[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Samuel Martin]

Openobex uses accept4 syscall, which is not available on Blackfin flavored uclibc.
So, diasble openobex on blackfin (do not add extra checks on the c-library since
uclibc is the only choice for Blackfin).

Fixed:
  http://autobuild.buildroot.net/results/78e033fe9f43845581a5d87b21a8451f67520e44

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
---
 package/openobex/Config.in | 1 +
 1 file changed, 1 insertion(+)

diff --git a/package/openobex/Config.in b/package/openobex/Config.in
index e611b8d803..4a4dc214ef 100644
--- a/package/openobex/Config.in
+++ b/package/openobex/Config.in
@@ -1,5 +1,6 @@
 config BR2_PACKAGE_OPENOBEX
 	bool "openobex"
+	depends on !BR2_bfin
 	help
 	  Free open source implementation of the Object Exchange (OBEX)
 	  protocol.
-- 
2.15.0

[Buildroot] [PATCH 2/4] package/opencv: fix build failure

[Samuel Martin]

Backport fix from upstream.

Fixes:
  http://autobuild.buildroot.net/results/b27d324331f6e351e95dd4742f4d0a50af60c590
  http://autobuild.buildroot.net/results/44ed0be0bd94028b7b37e7bf21233adc1753d94b

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
---
 .../0003-Update-OpenCVCompilerOptions.cmake.patch  | 33 ++++++++++++++++++++++
 1 file changed, 33 insertions(+)
 create mode 100644 package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch

diff --git a/package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch b/package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch
new file mode 100644
index 0000000000..290eef4d10
--- /dev/null
+++ b/package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch
@@ -0,0 +1,33 @@
+Fetch from upstream: https://github.com/opencv/opencv/commit/126de0cd95418811e302996161e03195bf7e631a
+
+From 126de0cd95418811e302996161e03195bf7e631a Mon Sep 17 00:00:00 2001
+From: neok-m4700 <neok-m4700@users.noreply.github.com>
+Date: Tue, 25 Jul 2017 10:25:20 +0200
+Subject: [PATCH] Update OpenCVCompilerOptions.cmake
+
+misplaced else
+
+Signed-off-by: Samuel Martin <s.martin49@gmail.com>
+---
+ cmake/OpenCVCompilerOptions.cmake | 4 ++--
+ 1 file changed, 2 insertions(+), 2 deletions(-)
+
+diff --git a/cmake/OpenCVCompilerOptions.cmake b/cmake/OpenCVCompilerOptions.cmake
+index 9b2a4e42e..7b079c5f9 100644
+--- a/cmake/OpenCVCompilerOptions.cmake
++++ b/cmake/OpenCVCompilerOptions.cmake
+@@ -18,9 +18,9 @@ if(ENABLE_CCACHE AND NOT CMAKE_COMPILER_IS_CCACHE)
+         message(STATUS "Unable to compile program with enabled ccache, reverting...")
+         set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE "${__OLD_RULE_LAUNCH_COMPILE}")
+       endif()
+-    else()
+-      message(STATUS "Looking for ccache - not found")
+     endif()
++  else()
++    message(STATUS "Looking for ccache - not found")
+   endif()
+ endif()
+ 
+-- 
+2.15.0
+
-- 
2.15.0

[Buildroot] [PATCH 2/4] package/opencv: fix build failure

[Thomas Petazzoni]

Hello,

On Wed,  1 Nov 2017 21:11:35 +0100, Samuel Martin wrote:
> Backport fix from upstream.
> 
> Fixes:
>   http://autobuild.buildroot.net/results/b27d324331f6e351e95dd4742f4d0a50af60c590
>   http://autobuild.buildroot.net/results/44ed0be0bd94028b7b37e7bf21233adc1753d94b
> 
> Signed-off-by: Samuel Martin <s.martin49@gmail.com>

Applied, with one small change, see below.

> ---
>  .../0003-Update-OpenCVCompilerOptions.cmake.patch  | 33 ++++++++++++++++++++++
>  1 file changed, 33 insertions(+)
>  create mode 100644 package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch
> 
> diff --git a/package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch b/package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch
> new file mode 100644
> index 0000000000..290eef4d10
> --- /dev/null
> +++ b/package/opencv/0003-Update-OpenCVCompilerOptions.cmake.patch
> @@ -0,0 +1,33 @@
> +Fetch from upstream: https://github.com/opencv/opencv/commit/126de0cd95418811e302996161e03195bf7e631a

By putting this comment here, you break the possibility of applying the
patch with "git am", i.e the patch is no longer a proper Git formatted
patch. Therefore, such comments should go...

> +
> +From 126de0cd95418811e302996161e03195bf7e631a Mon Sep 17 00:00:00 2001
> +From: neok-m4700 <neok-m4700@users.noreply.github.com>
> +Date: Tue, 25 Jul 2017 10:25:20 +0200
> +Subject: [PATCH] Update OpenCVCompilerOptions.cmake
> +
> +misplaced else

... here.

> +
> +Signed-off-by: Samuel Martin <s.martin49@gmail.com>

Applied with this change. Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 3/4] package/opencv: add hashes for patch and license files

[Samuel Martin]

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
---
 package/opencv/opencv.hash | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/package/opencv/opencv.hash b/package/opencv/opencv.hash
index 85a3e0b8f4..31bbaf6777 100644
--- a/package/opencv/opencv.hash
+++ b/package/opencv/opencv.hash
@@ -1,2 +1,6 @@
 # Locally calculated
 sha256 fb4769d0119c35426c3754b7fb079b407911e863958db53bdec83c7794582e41  opencv-2.4.13.3.tar.gz
+sha256 934825dcbd68ec46936d5b12f4734d07fffc864226a9122e46a23c55d0e5bb5c  0001-atomic.patch
+sha256 f211b79c126021de2f553e894bb5d0896be542289e8f55f5289eba45c86f5d6e  0002-CMakeLists.txt-Do-not-add-libdl-to-LINKER_LIBS-for-s.patch
+sha256 78622658634e0297588037e26e54ee37754f739e30336184921f42f4888cda7f  0003-Update-OpenCVCompilerOptions.cmake.patch
+sha256 46e42877dfc3ac65769292cb4dc1a04b204ad9e40859a390f4c035d6179fbcc4  LICENSE
-- 
2.15.0

[Buildroot] [PATCH 3/4] package/opencv: add hashes for patch and license files

[Thomas Petazzoni]

Hello,

On Wed,  1 Nov 2017 21:11:36 +0100, Samuel Martin wrote:
> Signed-off-by: Samuel Martin <s.martin49@gmail.com>
> ---
>  package/opencv/opencv.hash | 4 ++++
>  1 file changed, 4 insertions(+)
> 
> diff --git a/package/opencv/opencv.hash b/package/opencv/opencv.hash
> index 85a3e0b8f4..31bbaf6777 100644
> --- a/package/opencv/opencv.hash
> +++ b/package/opencv/opencv.hash
> @@ -1,2 +1,6 @@
>  # Locally calculated
>  sha256 fb4769d0119c35426c3754b7fb079b407911e863958db53bdec83c7794582e41  opencv-2.4.13.3.tar.gz
> +sha256 934825dcbd68ec46936d5b12f4734d07fffc864226a9122e46a23c55d0e5bb5c  0001-atomic.patch
> +sha256 f211b79c126021de2f553e894bb5d0896be542289e8f55f5289eba45c86f5d6e  0002-CMakeLists.txt-Do-not-add-libdl-to-LINKER_LIBS-for-s.patch
> +sha256 78622658634e0297588037e26e54ee37754f739e30336184921f42f4888cda7f  0003-Update-OpenCVCompilerOptions.cmake.patch
> +sha256 46e42877dfc3ac65769292cb4dc1a04b204ad9e40859a390f4c035d6179fbcc4  LICENSE

Since when are we adding hashes for patches that are stored in
Buildroot itself? We are not doing this for any other package, and the
package infrastructure is not even checking the hashes for patches
available locally I believe.

So I've tweaked your patch to only add the license file hash, and
applied.

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 4/4] package/ussp-push: removed deprecated/no longer needed deps on openobex-bluez

[Samuel Martin]

Commit 49a9fb0f13789298e65f21405502a9106e2c7382 deprecates the
BR2_PACKAGE_OPENOBEX_BLUEZ option, which should be turned in
BR2_PACKAGE_BLUEZ_UTILS instead.

Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Reported-by: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Samuel Martin <s.martin49@gmail.com>
---
 package/ussp-push/Config.in | 1 -
 1 file changed, 1 deletion(-)

diff --git a/package/ussp-push/Config.in b/package/ussp-push/Config.in
index e6695bb197..aefec85a20 100644
--- a/package/ussp-push/Config.in
+++ b/package/ussp-push/Config.in
@@ -5,7 +5,6 @@ config BR2_PACKAGE_USSP_PUSH
 	depends on BR2_TOOLCHAIN_HAS_THREADS # dbus, alsa-lib, libglib2
 	depends on BR2_USE_MMU # dbus
 	select BR2_PACKAGE_OPENOBEX
-	select BR2_PACKAGE_OPENOBEX_BLUEZ
 	select BR2_PACKAGE_BLUEZ_UTILS
 	help
 	  ussp-push is an OBEX object pusher for Linux, using the BlueZ
-- 
2.15.0

[Buildroot] [PATCH 4/4] package/ussp-push: removed deprecated/no longer needed deps on openobex-bluez

[Thomas Petazzoni]

Hello,

On Wed,  1 Nov 2017 21:11:37 +0100, Samuel Martin wrote:
> Commit 49a9fb0f13789298e65f21405502a9106e2c7382 deprecates the
> BR2_PACKAGE_OPENOBEX_BLUEZ option, which should be turned in
> BR2_PACKAGE_BLUEZ_UTILS instead.

Applied, thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com

[Buildroot] [PATCH 4/4] package/ussp-push: removed deprecated/no longer needed deps on openobex-bluez

[Thomas Petazzoni]

Hello,

On Wed,  1 Nov 2017 21:11:37 +0100, Samuel Martin wrote:
> Commit 49a9fb0f13789298e65f21405502a9106e2c7382 deprecates the
> BR2_PACKAGE_OPENOBEX_BLUEZ option, which should be turned in
> BR2_PACKAGE_BLUEZ_UTILS instead.
> 
> Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
> Reported-by: Romain Naour <romain.naour@gmail.com>
> Signed-off-by: Samuel Martin <s.martin49@gmail.com>

I clearly don't see how this could be related to your patch, but we've
started to see ussp-push build failures:

  http://autobuild.buildroot.net/?reason=ussp-push-0.11

Could you have a look? :-)

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Thomas Petazzoni]

Hello,

On Wed,  1 Nov 2017 21:11:34 +0100, Samuel Martin wrote:
> Openobex uses accept4 syscall, which is not available on Blackfin flavored uclibc.
> So, diasble openobex on blackfin (do not add extra checks on the c-library since
> uclibc is the only choice for Blackfin).
> 
> Fixed:
>   http://autobuild.buildroot.net/results/78e033fe9f43845581a5d87b21a8451f67520e44
> 
> Signed-off-by: Samuel Martin <s.martin49@gmail.com>
> ---
>  package/openobex/Config.in | 1 +
>  1 file changed, 1 insertion(+)
> 
> diff --git a/package/openobex/Config.in b/package/openobex/Config.in
> index e611b8d803..4a4dc214ef 100644
> --- a/package/openobex/Config.in
> +++ b/package/openobex/Config.in
> @@ -1,5 +1,6 @@
>  config BR2_PACKAGE_OPENOBEX
>  	bool "openobex"
> +	depends on !BR2_bfin
>  	help
>  	  Free open source implementation of the Object Exchange (OBEX)
>  	  protocol.

What about the Config.in comment, and the reverse dependencies of
BR2_PACKAGE_OPENOBEX ?

Also, I'd like to hear from Waldemar about accept4 availability on
Blackfin/uClibc-ng.

Thanks!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Waldemar Brodkorb]

Hi Thomas,
Thomas Petazzoni wrote,

> Hello,
> 
> On Wed,  1 Nov 2017 21:11:34 +0100, Samuel Martin wrote:
> > Openobex uses accept4 syscall, which is not available on Blackfin flavored uclibc.
> > So, diasble openobex on blackfin (do not add extra checks on the c-library since
> > uclibc is the only choice for Blackfin).
> > 
> > Fixed:
> >   http://autobuild.buildroot.net/results/78e033fe9f43845581a5d87b21a8451f67520e44
> > 
> > Signed-off-by: Samuel Martin <s.martin49@gmail.com>
> > ---
> >  package/openobex/Config.in | 1 +
> >  1 file changed, 1 insertion(+)
> > 
> > diff --git a/package/openobex/Config.in b/package/openobex/Config.in
> > index e611b8d803..4a4dc214ef 100644
> > --- a/package/openobex/Config.in
> > +++ b/package/openobex/Config.in
> > @@ -1,5 +1,6 @@
> >  config BR2_PACKAGE_OPENOBEX
> >  	bool "openobex"
> > +	depends on !BR2_bfin
> >  	help
> >  	  Free open source implementation of the Object Exchange (OBEX)
> >  	  protocol.
> 
> What about the Config.in comment, and the reverse dependencies of
> BR2_PACKAGE_OPENOBEX ?
> 
> Also, I'd like to hear from Waldemar about accept4 availability on
> Blackfin/uClibc-ng.

For me it looks like a uClibc bug.
When socketcalls.c was converted to use cancel.h for syscall
cancellation (19607f1113ef9916a0a0ac2bf99b5bc32526f0de)
it seems accept4 was left out for no specific reason.

For Linuxthreads the syscall wrapper for accept4 isn't compiled in.
I think for NPTL we do not see the issue.

I am going to sent out a patch soon.

best regards
 Waldemar

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Thomas Petazzoni]

Hello,

On Thu, 2 Nov 2017 03:14:34 +0100, Waldemar Brodkorb wrote:

> > Also, I'd like to hear from Waldemar about accept4 availability on
> > Blackfin/uClibc-ng.  
> 
> For me it looks like a uClibc bug.
> When socketcalls.c was converted to use cancel.h for syscall
> cancellation (19607f1113ef9916a0a0ac2bf99b5bc32526f0de)
> it seems accept4 was left out for no specific reason.
> 
> For Linuxthreads the syscall wrapper for accept4 isn't compiled in.
> I think for NPTL we do not see the issue.
> 
> I am going to sent out a patch soon.

Great, thanks a lot!

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Samuel Martin]

Hi all,

On Wed, Nov 1, 2017 at 11:23 PM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> Hello,
>
> On Wed,  1 Nov 2017 21:11:34 +0100, Samuel Martin wrote:
>> Openobex uses accept4 syscall, which is not available on Blackfin flavored uclibc.
>> So, diasble openobex on blackfin (do not add extra checks on the c-library since
>> uclibc is the only choice for Blackfin).
>>
>> Fixed:
>>   http://autobuild.buildroot.net/results/78e033fe9f43845581a5d87b21a8451f67520e44
>>
>> Signed-off-by: Samuel Martin <s.martin49@gmail.com>
>> ---
>>  package/openobex/Config.in | 1 +
>>  1 file changed, 1 insertion(+)
>>
>> diff --git a/package/openobex/Config.in b/package/openobex/Config.in
>> index e611b8d803..4a4dc214ef 100644
>> --- a/package/openobex/Config.in
>> +++ b/package/openobex/Config.in
>> @@ -1,5 +1,6 @@
>>  config BR2_PACKAGE_OPENOBEX
>>       bool "openobex"
>> +     depends on !BR2_bfin
>>       help
>>         Free open source implementation of the Object Exchange (OBEX)
>>         protocol.
>
> What about the Config.in comment, and the reverse dependencies of
> BR2_PACKAGE_OPENOBEX ?

Well, I was not really sure about what to mention... and also wanted
to heard Waldemar opinion.
FYI, this commit 53996bee433f09a91b12aa53b2be0f7d22c0acbe seems to fix
a similar issue.

>
> Also, I'd like to hear from Waldemar about accept4 availability on
> Blackfin/uClibc-ng.
>
> Thanks!
>
> Thomas
> --
> Thomas Petazzoni, CTO, Free Electrons
> Embedded Linux and Kernel engineering
> http://free-electrons.com


Regards,

-- 
Samuel

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Thomas Petazzoni]

Hello,

On Thu, 2 Nov 2017 09:01:28 +0100, Samuel Martin wrote:

> >> diff --git a/package/openobex/Config.in b/package/openobex/Config.in
> >> index e611b8d803..4a4dc214ef 100644
> >> --- a/package/openobex/Config.in
> >> +++ b/package/openobex/Config.in
> >> @@ -1,5 +1,6 @@
> >>  config BR2_PACKAGE_OPENOBEX
> >>       bool "openobex"
> >> +     depends on !BR2_bfin
> >>       help
> >>         Free open source implementation of the Object Exchange (OBEX)
> >>         protocol.  
> >
> > What about the Config.in comment, and the reverse dependencies of
> > BR2_PACKAGE_OPENOBEX ?  
> 
> Well, I was not really sure about what to mention... and also wanted
> to heard Waldemar opinion.

I don't understand what you mean here. If you're adding a new
dependency to a package, you must:

 - Propagate it to the Config.in comment of that package

 - Propagate it to the reverse dependencies of that package

So I'm not sure how you can be "not really sure about what to mention".

You must add a "depends on !BR2_bfin" in the Config.in comment, and
then a depends on !BR2_bfin to all reverse dependencies of openobex,
including their Config.in comments. There is nothing to "mention".

But since Waldemar has submitted a patch to fix the actual problem in
uClibc, this is all moot :)

> FYI, this commit 53996bee433f09a91b12aa53b2be0f7d22c0acbe seems to fix
> a similar issue.

We should revert this commit once Waldemar fix is merged, and we have
rebuilt the toolchains.

Best regards,

Thomas
-- 
Thomas Petazzoni, CTO, Free Electrons
Embedded Linux and Kernel engineering
http://free-electrons.com

[Buildroot] [PATCH 1/4] package/openobex: disable on Blackfin

[Samuel Martin]

On Thu, Nov 2, 2017 at 11:38 AM, Thomas Petazzoni
<thomas.petazzoni@free-electrons.com> wrote:
> Hello,
>
> On Thu, 2 Nov 2017 09:01:28 +0100, Samuel Martin wrote:
>
>> >> diff --git a/package/openobex/Config.in b/package/openobex/Config.in
>> >> index e611b8d803..4a4dc214ef 100644
>> >> --- a/package/openobex/Config.in
>> >> +++ b/package/openobex/Config.in
>> >> @@ -1,5 +1,6 @@
>> >>  config BR2_PACKAGE_OPENOBEX
>> >>       bool "openobex"
>> >> +     depends on !BR2_bfin
>> >>       help
>> >>         Free open source implementation of the Object Exchange (OBEX)
>> >>         protocol.
>> >
>> > What about the Config.in comment, and the reverse dependencies of
>> > BR2_PACKAGE_OPENOBEX ?
>>
>> Well, I was not really sure about what to mention... and also wanted
>> to heard Waldemar opinion.
>
> I don't understand what you mean here. If you're adding a new
> dependency to a package, you must:
>
>  - Propagate it to the Config.in comment of that package
>
>  - Propagate it to the reverse dependencies of that package
>
> So I'm not sure how you can be "not really sure about what to mention".
>
> You must add a "depends on !BR2_bfin" in the Config.in comment, and
> then a depends on !BR2_bfin to all reverse dependencies of openobex,
> including their Config.in comments. There is nothing to "mention".
>
> But since Waldemar has submitted a patch to fix the actual problem in
> uClibc, this is all moot :)
>
>> FYI, this commit 53996bee433f09a91b12aa53b2be0f7d22c0acbe seems to fix
>> a similar issue.
>
> We should revert this commit once Waldemar fix is merged, and we have
> rebuilt the toolchains.

That's the plan.

Regards,

>
> Best regards,
>
> Thomas
> --
> Thomas Petazzoni, CTO, Free Electrons
> Embedded Linux and Kernel engineering
> http://free-electrons.com



-- 
Samuel