[PATCH 1/6] systemd-boot: bump version to 246.2

[PATCH 1/6] systemd-boot: bump version to 246.2

[Alexander Kanavin]

The recipe includes systemd.inc and so shares SRCREV with the main systemd.

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
---
 .../systemd/{systemd-boot_246.1.bb => systemd-boot_246.2.bb}      | 0
 1 file changed, 0 insertions(+), 0 deletions(-)
 rename meta/recipes-core/systemd/{systemd-boot_246.1.bb => systemd-boot_246.2.bb} (100%)

diff --git a/meta/recipes-core/systemd/systemd-boot_246.1.bb b/meta/recipes-core/systemd/systemd-boot_246.2.bb
similarity index 100%
rename from meta/recipes-core/systemd/systemd-boot_246.1.bb
rename to meta/recipes-core/systemd/systemd-boot_246.2.bb
-- 
2.28.0

[PATCH 2/6] json-glib: inherit upstream-version-is-even

[Alexander Kanavin]

As specified in https://wiki.gnome.org/Projects/JsonGlib

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
---
 meta/recipes-gnome/json-glib/json-glib_1.4.4.bb | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta/recipes-gnome/json-glib/json-glib_1.4.4.bb b/meta/recipes-gnome/json-glib/json-glib_1.4.4.bb
index 5143d73ed7..c999d761bd 100644
--- a/meta/recipes-gnome/json-glib/json-glib_1.4.4.bb
+++ b/meta/recipes-gnome/json-glib/json-glib_1.4.4.bb
@@ -12,7 +12,7 @@ LIC_FILES_CHKSUM = "file://COPYING;md5=7fbc338309ac38fefcd64b04bb903e34"
 DEPENDS = "glib-2.0"
 
 GNOMEBASEBUILDCLASS = "meson"
-inherit gnomebase lib_package gobject-introspection gtk-doc gettext ptest-gnome manpages
+inherit gnomebase lib_package gobject-introspection gtk-doc gettext ptest-gnome manpages upstream-version-is-even
 
 SRC_URI += "file://run-ptest"
 SRC_URI[archive.md5sum] = "4d4bb9837f6d31e32d0ce658ae135f68"
-- 
2.28.0

[PATCH 3/6] packagegroup-core-device-devel: remove

[Alexander Kanavin]

The purpose and content of the group is not clear;
the only consumer (poky distro) is no longer using it.

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
---
 .../packagegroup-core-device-devel.bb            | 16 ----------------
 1 file changed, 16 deletions(-)
 delete mode 100644 meta/recipes-devtools/packagegroups/packagegroup-core-device-devel.bb

diff --git a/meta/recipes-devtools/packagegroups/packagegroup-core-device-devel.bb b/meta/recipes-devtools/packagegroups/packagegroup-core-device-devel.bb
deleted file mode 100644
index edee474eca..0000000000
--- a/meta/recipes-devtools/packagegroups/packagegroup-core-device-devel.bb
+++ /dev/null
@@ -1,16 +0,0 @@
-SUMMARY = "Provides a small set of tools for development on the device"
-
-PR = "r1"
-
-inherit packagegroup
-
-RPROVIDES_${PN} = "qemu-config"
-RREPLACES_${PN} = "qemu-config"
-RCONFLICTS_${PN} = "qemu-config"
-
-RDEPENDS_${PN} = "\
-    distcc-config \
-    nfs-export-root \
-    bash \
-    binutils-symlinks \
-    "
-- 
2.28.0

[PATCH 4/6] oeqa/x32lib: rework to use readelf from the host

[Alexander Kanavin]

It is difficult and error-prone to ensure binutils gets
installed into target images where this test may run;
on the other hand readelf is always present on the
host, as it is a part of HOSTTOOLS.

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
---
 meta/lib/oeqa/runtime/cases/x32lib.py | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/meta/lib/oeqa/runtime/cases/x32lib.py b/meta/lib/oeqa/runtime/cases/x32lib.py
index ddf220140e..f419c8f181 100644
--- a/meta/lib/oeqa/runtime/cases/x32lib.py
+++ b/meta/lib/oeqa/runtime/cases/x32lib.py
@@ -6,16 +6,21 @@ from oeqa.runtime.case import OERuntimeTestCase
 from oeqa.core.decorator.depends import OETestDepends
 from oeqa.core.decorator.data import skipIfNotInDataVar
 
+import subprocess
+
 class X32libTest(OERuntimeTestCase):
 
     @skipIfNotInDataVar('DEFAULTTUNE', 'x86-64-x32',
                         'DEFAULTTUNE is not set to x86-64-x32')
     @OETestDepends(['ssh.SSHTest.test_ssh'])
     def test_x32_file(self):
-        cmd = 'readelf -h /bin/ls | grep Class | grep ELF32'
-        status1 = self.target.run(cmd)[0]
-        cmd = 'readelf -h /bin/ls | grep Machine | grep X86-64'
-        status2 = self.target.run(cmd)[0]
-        msg = ("/bin/ls isn't an X86-64 ELF32 binary. readelf says: %s" % 
-                self.target.run("readelf -h /bin/ls")[1])
+        dest = self.td.get('T', '') + "/ls.x32test"
+        self.target.copyFrom("/bin/ls", dest)
+        cmd = 'readelf -h {} | grep Class | grep ELF32'.format(dest)
+        status1 = subprocess.call(cmd, shell=True)
+        cmd = 'readelf -h {} | grep Machine | grep X86-64'.format(dest)
+        status2 = subprocess.call(cmd, shell=True)
+        msg = ("/bin/ls isn't an X86-64 ELF32 binary. readelf says:\n{}".format(
+                subprocess.check_output("readelf -h {}".format(dest), shell=True).decode()))
+        os.remove(dest)
         self.assertTrue(status1 == 0 and status2 == 0, msg=msg)
-- 
2.28.0

[PATCH 5/6] oeqa/multilib: rework to use readelf from the host

[Alexander Kanavin]

It is difficult and error-prone to ensure binutils gets
installed into target images where this test may run;
on the other hand readelf is always present on the
host, as it is a part of HOSTTOOLS.

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
---
 meta/lib/oeqa/runtime/cases/multilib.py | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

diff --git a/meta/lib/oeqa/runtime/cases/multilib.py b/meta/lib/oeqa/runtime/cases/multilib.py
index 62e662b01c..56aed62eea 100644
--- a/meta/lib/oeqa/runtime/cases/multilib.py
+++ b/meta/lib/oeqa/runtime/cases/multilib.py
@@ -7,6 +7,8 @@ from oeqa.core.decorator.depends import OETestDepends
 from oeqa.core.decorator.data import skipIfNotInDataVar
 from oeqa.runtime.decorator.package import OEHasPackage
 
+import subprocess
+
 class MultilibTest(OERuntimeTestCase):
 
     def archtest(self, binary, arch):
@@ -14,8 +16,10 @@ class MultilibTest(OERuntimeTestCase):
         Check that ``binary`` has the ELF class ``arch`` (e.g. ELF32/ELF64).
         """
 
-        status, output = self.target.run('readelf -h %s' % binary)
-        self.assertEqual(status, 0, 'Failed to readelf %s' % binary)
+        dest = "{}/test_binary".format(self.td.get('T', ''))
+        self.target.copyFrom(binary, dest)
+        output = subprocess.check_output("readelf -h {}".format(dest), shell=True).decode()
+        os.remove(dest)
 
         l = [l.split()[1] for l in output.split('\n') if "Class:" in l]
         if l:
@@ -29,7 +33,6 @@ class MultilibTest(OERuntimeTestCase):
     @skipIfNotInDataVar('MULTILIBS', 'multilib:lib32',
                         "This isn't a multilib:lib32 image")
     @OETestDepends(['ssh.SSHTest.test_ssh'])
-    @OEHasPackage(['binutils'])
     @OEHasPackage(['lib32-libc6'])
     def test_check_multilib_libc(self):
         """
-- 
2.28.0

[PATCH 6/6] oeqa/multilib: un-skip the connman test

[Alexander Kanavin]

It was previously skipped as it was expecting plain connman
to not be in the image, and for core-image-sato connman is always there.

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
---
 meta/lib/oeqa/runtime/cases/multilib.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/meta/lib/oeqa/runtime/cases/multilib.py b/meta/lib/oeqa/runtime/cases/multilib.py
index 56aed62eea..0d1b9ae2c9 100644
--- a/meta/lib/oeqa/runtime/cases/multilib.py
+++ b/meta/lib/oeqa/runtime/cases/multilib.py
@@ -42,6 +42,6 @@ class MultilibTest(OERuntimeTestCase):
         self.archtest("/lib64/libc.so.6", "ELF64")
 
     @OETestDepends(['multilib.MultilibTest.test_check_multilib_libc'])
-    @OEHasPackage(['lib32-connman', '!connman'])
+    @OEHasPackage(['lib32-connman'])
     def test_file_connman(self):
         self.archtest("/usr/sbin/connmand", "ELF32")
-- 
2.28.0

Re: [OE-core] [PATCH 6/6] oeqa/multilib: un-skip the connman test

[Richard Purdie]

On Thu, 2020-08-27 at 12:59 +0200, Alexander Kanavin wrote:
> It was previously skipped as it was expecting plain connman
> to not be in the image, and for core-image-sato connman is always
> there.
> 
> Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
> ---
>  meta/lib/oeqa/runtime/cases/multilib.py | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/meta/lib/oeqa/runtime/cases/multilib.py
> b/meta/lib/oeqa/runtime/cases/multilib.py
> index 56aed62eea..0d1b9ae2c9 100644
> --- a/meta/lib/oeqa/runtime/cases/multilib.py
> +++ b/meta/lib/oeqa/runtime/cases/multilib.py
> @@ -42,6 +42,6 @@ class MultilibTest(OERuntimeTestCase):
>          self.archtest("/lib64/libc.so.6", "ELF64")
>  
>      @OETestDepends(['multilib.MultilibTest.test_check_multilib_libc'
> ])
> -    @OEHasPackage(['lib32-connman', '!connman'])
> +    @OEHasPackage(['lib32-connman'])
>      def test_file_connman(self):
>          self.archtest("/usr/sbin/connmand", "ELF32")

Hmm. I think this test is only meant to trigger in a lib32-core-image-
sato where lib32-connman would be present and connman would not?

The test name could certainly be clearer as its not a generic connman
test...

Cheers,

Richard

Re: [OE-core] [PATCH 6/6] oeqa/multilib: un-skip the connman test

[Alexander Kanavin]

[-- Attachment #1: Type: text/plain, Size: 698 bytes --]

On Thu, 27 Aug 2020 at 13:06, Richard Purdie <
richard.purdie@linuxfoundation.org> wrote:

> Hmm. I think this test is only meant to trigger in a lib32-core-image-
> sato where lib32-connman would be present and connman would not?
>
> The test name could certainly be clearer as its not a generic connman
> test...
>

There is no lib32-core-image-sato anywhere in config.json  - and I think
RPM_PREFER_ELF_ARCH in existing tests should ensure that 32 bit connman
package takes precedence.

Also note that those same tests install lib32-connman, which can only be
for the purpose of the test:
http://git.yoctoproject.org/cgit/cgit.cgi/yocto-autobuilder-helper/tree/config.json#n495

[-- Attachment #2: Type: text/html, Size: 1161 bytes --]

Re: [OE-core] [PATCH 6/6] oeqa/multilib: un-skip the connman test

[Richard Purdie]

On Thu, 2020-08-27 at 13:25 +0200, Alexander Kanavin wrote:
> On Thu, 27 Aug 2020 at 13:06, Richard Purdie <richard.purdie@linuxfoundation.org> wrote:
> > Hmm. I think this test is only meant to trigger in a lib32-core-image-
> > sato where lib32-connman would be present and connman would not?
> > 
> > The test name could certainly be clearer as its not a generic connman
> > test...
> 
> There is no lib32-core-image-sato anywhere in config.json  - and I think 
> RPM_PREFER_ELF_ARCH in existing tests should ensure that 32 bit connman package takes precedence.
> 
> Also note that those same tests install lib32-connman, which can only be for the purpose of the test:
> http://git.yoctoproject.org/cgit/cgit.cgi/yocto-autobuilder-helper/tree/config.json#n495

Right, those are the tests I was thinking this test is related to. It
looks like its only those tests that enable this TEST_SUITE so your
patch is probably correct. Its a shame this test suite is so tied to
the configuration in ab-helper though. Not sure how we could improve
that...

Cheers,

Richard

Re: [OE-core] [PATCH 4/6] oeqa/x32lib: rework to use readelf from the host

[Khem Raj]

On Thu, Aug 27, 2020 at 4:00 AM Alexander Kanavin
<alex.kanavin@gmail.com> wrote:
>
> It is difficult and error-prone to ensure binutils gets
> installed into target images where this test may run;
> on the other hand readelf is always present on the
> host, as it is a part of HOSTTOOLS.
>

this would mean that we are relying on build host provided binutils to
be compiled with all bfd support
since now we also have aarch64 build hosts it spreads he risk a bit
more. maybe we can
ignore this test if target does not have binutils installed or depend
on binutils-native for providing readelf


> Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
> ---
>  meta/lib/oeqa/runtime/cases/x32lib.py | 17 +++++++++++------
>  1 file changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/meta/lib/oeqa/runtime/cases/x32lib.py b/meta/lib/oeqa/runtime/cases/x32lib.py
> index ddf220140e..f419c8f181 100644
> --- a/meta/lib/oeqa/runtime/cases/x32lib.py
> +++ b/meta/lib/oeqa/runtime/cases/x32lib.py
> @@ -6,16 +6,21 @@ from oeqa.runtime.case import OERuntimeTestCase
>  from oeqa.core.decorator.depends import OETestDepends
>  from oeqa.core.decorator.data import skipIfNotInDataVar
>
> +import subprocess
> +
>  class X32libTest(OERuntimeTestCase):
>
>      @skipIfNotInDataVar('DEFAULTTUNE', 'x86-64-x32',
>                          'DEFAULTTUNE is not set to x86-64-x32')
>      @OETestDepends(['ssh.SSHTest.test_ssh'])
>      def test_x32_file(self):
> -        cmd = 'readelf -h /bin/ls | grep Class | grep ELF32'
> -        status1 = self.target.run(cmd)[0]
> -        cmd = 'readelf -h /bin/ls | grep Machine | grep X86-64'
> -        status2 = self.target.run(cmd)[0]
> -        msg = ("/bin/ls isn't an X86-64 ELF32 binary. readelf says: %s" %
> -                self.target.run("readelf -h /bin/ls")[1])
> +        dest = self.td.get('T', '') + "/ls.x32test"
> +        self.target.copyFrom("/bin/ls", dest)
> +        cmd = 'readelf -h {} | grep Class | grep ELF32'.format(dest)
> +        status1 = subprocess.call(cmd, shell=True)
> +        cmd = 'readelf -h {} | grep Machine | grep X86-64'.format(dest)
> +        status2 = subprocess.call(cmd, shell=True)
> +        msg = ("/bin/ls isn't an X86-64 ELF32 binary. readelf says:\n{}".format(
> +                subprocess.check_output("readelf -h {}".format(dest), shell=True).decode()))
> +        os.remove(dest)
>          self.assertTrue(status1 == 0 and status2 == 0, msg=msg)
> --
> 2.28.0
>
> 

Re: [OE-core] [PATCH 4/6] oeqa/x32lib: rework to use readelf from the host

[Alexander Kanavin]

[-- Attachment #1: Type: text/plain, Size: 899 bytes --]

On Thu, 27 Aug 2020 at 23:31, Khem Raj <raj.khem@gmail.com> wrote:

> On Thu, Aug 27, 2020 at 4:00 AM Alexander Kanavin
> <alex.kanavin@gmail.com> wrote:
> >
> > It is difficult and error-prone to ensure binutils gets
> > installed into target images where this test may run;
> > on the other hand readelf is always present on the
> > host, as it is a part of HOSTTOOLS.
> >
>
> this would mean that we are relying on build host provided binutils to
> be compiled with all bfd support
> since now we also have aarch64 build hosts it spreads he risk a bit
> more. maybe we can
> ignore this test if target does not have binutils installed or depend
> on binutils-native for providing readelf
>

The test needs to be explicitly enabled, and requires a very specific build
configuration (see config.json from the AB); it is unlikely someone will
stumble upon it randomly.

Alex

[-- Attachment #2: Type: text/html, Size: 1310 bytes --]

Re: [OE-core] [PATCH 4/6] oeqa/x32lib: rework to use readelf from the host

[Khem Raj]

On Thu, Aug 27, 2020 at 2:36 PM Alexander Kanavin
<alex.kanavin@gmail.com> wrote:
>
> On Thu, 27 Aug 2020 at 23:31, Khem Raj <raj.khem@gmail.com> wrote:
>>
>> On Thu, Aug 27, 2020 at 4:00 AM Alexander Kanavin
>> <alex.kanavin@gmail.com> wrote:
>> >
>> > It is difficult and error-prone to ensure binutils gets
>> > installed into target images where this test may run;
>> > on the other hand readelf is always present on the
>> > host, as it is a part of HOSTTOOLS.
>> >
>>
>> this would mean that we are relying on build host provided binutils to
>> be compiled with all bfd support
>> since now we also have aarch64 build hosts it spreads he risk a bit
>> more. maybe we can
>> ignore this test if target does not have binutils installed or depend
>> on binutils-native for providing readelf
>
>
> The test needs to be explicitly enabled, and requires a very specific build configuration (see config.json from the AB); it is unlikely someone will stumble upon it randomly.
>

that makes it less frequent but the logic problem still remains isn't it.

> Alex