[RFC i-g-t] kunit: Remove hard-coded test list from xe_live_ktest

Janusz Krzysztofik janusz.krzysztofik at linux.intel.com
Thu Aug 29 09:57:00 UTC 2024


Hi John,

On Wednesday, 28 August 2024 20:04:06 GMT+2 John Harrison wrote:
> On 8/26/2024 06:07, Janusz Krzysztofik wrote:
> > Hi John, Kamil,
> >
> > On Monday, 26 August 2024 13:46:45 GMT+2 Kamil Konieczny wrote:
> >> Hi John.C.Harrison,
> >> On 2024-08-23 at 11:24:18 -0700, John.C.Harrison at Intel.com wrote:
> >>> From: johnharr <johnharr at invalid-email.com>
> >> ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> >> Again invalid e-mail here.
> >>
> >>> The list of supported kunit tests is currently hard coded.
> > That pattern originates from i915_selftest, where there are 3 stable subtests,
> > "mock", "live" and "perf", each of them executing a (possibly long) list of
> > dynamic sub-subtests actually provided by the module for each category.  Also,
> > IIRC there was a separate module for each Xe test suite before, each mapped to
> > a separate IGT test, later merged into one module with multiple suites and one
> > test with multiple corresponding subtests.
> Not sure if you are just explaining the history of the test or making a 
> suggestion as to how it should evolve next?

Both, I think.  Maybe not the history, but origin of ideas standing behind the 
implementation, and how tests are expected to use it (and maybe evolve if now 
doing that in a different way).

> 
> >
> >>> Which means
> >>> that adding a new test to the kernel also requires patches to IGT as
> >>> well.
> >>>
> >>> The list of available kunit tests is already exported by the kernel.
> >>> So there is no need to bake a list into the IGT source code. So, add
> >>> support for querying the test list dynamically.
> >>>
> >>> NB: Currently, the test list can only be queried by root but the IGT
> >>> build system queries all tests at compile time. In theory this should
> >>> not be a problem. However, the kunit helper code is all written to run
> >>> inside a test and not inside the prep code, which means that when it
> >>> fails to open the root only interfaces, it calls 'skip'. And skip is
> >>> not allowed outside of running a test. Hence the build fails with:
> >>>    skipping is allowed only in fixtures, subtests or igt_simple_main
> >> Looks like we should fix it, move out skips from kunit libs.
> > I suggest you consider a different approach: for a module, call igt_kunit()
> > only once, with NULL suite argument.  As a result, you'll get results from one
> > IGT subtest called "<module_name>" with a bunch of IGT dynamic sub-subtests
> > called "<test_suite>-<test_case>", one for each test case provided by the
> > module.
> I'm not following. This is what my patch does, isn't it? 

No, your patch introduces a runtime determined list of subtests -- something 
not existent in IGT.

An IGT test may consist of one or more statically defined subtests with pre-
defined names.  The term dynamic subtest is usually used in two meanings.  It 
may mean a subtest of type igt_subtest_with_dynamic, still with a pre-defined 
name, but with a runtime determined list of sub-subtests, sometimes called 
dynamic sub-subtests, but often also called just dynamic subtests.  Names of 
dynamic sub-subtests are determined at runtime.

My approach tries to address your need for a maintenance-free kunit IGT test 
source file in a different way.  I'm following the IGT standard of statically 
defined list of subtests with pre-defined names: one subtest of type 
igt_subtest_with_dynamic, named after the kunit test module name which you 
have to enter into the test code anyway, and providing all test cases from 
all test suites contained in that module reported as dynamic sub-subtests 
named <test_suite>-<test_case>.

> There is a 
> single call to igt_kunit_get_test_names() which returns a list of 
> 'suite/case' name pairs having already registered those as dynamic 
> sub-tests. Then it loops through the list running the suites (with any 
> command line filtering being automatically applied to filter out sub-tests).

That's what breaks the IGT rule of statically defined lists of subtests with 
pre-defined names.

> 
> 
> >
> >>> And of course, putting all the query code inside an igt_fixture block
> >>> means it doesn't get run as part of --show-testlist or --list-subtests.
> >>>
> >>> Reworking the kunit code to fail cleanly rather than skipping
> >>> everywhere seems much more invasive. It is also not clear why the
> >>> kunit code is considered 'library' when it is execlusively used from
> >>> the kunit test alone!?
> > Current users:
> > 	igt$ grep -rl igt_kunit tests/
> > 	tests/intel/xe_live_ktest.c
> > 	tests/kms_selftest.c
> > 	tests/drm_mm.c
> > 	tests/drm_buddy.c
> >
> > The other 3 were first converted from i915_selftest standard to KUnit still
> > before Xe KUnit IGT tests started appearing.
> 
> Argh, I must have been inside the 'intel' directory when I did my 
> search! Oops.
> 
> The KMS one has another hard coded test list. 

But that's not a list of test suites, that's has a hardcoded list of module 
names -- something not avoidable when kunit test cases you want to group in a 
single IGT test are split among several modules.  IIRC, each of those modules 
provides only one test suite.  Names of test suites may be the same as module 
names (with _kunit or _test suffix stripped) or may be different.  It was 
decided to use module names as subtest names, then we are always passing NULL 
as suite.  If module and suite names don't match then IGT dynamic sub-subtests 
are called <test_suite>-<test_case>, otherwise just <test_case>.

> The drm_mm one has a commented hard coded test list 

I don't like the idea of documenting dynamic sub-subtest names in IGT test 
sources -- for me that breaks current IGT standards in the opposite direction.  
But maybe an important business need stands behind.

> but actually just runs everything with no compiled in lists and the
> drm_buddy test has no test lists at all. 

Both have a module name hardcoded, also used as a name of IGT subtest.

> Not 
> sure what tests/sub-tests they actually run because all three of those 
> just skip for me. Not sure why. 

Have you got those drm kunit modules built and available to the running 
kernel?

> The xe_live_ktest one runs everything 
> fine so I definitely have the ktest system compiled in to the kernel. 
> And this is running with a stock IGT tree, not with my patch, so it's 
> not that my patch is breaking it.

Your patch can't break anything since it doesn't modify any existing 
functions, only introduces a new one.  As long as you don't force IGT tests to 
use it, nothing can break.

Thanks,
Janusz

> 
> John.
> 
> >
> > Thanks,
> > Janusz
> >
> >>> So posting this as an RFC before going too far
> >>> down the possible rabbit hole.
> >>>
> >> As I see this, it is a lib because some other vendor (like amdpgu)
> >> could use it.
> >>
> >> +cc Janusz
> >> Cc: Janusz Krzysztofik <janusz.krzysztofik at linux.intel.com>
> >>
> >> btw could you split this into two patches, one for lib/* and
> >> one for tests/intel/xe_live_ktest ?
> >>
> >>> Signed-off-by: John Harrison <John.C.Harrison at Intel.com>
> >>> ---
> >>>   lib/igt_kmod.c              | 68 +++++++++++++++++++++++++++++++++++++
> >>>   lib/igt_kmod.h              |  6 ++++
> >>>   tests/intel/xe_live_ktest.c | 60 ++++++++++++++++----------------
> >>>   3 files changed, 103 insertions(+), 31 deletions(-)
> >>>
> >>> diff --git a/lib/igt_kmod.c b/lib/igt_kmod.c
> >>> index 464c0dcf484a..69c1254b397c 100644
> >>> --- a/lib/igt_kmod.c
> >>> +++ b/lib/igt_kmod.c
> >>> @@ -1414,6 +1414,74 @@ static void __igt_kunit(struct igt_ktest *tst,
> >>>   	}
> >>>   }
> >>>   
> >>> +/**
> >>> + * igt_kunit:
> >>> + * @module_name: the name of the module
> >>> + * @suite: the name of test suite to be executed, also used as subtest name;
> >>> + *	   if NULL then test cases from all test suites provided by the module
> >>> + *	   are executed as dynamic sub-subtests of one IGT subtest, which name
> >>> + *	   is derived from the module name by cutting off its optional trailing
> >>> + *	   _test or _kunit suffix
> >>> + * @opts: options to load the module
> >>> + *
> >>> + * Loads the test module, parses its (k)tap dmesg output, then unloads it
> >>> + */
> >>> +void igt_kunit_get_test_names(const char *module_name, const char *opts, struct igt_list_head *names)
> >>> +{
> >>> +	char debugfs_path[PATH_MAX] = { '\0', };
> >>> +	struct igt_ktest tst = { .kmsg = -1, };
> >>> +	struct igt_ktap_results *ktap = NULL;
> >>> +	const char *subtest;
> >>> +	char *suite_name = NULL, *case_name = NULL;
> >>> +	DIR *debugfs_dir = NULL;
> >>> +	IGT_LIST_HEAD(tests);
> >>> +
> >>> +	subtest = strdup(module_name);
> >>> +	if (!igt_debug_on(!subtest)) {
> >>> +		char *suffix = strstr(subtest, "_test");
> >>> +
> >>> +		if (!suffix)
> >>> +			suffix = strstr(subtest, "_kunit");
> >>> +
> >>> +		if (suffix)
> >>> +			*suffix = '\0';
> >>> +	}
> >>> +
> >>> +	igt_require(subtest);
> >>> +
> >>> +	igt_require(!igt_ktest_init(&tst, module_name));
> >>> +	igt_require(!igt_ktest_begin(&tst));
> >>> +
> >>> +	igt_ignore_warn(igt_kmod_load("kunit", NULL));
> >>> +
> >>> +	kunit_debugfs_path(debugfs_path);
> >>> +	igt_info("kunit path: %s\n", debugfs_path);
> >>> +	if (igt_debug_on(!*debugfs_path) ||
> >>> +	    !kunit_get_tests(&tests, &tst, NULL, opts, debugfs_path, &debugfs_dir, &ktap)) {
> >>> +		igt_info("kunit - no tests found!?\n");
> >>> +	} else {
> >>> +		struct igt_ktap_result *t;
> >>> +		igt_info("kunit tests:\n");
> >>> +		igt_list_for_each_entry(t, &tests, link) {
> >>> +			struct igt_kunit_names *name = malloc(sizeof(*name));
> >>> +			name->suite = strdup(t->suite_name);
> >>> +			name->sub = strdup(t->case_name);
> >>> +			igt_info("  %s / %s\n", t->suite_name, t->case_name);
> >>> +			igt_list_add(&name->link, names);
> >>> +		}
> >>> +	}
> >>> +
> >>> +	igt_ktap_free(&ktap);
> >>> +
> >>> +	kunit_results_free(&tests, &suite_name, &case_name);
> >>> +
> >>> +	if (debugfs_dir)
> >>> +		closedir(debugfs_dir);
> >>> +
> >>> +	igt_ktest_end(&tst);
> >>> +	igt_ktest_fini(&tst);
> >>> +}
> >>> +
> >>>   /**
> >>>    * igt_kunit:
> >>>    * @module_name: the name of the module
> >>> diff --git a/lib/igt_kmod.h b/lib/igt_kmod.h
> >>> index efb46da128d4..2f608eb69d48 100644
> >>> --- a/lib/igt_kmod.h
> >>> +++ b/lib/igt_kmod.h
> >>> @@ -71,7 +71,13 @@ static inline int igt_xe_driver_unload(void)
> >>>   int igt_amdgpu_driver_load(const char *opts);
> >>>   int igt_amdgpu_driver_unload(void);
> >>>   
> >>> +struct igt_kunit_names {
> >>> +	char *suite;
> >>> +	char *sub;
> >>> +	struct igt_list_head link;
> >>> +};
> >>>   void igt_kunit(const char *module_name, const char *name, const char *opts);
> >>> +void igt_kunit_get_test_names(const char *module_name, const char *opts, struct igt_list_head *names);
> >>>   
> >>>   void igt_kselftests(const char *module_name,
> >>>   		    const char *module_options,
> >>> diff --git a/tests/intel/xe_live_ktest.c b/tests/intel/xe_live_ktest.c
> >>> index 4376d5df7751..13265e87ba9d 100644
> >>> --- a/tests/intel/xe_live_ktest.c
> >>> +++ b/tests/intel/xe_live_ktest.c
> >>> @@ -9,40 +9,38 @@
> >>>    * Sub-category: kunit
> >>>    * Functionality: kunit test
> >>>    * Test category: functionality test
> >>> - *
> >>> - * SUBTEST: xe_bo
> >>> - * Description:
> >>> - *	Kernel dynamic selftests to check if GPU buffer objects are
> >>> - *	being handled properly.
> >>> - * Functionality: bo
> >>> - *
> >>> - * SUBTEST: xe_dma_buf
> >>> - * Description: Kernel dynamic selftests for dmabuf functionality.
> >>> - * Functionality: dmabuf test
> >>> - *
> >>> - * SUBTEST: xe_migrate
> >>> - * Description:
> >>> - *	Kernel dynamic selftests to check if page table migrations
> >>> - *	are working properly.
> >>> - * Functionality: migrate
> >>> - *
> >>> - * SUBTEST: xe_mocs
> >>> - * Description:
> >>> - *	Kernel dynamic selftests to check mocs configuration.
> >>> - * Functionality: mocs configuration
> >>>    */
> >>>   
> >> Does that mean we will not have any subtests here? The point
> >> into moving subtests names into documentation was to have a testlist
> >> generated during compilation, so it will not depend on running
> >> test with --list option. Adding also Katarzyna to cc.
> >> Cc: Katarzyna Piecielska <katarzyna.piecielska at intel.com>
> >>
> >> Regards,
> >> Kamil
> >>
> >>> -static const char *live_tests[] = {
> >>> -	"xe_bo",
> >>> -	"xe_dma_buf",
> >>> -	"xe_migrate",
> >>> -	"xe_mocs",
> >>> -};
> >>> -
> >>>   igt_main
> >>>   {
> >>> -	int i;
> >>> +	IGT_LIST_HEAD(names);
> >>> +	IGT_LIST_HEAD(done);
> >>> +	struct igt_kunit_names *name, *cmp;
> >>> +
> >>> +	igt_kunit_get_test_names("xe_live_test", NULL, &names);
> >>> +
> >>> +	while (!igt_list_empty(&names)) {
> >>> +		bool found = false;
> >>> +
> >>> +		name = igt_list_first_entry(&names, name, link);
> >>> +		igt_list_del(&name->link);
> >>> +
> >>> +		/*
> >>> +		 * Retuned list is sub-tests.
> >>> +		 * So, need filter out duplicated top level names.
> >>> +		 */
> >>> +		igt_list_for_each_entry(cmp, &done, link) {
> >>> +			if (strcmp(name->suite, cmp->suite) != 0)
> >>> +				continue;
> >>> +
> >>> +			found = true;
> >>> +		}
> >>> +
> >>> +		if (found)
> >>> +			continue;
> >>> +
> >>> +		igt_list_add(&name->link, &done);
> >>>   
> >>> -	for (i = 0; i < ARRAY_SIZE(live_tests); i++)
> >>> -		igt_kunit("xe_live_test", live_tests[i], NULL);
> >>> +		igt_kunit("xe_live_test", name->suite, NULL);
> >>> +	}
> >>>   }
> >
> >
> >
> 
> 






More information about the igt-dev mailing list