[PATCH 0/1] manifest.py: fix test_SDK_manifest_entries

[PATCH 0/1] manifest.py: fix test_SDK_manifest_entries

[Chen Qi]

*** BLURB HERE ***
The following changes since commit 27d60c5a812774f0e5c43161e5b514c4aebdf301:

  gstreamer1.0-libav: disable API documentation (2019-06-15 13:46:38 +0100)

are available in the git repository at:

  git://git.pokylinux.org/poky-contrib ChenQi/oeqa-manifest
  http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=ChenQi/oeqa-manifest

Chen Qi (1):
  manifest.py: fix test_SDK_manifest_entries

 meta/lib/oeqa/selftest/cases/manifest.py | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

-- 
1.9.1

[PATCH 1/1] manifest.py: fix test_SDK_manifest_entries

[Chen Qi]

TOOLCHAIN_OUTPUTNAME could be overridden. So use this variable directly
instead of its default value ${SDK_NAME}-toolchain-${SDK_VERSION}.

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
---
 meta/lib/oeqa/selftest/cases/manifest.py | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)

diff --git a/meta/lib/oeqa/selftest/cases/manifest.py b/meta/lib/oeqa/selftest/cases/manifest.py
index c0b25ab..5d13f35 100644
--- a/meta/lib/oeqa/selftest/cases/manifest.py
+++ b/meta/lib/oeqa/selftest/cases/manifest.py
@@ -86,11 +86,8 @@ class VerifyManifest(OESelftestTestCase):
         try:
             mdir = self.get_dir_from_bb_var('SDK_DEPLOY', self.buildtarget)
             for k in d_target.keys():
-                bb_vars = get_bb_vars(['SDK_NAME', 'SDK_VERSION'], self.buildtarget)
-                mfilename[k] = "{}-toolchain-{}.{}.manifest".format(
-                        bb_vars['SDK_NAME'],
-                        bb_vars['SDK_VERSION'],
-                        k)
+                toolchain_outputname = get_bb_var('TOOLCHAIN_OUTPUTNAME', self.buildtarget)
+                mfilename[k] = "{}.{}.manifest".format(toolchain_outputname, k)
                 mpath[k] = os.path.join(mdir, mfilename[k])
                 if not os.path.isfile(mpath[k]):
                     self.logger.debug("{}: {} does not exist".format(
-- 
1.9.1

Re: [PATCH 1/1] manifest.py: fix test_SDK_manifest_entries

[Richard Purdie]

On Mon, 2019-06-17 at 17:24 +0800, Chen Qi wrote:
> TOOLCHAIN_OUTPUTNAME could be overridden. So use this variable directly
> instead of its default value ${SDK_NAME}-toolchain-${SDK_VERSION}.
> 
> Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
> ---
>  meta/lib/oeqa/selftest/cases/manifest.py | 7 ++-----
>  1 file changed, 2 insertions(+), 5 deletions(-)
> 
> diff --git a/meta/lib/oeqa/selftest/cases/manifest.py b/meta/lib/oeqa/selftest/cases/manifest.py
> index c0b25ab..5d13f35 100644
> --- a/meta/lib/oeqa/selftest/cases/manifest.py
> +++ b/meta/lib/oeqa/selftest/cases/manifest.py
> @@ -86,11 +86,8 @@ class VerifyManifest(OESelftestTestCase):
>          try:
>              mdir = self.get_dir_from_bb_var('SDK_DEPLOY', self.buildtarget)
>              for k in d_target.keys():
> -                bb_vars = get_bb_vars(['SDK_NAME', 'SDK_VERSION'], self.buildtarget)
> -                mfilename[k] = "{}-toolchain-{}.{}.manifest".format(
> -                        bb_vars['SDK_NAME'],
> -                        bb_vars['SDK_VERSION'],
> -                        k)
> +                toolchain_outputname = get_bb_var('TOOLCHAIN_OUTPUTNAME', self.buildtarget)
> +                mfilename[k] = "{}.{}.manifest".format(toolchain_outputname, k)

Whilst I agree there is a problem here, this code is iterating over
multiple targets. Isn't TOOLCHAIN_OUTPUTNAME only going to be correct
for one of them?

Cheers,

Richard

Re: [PATCH 1/1] manifest.py: fix test_SDK_manifest_entries

[ChenQi]

On 06/17/2019 11:54 PM, Richard Purdie wrote:
> On Mon, 2019-06-17 at 17:24 +0800, Chen Qi wrote:
>> TOOLCHAIN_OUTPUTNAME could be overridden. So use this variable directly
>> instead of its default value ${SDK_NAME}-toolchain-${SDK_VERSION}.
>>
>> Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
>> ---
>>   meta/lib/oeqa/selftest/cases/manifest.py | 7 ++-----
>>   1 file changed, 2 insertions(+), 5 deletions(-)
>>
>> diff --git a/meta/lib/oeqa/selftest/cases/manifest.py b/meta/lib/oeqa/selftest/cases/manifest.py
>> index c0b25ab..5d13f35 100644
>> --- a/meta/lib/oeqa/selftest/cases/manifest.py
>> +++ b/meta/lib/oeqa/selftest/cases/manifest.py
>> @@ -86,11 +86,8 @@ class VerifyManifest(OESelftestTestCase):
>>           try:
>>               mdir = self.get_dir_from_bb_var('SDK_DEPLOY', self.buildtarget)
>>               for k in d_target.keys():
>> -                bb_vars = get_bb_vars(['SDK_NAME', 'SDK_VERSION'], self.buildtarget)
>> -                mfilename[k] = "{}-toolchain-{}.{}.manifest".format(
>> -                        bb_vars['SDK_NAME'],
>> -                        bb_vars['SDK_VERSION'],
>> -                        k)
>> +                toolchain_outputname = get_bb_var('TOOLCHAIN_OUTPUTNAME', self.buildtarget)
>> +                mfilename[k] = "{}.{}.manifest".format(toolchain_outputname, k)
> Whilst I agree there is a problem here, this code is iterating over
> multiple targets. Isn't TOOLCHAIN_OUTPUTNAME only going to be correct
> for one of them?
>
> Cheers,
>
> Richard
>
>

This code is iterating over d_target.keys(), which is 'target' and 
'host'. And the TOOLCHAIN_OUTPUTNAME is for self.buildtarget, which is 
core-image-minimal. Thus, we are basically checking 
${TOOLCHAIN_OUTPUTNAME}.target.manifest and 
${TOOLCHAIN_OUTPUTNAME}.host.target.
e.g.
tmp-glibc/deploy/sdk/poky-glibc-x86_64-core-image-minimal-core2-64-qemux86-64-toolchain-2.7+snapshot.host.manifest
tmp-glibc/deploy/sdk/poky-glibc-x86_64-core-image-minimal-core2-64-qemux86-64-toolchain-2.7+snapshot.target.manifest

So I think things are correct.

Best Regards,
Chen Qi

Re: [PATCH 1/1] manifest.py: fix test_SDK_manifest_entries

[richard.purdie]

On Tue, 2019-06-18 at 10:09 +0800, ChenQi wrote:
> On 06/17/2019 11:54 PM, Richard Purdie wrote:
> > On Mon, 2019-06-17 at 17:24 +0800, Chen Qi wrote:
> > > TOOLCHAIN_OUTPUTNAME could be overridden. So use this variable
> > > directly
> > > instead of its default value ${SDK_NAME}-toolchain-
> > > ${SDK_VERSION}.
> > > 
> > > Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
> > > ---
> > >   meta/lib/oeqa/selftest/cases/manifest.py | 7 ++-----
> > >   1 file changed, 2 insertions(+), 5 deletions(-)
> > > 
> > > diff --git a/meta/lib/oeqa/selftest/cases/manifest.py
> > > b/meta/lib/oeqa/selftest/cases/manifest.py
> > > index c0b25ab..5d13f35 100644
> > > --- a/meta/lib/oeqa/selftest/cases/manifest.py
> > > +++ b/meta/lib/oeqa/selftest/cases/manifest.py
> > > @@ -86,11 +86,8 @@ class VerifyManifest(OESelftestTestCase):
> > >           try:
> > >               mdir = self.get_dir_from_bb_var('SDK_DEPLOY',
> > > self.buildtarget)
> > >               for k in d_target.keys():
> > > -                bb_vars = get_bb_vars(['SDK_NAME',
> > > 'SDK_VERSION'], self.buildtarget)
> > > -                mfilename[k] = "{}-toolchain-
> > > {}.{}.manifest".format(
> > > -                        bb_vars['SDK_NAME'],
> > > -                        bb_vars['SDK_VERSION'],
> > > -                        k)
> > > +                toolchain_outputname =
> > > get_bb_var('TOOLCHAIN_OUTPUTNAME', self.buildtarget)
> > > +                mfilename[k] =
> > > "{}.{}.manifest".format(toolchain_outputname, k)
> > Whilst I agree there is a problem here, this code is iterating over
> > multiple targets. Isn't TOOLCHAIN_OUTPUTNAME only going to be
> > correct
> > for one of them?
> > 
> > Cheers,
> > 
> > Richard
> > 
> > 
> 
> This code is iterating over d_target.keys(), which is 'target' and 
> 'host'. And the TOOLCHAIN_OUTPUTNAME is for self.buildtarget, which
> is 
> core-image-minimal. Thus, we are basically checking 
> ${TOOLCHAIN_OUTPUTNAME}.target.manifest and 
> ${TOOLCHAIN_OUTPUTNAME}.host.target.
> e.g.
> tmp-glibc/deploy/sdk/poky-glibc-x86_64-core-image-minimal-core2-64-
> qemux86-64-toolchain-2.7+snapshot.host.manifest
> tmp-glibc/deploy/sdk/poky-glibc-x86_64-core-image-minimal-core2-64-
> qemux86-64-toolchain-2.7+snapshot.target.manifest
> 
> So I think things are correct.

You're right, I was going from memory and thought this code was
handling the case where there are two different targets in an SDK (e.g.
a multilib) but that isn't the case. I've queued the patch for testing.

Cheers,

Richard