/package/xenomai/xenomai.hash
+++ b/package/xenomai/xenomai.hash
@@ -1,2 +1,2 @@
 # Locally computed;
-sha256 be7a7a453a8d1f0763ed5feb3bee159fe60b8a42326b0ff564e94184bd4d0734  xenomai-v3.0.9.tar.bz2
+sha256 de72ef7a4522bf9ba5f72a674acac35a92eb38823d3631201a33bea958d9fb7d  xenomai-v3.0.10.tar.bz2
diff --git a/package/xenomai/xenomai.mk b/package/xenomai/xenomai.mk
index 1d38cffa30..ac823407f2 100644
--- a/package/xenomai/xenomai.mk
+++ b/package/xenomai/xenomai.mk
@@ -6,7 +6,7 @@
 
 XENOMAI_VERSION = $(call qstrip,$(BR2_PACKAGE_XENOMAI_VERSION))
 ifeq ($(XENOMAI_VERSION),)
-XENOMAI_VERSION = v3.0.9
+XENOMAI_VERSION = v3.0.10
 else
 BR_NO_CHECK_HASH_FOR += $(XENOMAI_SOURCE)
 endif
-- 
2.16.4

[Buildroot] [PATCH 4/4] package/xenomai: Clean up wrappers cleanup

[Jan Kiszka]

From: Jan Kiszka <jan.kiszka@siemens.com>

There is no posix.wrappers anymore, but cobalt and modechk. Those only
play a role when building in combination with wrap-link.sh.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
---
 package/xenomai/xenomai.mk | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/package/xenomai/xenomai.mk b/package/xenomai/xenomai.mk
index ac823407f2..fe0268b14c 100644
--- a/package/xenomai/xenomai.mk
+++ b/package/xenomai/xenomai.mk
@@ -68,6 +68,9 @@ define XENOMAI_REMOVE_UNNEEDED_FILES
 	for i in xeno xeno-config xeno-info wrap-link.sh ; do \
 		rm -f $(TARGET_DIR)/usr/bin/$$i ; \
 	done
+	for i in cobalt modechk ; do \
+		rm -f $(TARGET_DIR)/usr/lib/$$i.wrappers ; \
+	done
 	for i in autotune corectl hdb rtnet nomaccfg rtcfg rtifconfig \
 		rtiwconfig rtping rtroute tdmacfg rtps slackspot version; do \
 		rm -f $(TARGET_DIR)/usr/sbin/$$i ; \
@@ -127,9 +130,6 @@ XENOMAI_REMOVE_LIBS_LIST += $(if $(BR2_PACKAGE_XENOMAI_VRTX_SKIN),,vrtx)
 define XENOMAI_REMOVE_SKINS
 	for i in $(XENOMAI_REMOVE_LIBS_LIST) ; do \
 		rm -f $(TARGET_DIR)/usr/lib/lib$$i.* ; \
-		if [ $$i == "posix" ] ; then \
-			rm -f $(TARGET_DIR)/usr/lib/posix.wrappers ; \
-		fi ; \
 	done
 endef
 
-- 
2.16.4

[Buildroot] [PATCH 0/4] package/xenomai: Fixes, updates, enhancements

[Thomas Petazzoni]

Hello Jan,

On Thu, 12 Dec 2019 19:44:36 +0100
Jan Kiszka <jan.kiszka@siemens.com> wrote:

> Just a couple of improvements I came across while building me a virtual
> ppc32 test target.
> 
> Jan
> 
> PS: @CC, sorry, had to send twice as the list is not open.

Thanks a lot for sending those patches, it's nice to see patches from
you on our Xenomai package! I have a few comments/questions on the
patches, I'll reply individually to the patches.

> Jan Kiszka (4):
>   package/xenomai: Smokey's not a skin
>   package/xenomai: Pull from git
>   package/xenomai: Bump version to 3.0.10
>   package/xenomai: Clean up wrappers cleanup

A minor nit: we typically don't capitalize commit titles in Buildroot,
i.e "pull from git" and not "Pull from git". I know, that's arbitrary,
people like it one way or the other, we simply have more or less a
convention.

Thomas
-- 
Thomas Petazzoni, CTO, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com