dri-devel.lists.freedesktop.org archive mirror
 help / color / mirror / Atom feed
From: David Gow <davidgow@google.com>
To: "Maíra Canal" <maira.canal@usp.br>
Cc: "Harrison Chiu" <harrison.chiu@amd.com>,
	"Daniel Latypov" <dlatypov@google.com>,
	"Brendan Higgins" <brendanhiggins@google.com>,
	"Isabella Basso" <isabbasso@riseup.net>,
	andrealmeid@riseup.net, "Jun Lei" <jun.lei@amd.com>,
	"Rodrigo Siqueira" <Rodrigo.Siqueira@amd.com>,
	"Javier Martinez Canillas" <javierm@redhat.com>,
	amd-gfx@lists.freedesktop.org, "Leo Li" <sunpeng.li@amd.com>,
	mwen@igalia.com, "Sean Paul" <seanpaul@chromium.org>,
	"KUnit Development" <kunit-dev@googlegroups.com>,
	"Mark Yacoub" <markyacoub@chromium.org>,
	"Linux Kernel Mailing List" <linux-kernel@vger.kernel.org>,
	"Dmytro Laktyushkin" <Dmytro.Laktyushkin@amd.com>,
	"Nicholas Choi" <Nicholas.Choi@amd.com>,
	"Alex Deucher" <alexander.deucher@amd.com>,
	"Christian König" <christian.koenig@amd.com>
Subject: Re: [RFC 0/3] drm/amd/display: Introduce KUnit to Display Mode Library
Date: Thu, 16 Jun 2022 22:39:10 +0800	[thread overview]
Message-ID: <CABVgOSmesj5MGfQrtdWCgXzm1VXRoG0fAMCbkBCAvtqediqAjQ@mail.gmail.com> (raw)
In-Reply-To: <20220608010709.272962-1-maira.canal@usp.br>

On Wed, Jun 8, 2022 at 9:08 AM Maíra Canal <maira.canal@usp.br> wrote:
> This RFC is a preview of the work being developed by Isabella Basso [1],
> Maíra Canal [2], and Tales Lelo [3], as part of their Google Summer of Code
> projects [4], and Magali Lemes [5], as part of her capstone project.
> Our main goal is to bring unit testing to the AMDPGU driver; in particular,
> we'll focus on the Display Mode Library (DML) for DCN2.0 and some of the DCE
> functions. The modern AMD Linux kernel graphics driver is the single largest
> driver in the mainline Linux codebase [6]. As AMD releases new GPU models,
> the size of AMDGPU drivers is only becoming even larger.
> Assuring the drivers' quality and reliability becomes a complex task without
> systematic testing, especially for graphic drivers - which usually involve
> tons of complex calculations. Also, keeping bugs away becomes an increasingly
> hard task with the introduction of new code. Moreover, developers might want
> to refactor old code without fear of the introduction of new issues.
> In that sense, it is possible to argue for the benefits of implementing unit
> testing at the AMDGPU drivers. This implementation will help developers to
> recognize bugs before they are merged into the mainline and also makes it
> possible for future code refactors of the AMDGPU driver.
> When analyzing the AMDGPU driver, a particular part of the driver highlights
> itself as a good candidate for the implementation of unit tests: the Display
> Mode Library (DML), as it is focused on mathematical operations.
> For the implementation of the tests, we decided to go with the Kernel Unit
> Testing Framework (KUnit). KUnit makes it possible to run test suites on
> kernel boot or load the tests as a module. It reports all test case results
> through a TAP (Test Anything Protocol) in the kernel log.
> Moreover, KUnit unifies the test structure and provides tools to simplify the
> testing for developers and CI systems.
> That said, we developed a little snippet on what we intend to develop in our
> summer. We planned the basic structure on how the tests will be introduced
> into the codebase and, on the concern of the CI systems, developed a structure
> where the unit tests can be introduced as modules and run on IGT (the IGT patch
> will be introduced soon).

It's awesome to see this! It's definitely one of the more ambitious
KUnit test projects out there, and the DML code does seem particularly
well suited to unit-testings.

> The way the modules are implemented might seem a little unusual for KUnit
> developers. We need to call the KUnit init function inside the AMDGPU stack,
> otherwise, the test won't compile as a module. So, the solution to this
> problem was based on the unit tests for the Thunderbolt driver, which uses
> KUnit and also tests a physical driver.
> As kunit_test_suites() defines itself as an init_module(), it conflicts with
> the existing one at amdgpu_drv. So, if we use kunit_test_suites(), we won't
> be able to compile the tests as modules and, therefore, won't be able to use
> IGT to run the tests. This problem with kunit_test_suites() was already
> discussed in the KUnit mailing list, as can be seen in [7].

I'm not sure I fully understand why these tests need to be part of the
amdgpu module, though admittedly I've not played with IGT much. Would
it be possible to compile these tests as separate modules, which could
depend on amdgpu (or maybe include the DML stuff directly), and
therefore not have this conflict? I definitely was able to get these
tests working under kunit_tool (albeit as built-ins) by using
kunit_test_suites(). If each suite were built as a separate module (or
indeed, even if all the tests were in one module, with one list of
suites), then it should be possible to avoid the init_module()
conflict. That'd also make it possible to run these tests without
actually needing the driver to initialise, which seems like it might
require actual hardware(?)

There are two other reasons the 'thunderbolt'-style technique is one
we want to avoid:
1. It makes it much more difficult to run tests using kunit_tool and
KUnit-based CI tools: these tests would not run automatically, and if
they were built-in as-is, they'd need to be
2. We're planning to improve module support to replace the
init_module()-based implementation of kunit_test_suites() with one
which won't have these conflicts, so the need for this should be

If you're curious, an early version of the improved module support can
be found here, though it's out-of-date enough it won't apply or work

Now, that's unlikely to be ready very soon, but I'd be hesitant to
implement too extensive a system for avoiding kunit_test_suites()
given at some point it should work and we'll need to migrate back to

At the very least, having the dependency on KUNIT=m is a very bad
idea: it should be possible to have tests built as modules, even if
KUnit itself isn't, and ideally (even if this sort-of implementation
is required), it _should_ be possible to have these tests be built-in
if all their dependencies (KUnit, amdgpu) are, which would make it
possible to run the tests without a userland.

That being said, I've got basically no knowledge of amdgpu (or even
drm in general), so there could be something I'm missing.

> The first patch configures the basic structure of the KUnit Tests, setting the
> proper Makefile, Kconfig, and init function. It also contains a simple test
> involving DML logging, which is the pretext for building the testing structure.
> The second patch adds KUnit tests to bw_fixed functions. This patch represents
> what we intend to do on the rest of the DML modules: systematic testing of the
> public functions of the DML, especially mathematically complicated functions.
> Also, it shows how simple it is to add new tests to the DML with the structure
> we built.
> Any feedback or ideas for the project are welcome!
Looks great to me so far: I'll try to get a more detailed review in soon.

-- David

> [1] https://crosscat.me
> [2] https://mairacanal.github.io
> [3] https://tales-aparecida.github.io/
> [4] https://summerofcode.withgoogle.com/programs/2022/organizations/xorg-foundation
> [5] https://magalilemes.github.io/
> [6] https://www.phoronix.com/scan.php?page=news_item&px=AMDGPU-Closing-4-Million
> [7] https://groups.google.com/g/kunit-dev/c/hbJbh8L37FU/m/EmszZE9qBAAJ
> - Isabella Basso, Magali Lemes, Maíra Canal, and Tales Lelo
> Magali Lemes (1):
>   drm/amd/display: Introduce KUnit tests to the bw_fixed library
> Maíra Canal (2):
>   drm/amd/display: Introduce KUnit to DML
>   drm/amd/display: Move bw_fixed macros to header file
>  drivers/gpu/drm/amd/display/Kconfig           |   1 +
>  .../gpu/drm/amd/display/amdgpu_dm/Makefile    |   5 +
>  .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c |   3 +
>  .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.h |   3 +
>  .../drm/amd/display/amdgpu_dm/tests/Kconfig   |  41 +++
>  .../drm/amd/display/amdgpu_dm/tests/Makefile  |  18 +
>  .../amdgpu_dm/tests/calcs/bw_fixed_test.c     | 322 ++++++++++++++++++
>  .../amdgpu_dm/tests/display_mode_lib_test.c   |  83 +++++
>  .../amd/display/amdgpu_dm/tests/dml_test.c    |  26 ++
>  .../amd/display/amdgpu_dm/tests/dml_test.h    |  21 ++
>  .../drm/amd/display/dc/dml/calcs/bw_fixed.c   |  14 +-
>  drivers/gpu/drm/amd/display/dc/inc/bw_fixed.h |  14 +
>  12 files changed, 538 insertions(+), 13 deletions(-)
>  create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/tests/Kconfig
>  create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/tests/Makefile
>  create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/tests/calcs/bw_fixed_test.c
>  create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/tests/display_mode_lib_test.c
>  create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/tests/dml_test.c
>  create mode 100644 drivers/gpu/drm/amd/display/amdgpu_dm/tests/dml_test.h
> --
> 2.36.1

  parent reply	other threads:[~2022-06-16 14:39 UTC|newest]

Thread overview: 12+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-06-08  1:07 [RFC 0/3] drm/amd/display: Introduce KUnit to Display Mode Library Maíra Canal
2022-06-08  1:07 ` [RFC 1/3] drm/amd/display: Introduce KUnit to DML Maíra Canal
2022-06-08  2:36   ` Daniel Latypov
2022-06-15 15:05     ` Maíra Canal
2022-06-08  1:07 ` [RFC 2/3] drm/amd/display: Move bw_fixed macros to header file Maíra Canal
2022-06-08  1:07 ` [RFC 3/3] drm/amd/display: Introduce KUnit tests to the bw_fixed library Maíra Canal
2022-06-16 14:39 ` David Gow [this message]
2022-06-16 22:41   ` [RFC 0/3] drm/amd/display: Introduce KUnit to Display Mode Library Maíra Canal
2022-06-17  7:55     ` David Gow
2022-06-17 20:24       ` Maíra Canal
2022-06-18  9:08         ` David Gow
2022-06-22 22:55           ` Rodrigo Siqueira Jordao

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:

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CABVgOSmesj5MGfQrtdWCgXzm1VXRoG0fAMCbkBCAvtqediqAjQ@mail.gmail.com \
    --to=davidgow@google.com \
    --cc=Dmytro.Laktyushkin@amd.com \
    --cc=Nicholas.Choi@amd.com \
    --cc=Rodrigo.Siqueira@amd.com \
    --cc=alexander.deucher@amd.com \
    --cc=amd-gfx@lists.freedesktop.org \
    --cc=andrealmeid@riseup.net \
    --cc=brendanhiggins@google.com \
    --cc=christian.koenig@amd.com \
    --cc=dlatypov@google.com \
    --cc=dri-devel@lists.freedesktop.org \
    --cc=harrison.chiu@amd.com \
    --cc=isabbasso@riseup.net \
    --cc=javierm@redhat.com \
    --cc=jun.lei@amd.com \
    --cc=kunit-dev@googlegroups.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=magalilemes00@gmail.com \
    --cc=maira.canal@usp.br \
    --cc=markyacoub@chromium.org \
    --cc=mwen@igalia.com \
    --cc=seanpaul@chromium.org \
    --cc=sunpeng.li@amd.com \
    --cc=tales.aparecida@gmail.com \


* 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;
as well as URLs for NNTP newsgroup(s).