[igt-dev] [PATCH i-g-t v1] meson: Move MSM-specific tests to their own subdirectory

[Jessica Zhang]

On 4/20/2022 1:17 AM, Petri Latvala wrote:
> On Tue, Apr 19, 2022 at 04:43:39PM -0700, Jessica Zhang wrote:
>> Move all custom MSM tests to their own subdirectory and have them
>> installed in an msm/ subdirectory
>> ---
>>   meson.build                    |  9 +++++++++
>>   tests/meson.build              |  4 +---
>>   tests/msm/meson.build          | 15 +++++++++++++++
>>   tests/{ => msm}/msm_mapping.c  |  0
>>   tests/{ => msm}/msm_recovery.c |  0
>>   tests/{ => msm}/msm_submit.c   |  0
>>   6 files changed, 25 insertions(+), 3 deletions(-)
>>   create mode 100644 tests/msm/meson.build
>>   rename tests/{ => msm}/msm_mapping.c (100%)
>>   rename tests/{ => msm}/msm_recovery.c (100%)
>>   rename tests/{ => msm}/msm_submit.c (100%)
>>
>> diff --git a/meson.build b/meson.build
>> index 3e937f5a5b98..a42414c1d22d 100644
>> --- a/meson.build
>> +++ b/meson.build
>> @@ -260,6 +260,7 @@ includedir = get_option('includedir')
>>   libdir = get_option('libdir')
>>   libexecdir = join_paths(get_option('libexecdir'), 'igt-gpu-tools')
>>   amdgpudir = join_paths(libexecdir, 'amdgpu')
>> +msmdir = join_paths(libexecdir, 'msm')
>>   mandir = get_option('mandir')
>>   pkgconfigdir = join_paths(libdir, 'pkgconfig')
>>   python3 = find_program('python3', required : true)
>> @@ -295,10 +296,18 @@ if get_option('use_rpath')
>>   		amdgpudir_rpathdir = join_paths(amdgpudir_rpathdir, '..')
>>   	endforeach
>>   	amdgpudir_rpathdir = join_paths(amdgpudir_rpathdir, libdir)
>> +
>> +	msmdir_rpathdir = '$ORIGIN'
>> +	foreach p : msmdir.split('/')
>> +		msmdir_rpathdir = join_paths(msmdir_rpathdir, '..')
>> +	endforeach
>> +	msmdir_rpathdir = join_paths(msmdir_rpathdir, libdir)
>> +
>>   else
> 
> Extra empty line there.

Noted.

> 
> 
> 
>>   	bindir_rpathdir = ''
>>   	libexecdir_rpathdir = ''
>>   	amdgpudir_rpathdir = ''
>> +	msmdir_rpathdir = ''
>>   endif
>>   
>>   subdir('lib')
>> diff --git a/tests/meson.build b/tests/meson.build
>> index 7261e9aa2950..4ee06677a1eb 100644
>> --- a/tests/meson.build
>> +++ b/tests/meson.build
>> @@ -63,9 +63,6 @@ test_progs = [
>>   	'kms_vrr',
>>   	'kms_writeback',
>>   	'meta_test',
>> -	'msm_mapping',
>> -	'msm_recovery',
>> -	'msm_submit',
>>   	'panfrost_get_param',
>>   	'panfrost_gem_new',
>>   	'panfrost_prime',
>> @@ -420,6 +417,7 @@ test_executables += executable('sw_sync', 'sw_sync.c',
>>   test_list += 'sw_sync'
>>   
>>   subdir('amdgpu')
>> +subdir('msm')
>>   
>>   gen_testlist = find_program('generate_testlist.sh')
>>   test_list_target = custom_target('testlist',
>> diff --git a/tests/msm/meson.build b/tests/msm/meson.build
>> new file mode 100644
>> index 000000000000..9f1788a38d20
>> --- /dev/null
>> +++ b/tests/msm/meson.build
>> @@ -0,0 +1,15 @@
>> +msm_progs = [
>> +	'msm_mapping',
>> +	'msm_recovery',
>> +	'msm_submit'
>> +]
>> +msm_deps = test_deps
>> +
>> +foreach prog : msm_progs
>> +	test_executables += executable(prog, prog + '.c',
>> +				       dependencies : msm_deps,
>> +				       install_dir : msmdir,
>> +				       install_rpath : msmdir_rpathdir,
>> +				       install : true)
>> +	test_list += join_paths('msm', prog)
>> +endforeach
>> diff --git a/tests/msm_mapping.c b/tests/msm/msm_mapping.c
>> similarity index 100%
>> rename from tests/msm_mapping.c
>> rename to tests/msm/msm_mapping.c
>> diff --git a/tests/msm_recovery.c b/tests/msm/msm_recovery.c
>> similarity index 100%
>> rename from tests/msm_recovery.c
>> rename to tests/msm/msm_recovery.c
>> diff --git a/tests/msm_submit.c b/tests/msm/msm_submit.c
>> similarity index 100%
>> rename from tests/msm_submit.c
>> rename to tests/msm/msm_submit.c
>> -- 
>> 2.31.0
>>
> 
> 
> Matches what's done with amdgpu dir,
> Acked-by: Petri Latvala <petri.latvala at intel.com>
> 
> The major change by this is that the piglit-names for the tests become
> igt at msm/msm_*@* instead of igt at msm_*@* and I don't know who's affected
> by that. Tomeu's patch series maybe, the one for integrating msm
> testing to gitlab-CI...

Ah yes, looks like it will conflict with this patch in the series [1]. 
Let me discuss with Tomeu on how best to resolve this.

[1] https://patchwork.freedesktop.org/patch/478042/?series=101195&rev=3

Thanks,

Jessica Zhang

> 
> 
> -- 
> Petri Latvala