Currently, the containers are used only to build IGT. We also want to
have an option to run IGT inside a container. For this, new build stage
is added.

Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

Since different images can be present at the same time, it is better to
always use the one built for this particular git reference.

Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

V3:
 - Change $CI_COMMIT_REF_NAME to commit-$CI_COMMIT_SHA

Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

Base images rarely change, there is not much sense in rebuilding it on
every commit. GitLab already has mechanism for detecting such changes.
However, it is only able to prevent rebuilding whenever there is no
changes within the same ref. Since in our CI system, git tag is created
on every series, the mechanism doesn't work.

One possible way to workaround that is to compute a checksum of a
Dockerfile, and rebuilding only if there was no image built with the
same checksum.

V2:
 - Rename gitlab-ci to .gitlab-ci (Arek)
 - Add commit-$CI_COMMIT_SHA

Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

Dockerfile.build-* is consistent with build-* naming.

Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

igt-* better suits for containers which contain IGT inside, while
build-* is better for containers which are used to build IGT.

Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

No functional changes. This order is consistent with the order of
execution stages.

Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

Comments say that a connector can only be forced if it's disconnected. This was
a wrong assumption which has been fixed, but the comments went out of sync.

Fixes: dc3a0934 ("tests/kms_hdmi_inject: Accept any HDMI connector")
Signed-off-by: Simon Ser <simon.ser@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Reviewed-by: Ville Syrjälä <ville.syrjala@linux.intel.com>

We were missing the level and control method.

Reviewed-by: Simon Ser <simon.ser@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

Similar to other panel type indexed arrays.

Reviewed-by: Simon Ser <simon.ser@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

The test assumes that the dmabuf for the batch buffer has no exclusive
fence. This is only true for the batch as it executes on the GPU; prior
to that we may be in the process of moving it into memory and the
ppGTT which cause the batch to carry an exclusive fence. Keep the test
simple and prebind the batch buffer so that when we poll the dmabuf fd,
we only expect it to be the spinning batch.

Alternatively we could wait until we start writing values into scratch
(i.e. igt_spin_busywait_until_started()). This seemed simpler.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Matthew Auld <matthew.auld@intel.com>
Acked-by: Matthew Auld <matthew.auld@intel.com>

e points into the buffer pointed to by p.

Fixes: dfda0b6a ("intel_reg: introduce one intel_reg tool to rule them all")
Reviewed-by: Simon Ser <simon.ser@intel.com>
Signed-off-by: Jani Nikula <jani.nikula@intel.com>

The available range reported by the driver is the inclusive [min_bpc,
max_bpc], so fix the upper bound used by the test accordingly. This
issue had the side-effect (on ICL for instance) of limiting 'max bpc' to
10 (instead of restoring it to the default 12) for all the tests
following kms_properties, as seen from the logs in the referenced bug
below.

Reference: https://bugs.freedesktop.org/show_bug.cgi?id=109593


Cc: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>
Reviewed-by: Simon Ser <simon.ser@intel.com>

If we try to idle while another thread is submitting, we will be forced
to wait until that other thread is finished -- effectively serialising
the parallel workloads, defeating said purpose.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Stuart Summers <stuart.summers@intel.com>

Getting network back after waking up from suspend takes some time.
If it's not quick enough, it may cause any following chamelium_rpc
to fail.

This includes the chamelium port reset in exit handler:

	igt_chamelium-CRITICAL: Test assertion failure function chamelium_rpc,
				file ../lib/igt_chamelium.c:349:
	igt_chamelium-CRITICAL: Failed assertion: !chamelium->env.fault_occurred
	igt_chamelium-CRITICAL: Last errno: 113, No route to host
	igt_chamelium-CRITICAL: Chamelium RPC call failed: libcurl failed to execute
				the HTTP POST transaction, explaining:
				Failed to connect to 192.168.1.224 port 9992: No route to host

	Stack trace:
	  #0 ../lib/igt_core.c:1674 __igt_fail_assert()
	  #1 [chamelium_rpc+0x1ea]
	  #2 ../lib/igt_chamelium.c:2274 chamelium_reset()
	  #3 ../lib/igt_chamelium.c:2361 chamelium_deinit()
	  #4 ../lib/igt_core.c:2124 call_exit_handlers()

So let's wait for Chamelium to be reachable after each suspend_autoresume.

v2: Bump the network timeout to 20s,
    the slowest observed bring-up was 17s. (Petri)
    Do not DECREF if we get NULL from __chmelium_rpc(). (Simon & CI)

Cc: Simon Ser <simon.ser@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=109380


Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Simon Ser <simon.ser@intel.com>

Antonio Argenziano authored 5 years ago and Petri Latvala committed 5 years ago

dmabuf_selftests has been recently added to the test list but not added
to the special handling in igt_command_line.sh and therefore build
testing would fail.

Signed-off-by: Antonio Argenziano <antonio.argenziano@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
[Fixed syntax error and test name]
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

Just a plain igt_kmod() wrapper for dmabuf_seltests.ko

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Tomi Sarvela <tomi.p.sarvela@intel.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>

The kms_plane_multiple@* tests fail on amdgpu with a dmesg-warn due to

"atomic remove_fb failed with -22"
"WARNING: CPU: 6 PID: 183 at drivers/gpu/drm/drm_framebuffer.c:986"

The call to drmModeRmFB essentially creates an atomic commit internally
when we drop the last reference to the FB in IGT. This commit disables
the plane associated with framebuffer.

Since the planes are iterated by increasing index in IGT the last
plane to have its framebuffer removed is the cursor plane on AMDGPU.

On AMDGPU we can't enable the cursor plane without any active hardware
planes (primary, overlay) since there is no dedicated cursor plane.

The commits to release the overlay plane fail since it's the last
remaining plane and the cursor plane is still active.

The fix to the test is to just perform a commit that disables all 3
planes at the same time before calling remove_fb.

This improves performance enables support for drivers that don't
support cursor planes being enabled without a primary or overlay.

Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Reviewed-by: Leo Li <sunpeng.li@amd.com>

Oleg Vasilev authored 5 years ago and Simon Ser committed 5 years ago

The test was introduced in earlier series, but was only added
to meson test list.

Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>
Cc: Simon Ser <simon.ser@intel.com>
Fixes: 93129f55 ("tests/kms_prime: add vendor-agnostic kms prime tests")
Reviewed-by: Simon Ser <simon.ser@intel.com>

test_edid_read sets an EDID on a port, plugs it and reads it back. Sometimes
retrieving the blob fails. Check whether the blob is set at all before trying
to get it. This makes it obvious it's not drmModeGetPropertyBlob failing for
another reason.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=106766


Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

Oleg Vasilev authored 5 years ago and Simon Ser committed 5 years ago

Currently, there are different sets of prime tests:
 - vgem + i915
 - amdgpu + i915
 - nouveau + i915

The idea is to create a set of tests which are expected to work on any
prime-compatible driver. It can be run with any combination of
exporter + importer, while those devices have respective prime
capabilities. In order to be vendor-agnostic, tests should use generic
KMS API.

Since vgem can be used as both exporter and importer, and there aren't
any generic kms features which vgem doesn't support, it is sufficient
to test vgem as exporter with any other KMS driver.

The first test is simple:
1. Exporter creates a dumb FB and fills it with a plain color
2. FB is transferred to the importer
3. Importer modesets and computes pipe CRC
4. Importer draws the same color through cairo and compares CRC

The initial motivation comes from the need to test prime support in
vkms.

V2:
 - Chris: rename the file
 - Simon: memory hygiene, codestyle

Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Reviewed-by: Simon Ser <simon.ser@intel.com>
Cc: Rodrigo Siqueira <rodrigosiqueiramelo@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: Simon Ser <daniel@ffwll.ch>
Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>

Oleg Vasilev authored 5 years ago and Simon Ser committed 5 years ago

Make it public.

Signed-off-by: Oleg Vasilev <oleg.vasilev@intel.com>
Reviewed-by: Simon Ser <simon.ser@intel.com>

To run the independent subtests requires to issue the requests
not-in-submission order, we need a scheduler.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>

This test generates an EDID with support for modes tied to a specific aspect
ratio. It then plugs a connector with this EDID and iterates the list of modes
to find the one with the expected aspect ratio flag. The connector is then
enabled and the Chamelium board is used to capture the AVI InfoFrame. The
InfoFrame fields are then checked.

To advertise support for a mode with a specific aspect ratio, sinks add Video
Identification Codes in the CEA extension. The VIC chosen by the source is
contained in the AVI InfoFrame. The InfoFrame also contains a picture aspect
ratio field, but it's only able to indicate whether the aspect ratio is 16:9 or
4:3.

For now the test is only enabled for HDMI. I've tried to make it work on DP
too, but after reading the kernel code it seems like we don't support AVI
InfoFrames on DP.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

This commit adds partial support for AVI InfoFrames. Only bytes 1 and 2 of the
InfoFrame payload are parsed.

These bytes contain (among other things) information about the aspect ratio of
the frames.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

This commit adds an error message to connector status assertions.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

Explicit Chamelium port mappings are now optional.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

This makes it easier to tell the code block apart from the rest of the
document.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

GCC doesn't seem smart enough to figure out that everything is fine. Let's just
initialize the variable to silence the warning.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

This test first checks that on normal hotplug, the link-status property is set
to GOOD. It then triggers a link failure, checks that the property is updated
to BAD. Checks are performed to make sure the new mode list is a subset of the
previous one. After that, a new modeset is performed if necessary and the
link-status property is set back to GOOD.

This whole dance is repeated until we reach fallback modes.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

This patch simply adds link-status to our list of properties.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Manasi Navare <manasi.d.navare@intel.com>

This new function triggers a link failure on the provided port. Another
function called chamelium_supports_trigger_link_failure has been added to
check whether the Chamelium board supports the new method.

Support for TriggerLinkFailure in Chamelium has been added in [1].

[1]: https://chromium.googlesource.com/chromiumos/platform/chameleon/+/44866c7bc2af44da1f7bac4e782d0ed6d7c1400a



Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

Callers no longer need to call edid_ext_update_cea_checksum on each extension
block. This makes it less error-prone to miss one block by mistake.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

This array is filled manually, some bytes might have the MSB set.

This fixes the following warning:

    ../lib/igt_kms.c:357:18: warning: implicit conversion from 'int' to 'char' changes value from 128 to -128 [-Wconstant-conversion]
            hdmi->data[0] = HDMI_VSDB_VIDEO_3D_PRESENT; /* HDMI video flags */
    	                      ~ ^~~~~~~~~~~~~~~~~~~~~~~~~~

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

EDID_LENGTH is misleading because EDIDs are a variable size (they contain one
or more 128-byte EDID blocks). This commit renames it to EDID_BLOCK_SIZE which
makes it clear users need to call edid_get_size to get the total size.

The declaration has also been moved to igt_edid.

("Size" has been chosen over "length" because it's clearer that it's a number
of bytes, not a number of elements)

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

This has several advantages:

* No more need to convert back and forth between these two (everybody should
  use struct edid, the exception being lib/tests/igt_edid which performs sanity
  checks)
* Makes it clearer that users can call edid_get_size on a returned EDID blob
* Improves type safety (it's more obvious is a random blob is used as an EDID)

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

If the DUT sends an InfoFrame, check that its values are correct. Per the HDMI
and DP specs, most of the fields can be set to "refer to stream header" because
some information is duplicated. The DP spec also says that InfoFrames are
optional if mono or stereo audio is used (because there's no channel speaker
allocation).

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>

This new call retrieves the last InfoFrame received by the Chamelium board.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>

This commit introduces a new igt_infoframe library, used to parse InfoFrames.
For now only audio InfoFrames are supported. Support for AVI and other types of
InfoFrames is planned (and will come with the matching tests).

Unlike igt_edid, InfoFrames are parsed into a higher-level user-friendly
struct.

Signed-off-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>

Ramalingam C authored 5 years ago and Petri Latvala committed 5 years ago

At the start of every test, reset HDCP state to UNDESIRED on all
connectors, so that we will not have any hdcp related noises in
the subsequent tests.

v2:
  igt_output_set_prop_enum is used.
  Documentation of display_reset is updated.

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111390


Reviewed-by: Simon Ser <simon.ser@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

Still trying to hit a deadlock with userptr from kcompatcd.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Reveiwed-by: Andi Shyti <andi.shyti@intel.com>