Debian Busters is the new stable so let's upgrade.

v2: stay on stretch for MIPS

Cc: Guillaume Tucker <guillaume.tucker@collabora.com>
Cc: Petri Latvala <petri.latvala@intel.com>
Issue: https://bugs.debian.org/932725


Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

Prior to this commit, audio tests play sound on all HDMI/DisplayPort PCM
devices (by filtering devices with the "HDMI" prefix). This had several issues.
First, the PCM device → connector mapping via the ELD wasn't checked. Second,
it makes debugging more complicated: if ALSA doesn't detect the screen, we just
get a Chamelium failure (because it doesn't receive any audio stream). Third,
it would play sounds on real screens too (not only Chamelium) and this could
theorically damage bad hardware (though this is just speculation).

To mitigate this issue, check the monitor name of PCM devices we open and only
keep them if it's an IGT EDID (ie. connected to the Chamelium board).

Note that while this is a good improvement, it isn't a complete fix: the ELD
is only exposed by the HDA Intel driver. This means non-IGT PCM devices won't
be filtered out on non-Intel hardware. There is no driver-agnostic way to get
the screen name, I'm afraid.

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

This function parses the monitor name from an ELD blob and checks whether it
comes from an IGT EDID.

v2: replace a bunch of random numbers with constants (Chris, Andi)
v3: rabase and fix subject (Arek)

Signed-off-by: Simon Ser <simon.ser@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Andi Shyti <andi.shyti@intel.com>
Cc: Martin Peres <martin.peres@linux.intel.com>
Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>

We test so many audio configurations that igt_debug's ring buffer gets
overflown which often puts the actual failure out of our reach.

Let's promote failure-indicating messages to critical log level, so we
see which exact configuration has borked.

Cc: Martin Peres <martin.peres@linux.intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=111696


Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Martin Peres <martin.peres@linux.intel.com>

Once the fence is retired, we discard its references to the context
state, including its timeline name. So the timeline query is only
accurate before being signaled, so keep the fence alive until after the
query.

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

This igt relies on using privileged batches to read and modify protected
registers. This is no longer supported and so the test is entirely
defunct, and for as much as we can has been subsumed into a kernel
selftest.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Martin Peres <martin.peres@linux.intel.com>
Acked-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>

The implicit soft-pinning we use to probe the vm layout using execbuf,
depends on the batch remaining active (not retired) between execbufs.
Naturally, if the background retire worker runs the batch is retired and
the implicit soft-pinning is free to use a fresh address.

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

Chuansheng Liu authored 5 years ago and Chris Wilson committed 5 years ago

Reference:
https://bugs.freedesktop.org/show_bug.cgi?id=112126



The issue we hit is the GPU keeps very high load after running
the subtest min-max-config-loaded.

Some background of the issue:
Currently the rps is not fully enabled yet on TGL, and running
the subtest min-max-config-loaded will hit below assertion:
==
(i915_pm_rps:1261) CRITICAL: Test assertion failure function loaded_check, file ../tests/i915/i915_pm_rps.c:505:
(i915_pm_rps:1261) CRITICAL: Failed assertion: freqs[MAX] <= freqs[CUR]
(i915_pm_rps:1261) CRITICAL: Last errno: 2, No such file or directory
==

with igt stress test, we find the GT keeps busy after running
this subtest, it is due to the igt_spin_end() is not called
randomly.

The root cause analysis is:
When the main process i915_pm_rps for running the subtest
min-max-config-loaded hits the assertion, the main process will
try to send signal SIGTERM to the child process loader_helper
which is created by main process for starting GT load, then the
main process itself will exit.

The SIGTERM handler for loader_helper is the default one, which
will cause the loader_helper exits directly. That is unsafe, we
always expect the igt_spin_end() is called before loader_helper
process exits, which is used to stop the load of GT.

Furthermore, in normal scenario, before main process exits,
it will send SIGUSR1 to child process for stopping GT loading
in safe way.

So here we install the proper handler for signal SIGTERM in the
similar way. Without this patch, the GT may keep busy after
running this subtest. Enabling rps should be tracked on the
other side.

V3: As suggested by Chris, s/SIGUSR1/SIGTERM, since both signals
take the same function.

Signed-off-by: Chuansheng Liu <chuansheng.liu@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Lionel Landwerlin authored 5 years ago and Chris Wilson committed 5 years ago

v2: Fix mistake in syncobj_busy() (Lionel)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Lionel Landwerlin authored 5 years ago and Chris Wilson committed 5 years ago

Drop local defines etc..

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Lionel Landwerlin authored 5 years ago and Chris Wilson committed 5 years ago

We have copies of the DRM uAPI headers in the repo, so drop the local
declaration of syncobj defines/types.

v2: s/drmIoctl/igt_ioctl/ (Chris)
    Clear errno after report it through return value (Chris)

Signed-off-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Fixes: 3fa72891 ("i915/gem_ctx_persistence: Double the fput hammer!")
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Mika Kuoppala <mika.kuoppala@linux.intel.com>

Tvrtko Ursulin authored 5 years ago and Chris Wilson committed 5 years ago

Check that the engine query is not polluting the buffer past the size it
indicated it would write.

Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>

Abdiel Janulgue authored 5 years ago and Chris Wilson committed 5 years ago

If we accidentally modify the mmap_gtt ioctl struct sizes in our headers,
we might end up feeding garbage into the ioctl and start validating
that as our flags for the upcoming mmap_offset ioctl. This ensures header
changes don't mess up with our existing mmap_gtt ioctl.

Signed-off-by: Abdiel Janulgue <abdiel.janulgue@linux.intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>

Deferred rcu work is tricky to pin down and encourage to run, so try
again...

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=112277


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

Venkata Sandeep Dhanalakota authored 5 years ago and Chris Wilson committed 5 years ago

Make invalid context subtest independent of engines, as the intent
of the subtest is to verify if using an invalid context returns
an error.

Cc: Andi Shyti <andi.shyti@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Venkata Sandeep Dhanalakota <venkata.s.dhanalakota@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>

GEN12 CCS render compression support is missing from the test.
This causes a SKIP when executing a subtest 'getfb-reject-ccs'.

For TGL the main surface is 4x4 tiles aligned. Therefore, the
pitch for 32bpp is 4*4*32 bytes and the height is 4 rows aligned.

The CCS surface is 64 bytes, which corresponds to 4 tiles on the
main surface. The height of the CCS surface is aligned by 4 rows.

v2: Fix incorrect size and updates on comments (Imre)

Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>

On TGL till B0 stepping FF_MODE2(0x6604) register is not readable. But
we need to program this register for Wa_1604555607 for better
performance. Hence to skip the verification attempt on this register
through read, we are marking this register as write only.

As of now I have marked this register as write only for gen12, we can
narrow it down to TGL <= B0, if we can retrieve the stepping of the
platform at IGT.

v2:
  comma is added [Chris]

Signed-off-by: Ramalingam C <ramalingam.c@intel.com>
cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>

Debian has two Dockerfiles associated - minimal and regular one that
bases on the minimal one.

Let's mention them by name and add a comment providing some context to
the split so it's easier to figure out what to install on a new system.

Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

Since engines are now looked up based on context engine map,
context should be setup with engines before using in execbuf.

Fixes: a9998c23 igt("gem_exec_parallel: use the gem_engine_topology")

Cc: Andi Shyti <andi.shyti@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@linux.intel.com>
Signed-off-by: Venkata Sandeep Dhanalakota <venkata.s.dhanalakota@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

* chamelium was renamed to kms_chamelium and is now part of the KMS
   Tests section
 * GVT tests were removed.

v2: don't remove the API docs (Petri)

Fixes: 8809638e ("tests: Rename chamelium to kms_chamelium.")
Fixes: 8781fd89 ("igt: Remove gvt_basic")
Reported-by: Petri Latvala <petri.latvala@intel.com>
Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

Not used since 8781fd89 ("igt: Remove gvt_basic").

Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

The final naming for the magic block identifiers is:

igt_subtest_with_dynamic("do-stuff-on-things") {
    for_each_thing(x) {
        igt_dynamic_f("%s", x->name) {
	    do_stuff(x);
	}
    }
}

Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Cc: Katarzyna Dec <katarzyna.dec@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Acked-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

Turns out that `readlink()` does not NULL-terminate the string and since
it resides on the stack we may end up with some junk:

  `/proc/self/exe point to /opt/igt/libexec/igt-gpu-tools/tools_test7<CD>\t<8A>^?`

That in turn confuses `dirname()` and we end up doing `chdir("/")`,
which explain the sporadic failures of this test where it was not able
to locate the tools.

Let's zero out the variable first and allow `readlink()` to write at
most `sizeof()-1` bytes to it, so it is always properly terminated.

Cc: Petri Latvala <petri.latvala@intel.com>
Cc: Tomi Sarvela <tomi.p.sarvela@intel.com>
Issue: #12


Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>

i915_pm_dc have implemented a function to check if PSR2 is enabled,
that may be util to future tests so move it to PSR lib.

Cc: Jeevan B <jeevan.b@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>

has_runtime_pm was not used outside of igt_fixture also no need to
print this info, if not supported IGT skip log will contain the line
that was not meet.

Cc: Jeevan B <jeevan.b@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>

Also moving the sink check to igt_fixture as it do not change
overtime.

Cc: Jeevan B <jeevan.b@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>

Writing 0xf to i915_edp_psr_debug is expected to have -EINVAL
returned in newer kernels but this error actually comes from errno
not from the return of write()(check writeN()), so unseting the
expected errno so this do not cause tests to skips.

Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=112257


Cc: Jeevan B <jeevan.b@intel.com>
Cc: Anshuman Gupta <anshuman.gupta@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>

They have lived past their usefulness as now we frequently lose vital
debug information because we are spamming unrelated GEM traces. We can
add the spam directly to GEM_BUG_ON(), and enable it ourselves if we
truly need the trace-on-oops.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Martin Peres <martin.peres@linux.intel.com>
Cc: Petri Latvala <petri.latvala@intel.com>
Cc: Tomi Sarvela <tomi.p.sarvela@intel.com>
Acked-by: Petri Latvala <petri.latvala@intel.com>

Mika Kahola authored 5 years ago and Imre Deak committed 5 years ago

Add GEN12 CCS format modifiers for render compression. Render
compression is not supported for 90/270 rotation.

v2: Added new subtests for each format modifier.
v3: Add IGT description for each subtest (GitLab)
v4: Fix errorneous subtest execution and listing
v5: Fix AUX surface units for GEN12 (Imre)

Bspec: 49252

Signed-off-by: Mika Kahola <mika.kahola@intel.com>
[Keep existing subtest names.]
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>

Mika Kahola authored 5 years ago and Imre Deak committed 5 years ago

Update tile sizes for GEN12 CCS.

BSpec: 47709

v2: Realign main and aux surfaces (Imre)

Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Imre Deak <imre.deak@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>

Mika Kahola authored 5 years ago and Imre Deak committed 5 years ago

Gen-12 has a new compression format for render compression.
For this, a new modifier is needed to indicate that.

Signed-off-by: Mika Kahola <mika.kahola@intel.com>
Reviewed-by: Radhakrishna Sripada <radhakrishna.sripada@intel.com>
Signed-off-by: Imre Deak <imre.deak@intel.com>

Our worst case is blocking on a infinite batch, as the hangcheck allows
it to remain in place indefinitely. This is too slow for CI as it only
fails via a lengthy timeout. However, for our purposes a fail due to
hangcheck is an equally good indicator of the bug, so use a hostile
spinner to provoke hangcheck.

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

Since the buffer is always valid up to the next page, we can just round
up to the oword boundary.

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

Although a virtual engine itself has no hang detection; that is on the
underlying physical engines, it does provide a unique means for clients
to try and break the system. Try and break it before they do.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>

Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

If binary 'bin' has a subtest 'sub', which has dynamic subtests 'foo'
and 'bar', results.json will now have "subtests" by the names
igt@bin@sub@foo and igt@bin@sub@bar, with data as expected of normal
subtests.

Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

v2: Also test that calling igt_describe in a dynamic context asserts

Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>

Dynamic subsubtests, or subtests of subtests, are individual pieces of
tests that are not statically available all the time.

A good example of a need for a dynamic subsubtest is i915 engine
listing: A normal subtest for each engine class ("bsd"), and a dynamic
subsubtest for each instance ("bsd0", "bsd2", etc). Or a normal
subtest for an operation with a dynamic subsubtest for every engine
there is.

Another example is dynamic subsubtests for pipes: Instead of using
foreach_pipe_static, make one subtest and use foreach_pipe with
dynamic subsubtests for each pipe.

v2: Rebase and adapt to igt_describe changes

v3: Rename to igt_subtest_with_dynamic_subsubtests &
    igt_dynamic_subsubtest, better docs, make igt_describe fail loudly
    if it's used in an impossible context.

v4: Mention dynamic blocks in the warning for longjmp in core docs.

Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>