Currently the last line of the spreadsheet is not processed.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>

Automatically creating/updating wildcards is not an easy task.
This script can't do it, so make it clearer that it will be
skipped.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>

Newer platforms (Xe2) supports compression on system memory so subtests
which might be executed are not only limited to discrete/vram. Enabling
dynamic subtest in ctrl-surf-copy test depends on surface compression
property. Lets use introduced in previous patch helper which returns
does region supports surface compression.

Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20240313104520.148338-3-zbigniew.kempczynski@intel.com


Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>

Some constructs like querying if device is discrete or getting memory
region origin are common so add new file to place those functions into.

Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20240313104520.148338-2-zbigniew.kempczynski@intel.com


Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>

Test submits long kernel with a higher threadgroup count, lower
iteration kernel and a short opencl kernel to exercise threadgroup
preemption scenario with WMTP disabled.

v2: Use macros instead of const values. Add documentation for
    validation check. (Nirmoy)

Cc: Nirmoy Das <nirmoy.das@intel.com>
Signed-off-by: Janga Rahul Kumar <janga.rahul.kumar@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>

Add printing of version information to basic subtest.

Cc: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab@kernel.org>

Create one basic subtest with the ioctl check.

Cc: Helen Koike <helen.koike@collabora.com>
Signed-off-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Reviewed-by: Mauro Carvalho Chehab <mchehab@kernel.org>

The flag to force remove a module should never be used. The flag is
intended for end users to decide if they want to have a broken system
left after that. Maybe it's a "controlled breakage". It's not something
for a testsuite to use.

If the kernel ever resorts to honoring the flag (due to module's kref
not being 0), we will have a tainted kernel, potential leaks and memory
corruption. It's better to just handle the error returned by libkmod and
fix the root cause.

Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

Matt Roper authored 1 year ago and Kamil Konieczny committed 1 year ago

The MI_STORE_REGISTER_MEM macro in IGT's instruction header is the "old"
version of this define, intended for use on pre-gen8 platforms since it
bakes in the fixed 3-dword instruction length in the instruction header.
For gen8 and beyond (i.e., every platform the Xe driver supports), the
MI_STORE_REGISTER_MEM_GEN8 macro should be used instead, which properly
bakes in the 4-dword instruction length.  Since the xe_spin library was
using the old macro, but then manually |'ing in a "2" at the end of the
expression, it was effectively winding up with a field value of 3 (which
would represent a 5-dword instruction).

Switch xe_spin to using the "GEN8" macro and drop the redundant "| 2"
which is already handled by the macro itself.

Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Reviewed-by: Nirmoy Das <nirmoy.das@intel.com>

Himanshu Girotra authored 1 year ago and Kamil Konieczny committed 1 year ago

Unblock core hotunplug for XE as it is one of the core driver test and
should not create problem anymore since XE no more in enabling stage.

Signed-off-by: Himanshu Girotra <himanshu.girotra@intel.com>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

Be consistent with other drm_open_xxx functions.

Signed-off-by: Marcin Bernatowicz <marcin.bernatowicz@linux.intel.com>
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Reviewed-by: Lukasz Laguna <lukasz.laguna@intel.com>

Add a verbosity parameter to hide messages used mostly for debugging
purposes, as, on several cases, the output may be misleading.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

Add a short description of what function does.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

The current logic relies on dict = sorted(dict) to preserve the
sorted order. Ordered dicts were introduced only on python 3.7.

This is a silly requirement, as all we want is to sort the
dict.item() tuple. Change the logic to avoid the need of
checking for an specific python version.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

- add a new parameter to optionally generate JSON debug files;
- don't use default formatter;
- don't use args SUPRESS;
- don't use an empty epilog.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

Use autopep8 and some manual adjustments to make pylint
happier.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

Move the main code to a function, called only when the script
is executed directly from command line.

That allows the method to be re-used by other scripts if needed.

No functional changes.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>

https://patchwork.freedesktop.org/patch/582554/?series=131032&rev=1


groups all the live tests into a single kernel module. Use that
module to run any of the tests triggered by igt.

Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20240312174533.2417630-1-lucas.demarchi@intel.com


Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

The assert is unnecessary at this early while init phase
of the test so skip the test instead when are no drm devices.

v2: fix a typo.
v3:  s/igt_assert_fd/igt_require_fd(Janusz)

Closes: https://gitlab.freedesktop.org/drm/intel/-/issues/10087


Signed-off-by: Nirmoy Das <nirmoy.das@intel.com>
Reviewed-by: Janusz Krzysztofik <janusz.krzysztofik@linux.intel.com>

Currently the tests are failing with OOO memory error.
Test requires SRAM to be as big as VRAM. For example, small-cm creates
(448 / 2) BOs with a size (1 / 128) of the total VRAM size. For
simplicity ensure the SRAM size >= VRAM before running this test.
Added code to skip the test if the system memory size is less than local
memory size.

v2: Added memory size in the skip print.(kamil)
Also added reason as a comment for the skip.(Matt)

Cc: Matthew Brost <matthew.brost@intel.com>
Cc: Venkata Ramana Nayana <venkata.ramana.nayana@intel.com>
Signed-off-by: Priyanka Dandamudi <priyanka.dandamudi@intel.com>
Reviewed-by: Matthew Brost <matthew.brost@intel.com>
Reviewed-by: Sai Gowtham Ch <sai.gowtham.ch@intel.com>

Update the query test to include the GT's IP version in the output and
to also perform some simple sanity checks on the version received.  We
don't want to try to match any specific version numbers (since we expect
new versions to show up in the future), but we do know that the uapi
should never return version numbers older than 12.70 for the primary GT
or 13.00 for the media GT since those were the first IP versions to
include GMD_ID support.  We also know that MTL/ARL are the only pre-Xe2
platforms to support GMD_ID, so if we're not running on MTL/ARL, the
version numbers must always be 20.00 or greater.

Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20240312205431.2869411-3-matthew.d.roper@intel.com


Signed-off-by: Matt Roper <matthew.d.roper@intel.com>

Align with commit ("drm/xe/uapi: Add IP version and stepping to GT list
query")

v2:
 - s/revid/rev/ to match rename on kernel series

Reviewed-by: Gustavo Sousa <gustavo.sousa@intel.com>
Link: https://lore.kernel.org/r/20240312205431.2869411-2-matthew.d.roper@intel.com


Signed-off-by: Matt Roper <matthew.d.roper@intel.com>

Francois Dugast authored 1 year ago and Kamil Konieczny committed 1 year ago

This aligns with kernel commit ("drm/xe: Extend uAPI to query HuC
micro-controler firmware version").

v2:
- Fix printing branch (Francois Dugast)
- Make ENODEV the only accepted error (José Roberto de Souza)

v3: Share code between guc and huc tests (John Harrison)

v4: Better use of asserts to report errors (John Harrison)

v5: Add lore link to corresponding kernel commit (Francois Dugast)

Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Link: https://lore.kernel.org/intel-xe/20240208183539.185095-2-jose.souza@intel.com/


Reviewed-by: José Roberto de Souza <jose.souza@intel.com>

Francois Dugast authored 1 year ago and Kamil Konieczny committed 1 year ago

This aligns with kernel commit ("drm/xe: Add uAPI to query GuC firmware
submission version").

v2: Print branch first, use memset to zero the query (John Harrison)

v3: Update header to match the KMD version with branch first (John Harrison)

v4: Add lore link to corresponding kernel commit (Francois Dugast)

Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Signed-off-by: Francois Dugast <francois.dugast@intel.com>
Link: https://lore.kernel.org/intel-xe/20240208183539.185095-1-jose.souza@intel.com/


Reviewed-by: José Roberto de Souza <jose.souza@intel.com>

Mauro Carvalho Chehab authored 1 year ago and Kamil Konieczny committed 1 year ago

The wildcard is wrong there, causing the ReST output to be:

	``igt@xe_exec_queue_property@timeslice_duration_us-property-min-max``

	:Description: Test to check if %s timeslice duration us schedule parameter checks for min max values.

After the change, the description is now correct:

	``igt@xe_exec_queue_property@timeslice_duration_us-property-min-max``

	:Description: Test to check if timeslice duration schedule parameter checks for min max values.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>

echo -n "" > /sys/module/<modulename>/parameters/<param>

doesn't really work as it will just create a open() + close() expecting
the file to be truncated. The same issue happens with igt as it will
stop writing when there are 0 chars to write. Special case the empty
string so it always write a '\0' and make sure igt_sysfs_set() accounts
for the extra null char.

Shell example:
	# echo -n "/foo" > /sys/module/firmware_class/parameters/path
	# cat /sys/module/firmware_class/parameters/path
	/foo
	# echo -n "" > /sys/module/firmware_class/parameters/path
	/foo
	# # make sure to actually write a \0:
	echo -ne "\0" > /sys/module/firmware_class/parameters/path
	# cat /sys/module/firmware_class/parameters/path

Same thing happens when testing igt_sysfs_set():
       int dirfd = open("/sys/module/firmware_class/parameters", O_RDONLY);
       igt_sysfs_set(dirfd, "path", "");

Previously it was not really setting the param.

v2:
  - Fix return code from igt_sysfs_vprintf() to differentiate between
    writing 1 or 0 chars (Janusz)
  - Document the behavior of igt_sysfs_set() as being a higher-level
    than igt_sysfs_write().
v3:
  - Partial "back to v1": include the \0 in the write only when writing
    an empty string: there are some files in sysfs like pm_test that
    don't like
v4:
  - Fix return code in igt_sysfs_vprintf() when fmt string is something
    like "%c<more_fmt>", c, ..." so it still returns > 0 (Janusz)

Reviewed-by: Janusz Krzysztofik <janusz.krzysztofik@linux.intel.com>
Link: https://lore.kernel.org/r/20240228223134.3908035-4-lucas.demarchi@intel.com


Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

vsnprintf() should receive the buffer size as argument, here called `len`,
including the trailing '\0'. There was truncation if the return is "size
or more". In this second call to vsnprintf() the value should be exactly
the same as in the first call, otherwise something really unexpected
happened.

Acked-by: Janusz Krzysztofik <janusz.krzysztofik@linux.intel.com>
Link: https://lore.kernel.org/r/20240228223134.3908035-3-lucas.demarchi@intel.com


Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

Make sure to close the fd before returning.

Acked-by: Janusz Krzysztofik <janusz.krzysztofik@linux.intel.com>
Link: https://lore.kernel.org/r/20240228223134.3908035-2-lucas.demarchi@intel.com


Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

Let's guarantee we always pass the same pointer to vsnprintf: if we are
using the array size, pass that instead of buf that could point to
something else in an eventual refactor.

Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: https://lore.kernel.org/r/20240228223134.3908035-1-lucas.demarchi@intel.com


Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

Instead of manipulating param width/heigth add dedicated variable
used in inc-dimension subtests. This fixes issue when all subtests
are run one-by-one what causes failures on comparing surfaces due
to different sizes.

Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20240311114730.77947-1-zbigniew.kempczynski@intel.com


Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>

devcoredump holds a module reference, blocking the module removal.

It is intentional from the devcoredump perspective to keep the
log available even after the unbind/unprobe. However it blocks
our module removal here.

v2: Accepting many suggestions from Lucas.

Cc: Lucas De Marchi <lucas.demarchi@intel.com>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Lucas De Marchi <lucas.demarchi@intel.com>
Link: https://lore.kernel.org/r/20240308221800.48860-1-rodrigo.vivi@intel.com


Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>

On musl there is already defined PAGE_SIZE and it is casing
compilation error:

../tests/kms_atomic.c: In function 'crtc_invalid_params_fence'
../tests/kms_atomic.c:893:25: error: expected identifier or '(' before numeric constant
  893 |         const ptrdiff_t PAGE_SIZE = sysconf(_SC_PAGE_SIZE);

Change this variable name in order to fix it.

Link: #138


Signed-off-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

Such change allows using the code on some other python script that
would be converting IGT doc test documentation into XLS files.

Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Acked-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>
Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>

Instead of only allowing to load the from the module name, also allow it
from alias by using kmod_module_new_from_lookup(). With that a user can
have their on aliases and the kernel can also export a compat alias for
future module renames.

There was a bug in libkmod <= 29 that would make
kmod_module_new_from_lookup() return prematurely after a failed alias
lookup when we don't have a modules.alias file in the modules dir. So
add a workaround to check for -ENOSYS return code too. It should still
be compatible with what we are doing here.

v2: Fallback to kmod_module_new_from_name() since the kmod ctx is also
    used on builder machines to determine the test lists.
v3: Add workaround to bug in libkmod
v4: Remove unneeded braces

Reviewed-by: Kamil Konieczny <kamil.konieczny@linux.intel.com>
Link: https://lore.kernel.org/r/20240308193149.2220807-1-lucas.demarchi@intel.com


Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>

Currently, some parmeters like Run Type are imported from files.
Don't add documentation for them by default, preserving the way
it was originally meant to be filled.

Acked-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

There are two functions there that aren't unused.

Drop the dead code.

Acked-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

There were some changes at test_list that caused it to stop
working. Fix them.

Acked-by: Katarzyna Piecielska <katarzyna.piecielska@intel.com>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>

Fill the gap with testing render-copy for xe driver. Currently only
few basic render-copy checks are executing within xe_intel_bb@render.

xe_render_copy introduces new checks comparing to gem_render_copy
test. During my work on enabling render-copy on LunarLake I've
used same r6 subregisters like on previous platforms what causes
full render-copy between surfaces worked but partials were not.

I decided to include square and v/h stripes tests to verify cofficients
for barycentric coordinates matches expected <x,y> position. More
stressful is random subtest which randomizes width and height as well
as division point (it copies four rectangles to destination surface).

Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20240306150219.301977-5-zbigniew.kempczynski@intel.com


Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>

Refactor a bit handling different tilings in render-copy pipeline
and add Tile64 which I want to exercise in xe_render_copy test.

Cc: Matthew Auld <matthew.auld@intel.com>
Acked-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20240306150219.301977-4-zbigniew.kempczynski@intel.com


Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>

Dumping to png where we want to pass width, height and/or other
values is much easier when we use printf style filename formatting
instead preparing string with filename separately. Replace current
 write to raw png from single string to string formatting.

Cc: Matthew Auld <matthew.auld@intel.com>
Reviewed-by: Matthew Auld <matthew.auld@intel.com>
Link: https://lore.kernel.org/r/20240306150219.301977-3-zbigniew.kempczynski@intel.com


Signed-off-by: Zbigniew Kempczyński <zbigniew.kempczynski@intel.com>