From: "Melkaveri, Arjun" <arjun.melkaveri@intel.com>
To: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Cc: igt-dev@lists.freedesktop.org
Subject: Re: [igt-dev] [PATCH] [PATCH i-g-t] gem_ctx_create:Adjusted test to use __for_each_physical_engine to utilize all available engines
Date: Wed, 5 Feb 2020 14:43:06 +0530 [thread overview]
Message-ID: <20200205091306.GB1580@arjun-NUC8i7BEH> (raw)
In-Reply-To: <7c117c23-98a3-efbe-ac59-f85ac2b63b19@linux.intel.com>
On Tue, Feb 04, 2020 at 04:49:47PM +0000, Tvrtko Ursulin wrote:
>
> On 03/02/2020 02:52, Arjun Melkaveri wrote:
> > Changes :-
> > 1) Added __for_each_physical_engine for subtest engines.
> > 2) Moved test cases from static to dynamic.
> > 3) used gem_context_clone_with_engines for creating contexts.
> > 4) Passing NULL in active test to run test for all engines.
> >
> > Cc: Chris Wilson <chris@chris-wilson.co.uk>
> > Cc: Ursulin Tvrtko <tvrtko.ursulin@intel.com>
> > Signed-off-by: Arjun Melkaveri <arjun.melkaveri@intel.com>
> > ---
> > tests/i915/gem_ctx_create.c | 56 ++++++++++++++++++++++---------------
> > 1 file changed, 34 insertions(+), 22 deletions(-)
> >
> > diff --git a/tests/i915/gem_ctx_create.c b/tests/i915/gem_ctx_create.c
> > index d9a820e2..7100e75a 100644
> > --- a/tests/i915/gem_ctx_create.c
> > +++ b/tests/i915/gem_ctx_create.c
> > @@ -124,22 +124,22 @@ static void files(int core, int timeout, const int ncpus)
> > gem_close(core, batch);
> > }
> > -static void active(int fd, unsigned engine, int timeout, int ncpus)
> > +static void active(int fd, const struct intel_execution_engine2 *e,
> > + int timeout, int ncpus)
> > {
> > const uint32_t bbe = MI_BATCH_BUFFER_END;
> > struct drm_i915_gem_execbuffer2 execbuf;
> > struct drm_i915_gem_exec_object2 obj;
> > unsigned int nengine, engines[16];
> > unsigned *shared;
> > -
> > - if (engine == ALL_ENGINES) {
> > + /* When e is NULL, test would run for all engines */
> > + if (!e) {
> > igt_require(all_nengine);
> > nengine = all_nengine;
> > memcpy(engines, all_engines, sizeof(engines[0])*nengine);
> > } else {
> > - gem_require_ring(fd, engine);
> > nengine = 1;
> > - engines[0] = engine;
> > + engines[0] = e->flags;
> > }
> > shared = mmap(NULL, 4096, PROT_WRITE, MAP_SHARED | MAP_ANON, -1, 0);
> > @@ -157,7 +157,7 @@ static void active(int fd, unsigned engine, int timeout, int ncpus)
> > igt_fork(child, ppgtt_nengine) {
> > unsigned long count = 0;
>
> After fork you need to transfer the engine map from parent fd default
> context. (gem_context_copy_engines)
>
> There are two more forks which needs this treatment on a quick look.
>
> Also, since you have __for_each_physical_engine in the top level
> igt_fixture, I think _all_ subtests run with default ctx engine map. So all
> which submit to one from either all_engines or ppgtt_engines need to make
> sure engine maps are aligned. I can see files(), which should be covered
> already when you handle the igt_fork properly. But please double check me
> and audit everything.
>
Will make changes , I need to modify files() too . It open Drm driver
directly . Will change it to gem_reopn_driver .
> > - if (ppgtt_engines[child] == engine)
> > + if (ppgtt_engines[child] == e->flags)
> > continue;
>
> What is continue doing outside a loop, I'm confused? Does it relate to
> igt_fork? Spawns N children and then exists some immediately?
>
you are correct , This is related to igt_fork
> > execbuf.flags = ppgtt_engines[child];
> > @@ -183,7 +183,7 @@ static void active(int fd, unsigned engine, int timeout, int ncpus)
> > clock_gettime(CLOCK_MONOTONIC, &start);
> > do {
> > do {
> > - execbuf.rsvd1 = gem_context_create(fd);
> > + execbuf.rsvd1 = gem_context_clone_with_engines(fd, 0);
> > for (unsigned n = 0; n < nengine; n++) {
> > execbuf.flags = engines[n];
> > gem_execbuf(fd, &execbuf);
> > @@ -276,7 +276,9 @@ static void maximum(int fd, int ncpus, unsigned mode)
> > err = -ENOMEM;
> > if (avail_mem > (count + 1) * ctx_size)
> > - err = __gem_context_create(fd, &ctx_id);
> > + err = __gem_context_clone(fd, 0,
> > + I915_CONTEXT_CLONE_ENGINES,
> > + 0, &ctx_id);
> > if (err) {
> > igt_info("Created %lu contexts, before failing with '%s' [%d]\n",
> > count, strerror(-err), -err);
> > @@ -370,6 +372,7 @@ static void basic_ext_param(int i915)
> > gem_context_destroy(i915, create.ctx_id);
> > +
>
> Try to avoid noise.
>
will note it
> > /* Having demonstrated a valid setup, check a few invalids */
> > ext.param.ctx_id = 1;
> > igt_assert_eq(create_ext_ioctl(i915, &create), -EINVAL);
> > @@ -524,6 +527,7 @@ igt_main
> > {
> > const int ncpus = sysconf(_SC_NPROCESSORS_ONLN);
> > struct drm_i915_gem_context_create create;
> > + const struct intel_execution_engine2 *e;
> > int fd = -1;
> > igt_fixture {
> > @@ -531,8 +535,8 @@ igt_main
> > igt_require_gem(fd);
> > gem_require_contexts(fd);
> > - for_each_physical_engine(e, fd)
> > - all_engines[all_nengine++] = eb_ring(e);
> > + __for_each_physical_engine(fd, e)
> > + all_engines[all_nengine++] = e->flags;
> > igt_require(all_nengine);
> > if (gem_uses_full_ppgtt(fd)) {
> > @@ -572,20 +576,28 @@ igt_main
> > igt_subtest("forked-files")
> > files(fd, 20, ncpus);
> > + /* NULL value means all engines */
> > igt_subtest("active-all")
> > - active(fd, ALL_ENGINES, 20, 1);
> > + active(fd, NULL, 20, 1);
> > igt_subtest("forked-active-all")
> > - active(fd, ALL_ENGINES, 20, ncpus);
> > -
> > - for (const struct intel_execution_engine *e = intel_execution_engines;
> > - e->name; e++) {
> > - igt_subtest_f("active-%s", e->name)
> > - active(fd, eb_ring(e), 20, 1);
> > - igt_subtest_f("forked-active-%s", e->name)
> > - active(fd, eb_ring(e), 20, ncpus);
> > - if (e->exec_id) {
> > - igt_subtest_f("hog-%s", e->name)
> > - active(fd, eb_ring(e), 20, -1);
> > + active(fd, NULL, 20, ncpus);
> > +
> > + igt_subtest_with_dynamic("active") {
> > + __for_each_physical_engine(fd, e) {
> > + igt_dynamic_f("%s", e->name)
>
> Does igt_dynamic_f(e->name) work?
>
This does work , verified it when test is executed . engine name is
logged as " Dynamic subtest "
> > + active(fd, e, 20, 1);
> > + }
> > + }
> > + igt_subtest_with_dynamic("forked-active") {
> > + __for_each_physical_engine(fd, e) {
> > + igt_dynamic_f("%s", e->name)
> > + active(fd, e, 20, ncpus);
> > + }
> > + }
> > + igt_subtest_with_dynamic("hog") {
> > + __for_each_physical_engine(fd, e) {
> > + igt_dynamic_f("%s", e->name)
> > + active(fd, e, 20, -1);
> > }
> > }
> >
>
> And finally, you can send this to upstream igt-dev in the next incarnation.
>
> Regards,
>
> Tvrtko
_______________________________________________
igt-dev mailing list
igt-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/igt-dev
next prev parent reply other threads:[~2020-02-05 9:13 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-02-03 2:52 [igt-dev] [PATCH] [PATCH i-g-t] gem_ctx_create:Adjusted test to use __for_each_physical_engine to utilize all available engines Arjun Melkaveri
2020-02-03 3:37 ` [igt-dev] ✓ Fi.CI.BAT: success for " Patchwork
2020-02-03 8:32 ` [igt-dev] ✗ GitLab.Pipeline: failure " Patchwork
2020-02-04 16:49 ` [igt-dev] [PATCH] [PATCH i-g-t] " Tvrtko Ursulin
2020-02-05 9:13 ` Melkaveri, Arjun [this message]
2020-02-05 12:33 ` [igt-dev] ✗ Fi.CI.IGT: failure for " Patchwork
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200205091306.GB1580@arjun-NUC8i7BEH \
--to=arjun.melkaveri@intel.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=tvrtko.ursulin@linux.intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox