igt_enable_connectors forces connectors to be enabled where doing so is
known to work well. igt_reset_connectors resets the force state on all
connectors.
Add a function to set an EDID data block on a connector and include a
set of generic EDID blocks for testing.
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Add an API function and a test program to force a particular state on a
connector.
v2: mask the correct part of the minor number to get the card number (Chris
Wilson)
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
If the parent passes a userptr to some private memory, we expect to
still be able to use the userptr in the child.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
The forked tests allocate the bo (and thus for userptr, the memory) in
the parent and pass them to all children. The difference for userptr is
that we allocate system memory which the kernel then copies into each
child. As the children need to access the memory for their checks, it
does need to be shared - so allocate the userptr from shared memory!
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=80208
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
When waiting for the forked tests, we can respond quicker to a failure
(such as oom) by waiting for any child to exit rather than waiting for
each child in order. Then when we see that a test failed, we can kill
all other children before aborting.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
New stuff caught.
Plus manually simplify the massive igt_fail_on_f(file == NULL, ...) to
a simple igt_assert(file). We already print the errno (if
applicapable) and the condition, which is equally informative.
Cc: Yi Sun <yi.sun@intel.com>
Cc: Matt Roper <matthew.d.roper@intel.com>
Cc: Wendy Wang <wendy.wang@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We want to unbind fbcon, but only fbcon and only if it's there.
This was broken by the recent patch in 3.16-rc to kick out the vgacon
driver. I've forgotten to push out the relevant fix from the machine
used to create the kick vgacon patches.
v2: Dropped duplicated /, spotted by Damien.
Reported-by: Damien Lespiau <damien.lespiau@intel.com>
Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
We weren't returning straight away when failing to unload the driver, so
the test happilly executed gem_suspend and printed ""module successfully
loaded again".
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Just like the it was done for the requirement message, display the errno
message only if errno is set, and display it at the end of the assert
message.
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
The errno message was a bit in the middle here, it makes more sense to
group the messages about why the test requirement wasn't met together.
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
... since it uses flink. Fixes a regression due to:
commit 6d6dfcfb883818b40b58bac61cc72cab428a7a03
Author: David Herrmann <dh.herrmann@gmail.com>
AuthorDate: Sun Mar 16 14:38:40 2014 +0100
drm: enable render-nodes by default
Cc: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
kms_univeral_plane is a new test that requires cairo.
I have added it to the list of tests not to build on
Android unless ANDROID_HAS_CAIRO is set.
Signed-off-by: Tim Gore <tim.gore@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When changing the pipe we were using, test_grab_crc() wasn't correctly
setting the pipe constraint before waiting for the CRC on the pipe, and
so we ended up waiting for a CRC on a pipe that wasn't lit up.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Get CRCs of a full red and a full blue surface as reference.
Create a big framebuffer that is twice width and twice height as the
current display mode.
Fill the top left quarter with red, bottom right quarter with blue
Check the scanned out image with the CRTC at position (0, 0) of the
framebuffer and it should be the same CRC as the full red fb
Check the scanned out image with the CRTC at position (hdisplay,
vdisplay) and it should be the same CRC as the full blue fb
v2: Fix a few things here and there (Damien)
Cc: Lei Liu <lei.a.liu@intel.com>
Cc: Yi Sun <yi.sun@intel.com>
Signed-off-by: Lei Liu <lei.a.liu@intel.com>
Signed-off-by: Yi Sun <yi.sun@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
We're going to add tests for panning. Both position and panning tests
share the need to grab reference CRCs of single coloured fb, so let's
separate this function from the position test.
Cc: Lei Liu <lei.a.liu@intel.com>
Cc: Yi Sun <yi.sun@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
This is what we read when the CRC logic in in a powered down well. We
really don't want that to happen.
In theory, it's possible 0xffffffff to be a valid CRC value, so I don't
assert here.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
So the next commit won't trigger a drmModeSetCrtc() if the primary plane
doesn't have any update needing it.
This shouldn't be a problem at the moment as we don't allow the primary
plane to be of a different size than the CRTC viewport, but it will most
likely change in the future and we don't want to have that bug there.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Let's make the test a bit more generic and have the number of CRCs we're
collecting a define so it can be changed easily.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Instead of just testing if the CRCs are stable, we also test 2 different
fbs to make sure that the CRC is actually changing.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Let's put the per-output test in its own function to get rid of 1 level
of indentation. We'll need it to cycle through 2 different framebuffers
to make sure we compute different CRCs if the fbs are different.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
There hasn't been a libdrm release containing the universal plane
definitions yet, so add them to igt_kms to allow compilation to succeed
in the meantime.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
- Drop the flip from the name - we don't do that. And the blt is
really just to have a bit of fun with the domain tracking.
- The real test is 1) dirty with gpu 2) grab 1st crc 3) set_domain for
cpu access 4) wait a long time 5) dirty more with cpu 6) grab 2nd
crc.
This fixes failures since with the old tests we wouldn't have noticed
the cpu rendering really.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Doing otherwise breaks listing the subtests. The test was throwing an
error out when universal planes were disabled as well because of that.
Cc: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
We're calling drmModeSetCursor() to change the cursor image and never
actually doing a display commit (aside from when we display the cursor),
so call the move ioctl directly rather than igt_plane_set_position() to
ensure the changes actually take effect.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
If a subtest fails, cleanup_crtc() never gets called and then the
test_data_t structure for the test is lost, including the CRC file
descriptor that we never got a chance to release; this causes all
subsequent tests to fail with -EBUSY at igt_pipe_crc_new().
The split between permanent data_t and temporary test_data_t doesn't
seem to serve a purpose, so just combine the fields from both into
data_t. This will prevent us from losing the CRC filedescriptor so that
we can properly close and reopen it after a failed test.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Add a simple test to exercise universal plane support.
v6:
- Update to new universal plane interface (commit parameter rather than
state-changing function). It should now be a lot more explicit which
steps are being taken with legacy API's vs universal API's now.
v5:
- Check that we don't have more than one primary or cursor. This will
catch accidental calls to drm_plane_init() in the kernel where
drm_universal_plane_init() was intended (these don't cause a compile
warning due to type compatibility between enum and bool).
v4:
- Test disabling the primary plane explicitly when it has previously
been implicitly disabled due to clipping.
- Skip test if igt_pipe_crc_new() fails
v3:
- For testing while crtc is off, switch between several different
primary plane fb's before reenabling the crtc. This will help
catch pin/unpin mistakes.
v2:
- Test that pageflips error out gracefully when the primary plane
is disabled before the ioctl, or between ioctl and pageflip
execution.
- Test that nothing blows up if we try to disable the primary plane
immediately after a pageflip (presumably before the pageflip actually
completes).
- Test that we can setup primary + sprite planes with the CRTC off and
then have them show up properly when we enable the CRTC
(drmModeSetCrtc with fb = -1).
- Test that we can modeset properly after having disabled the primary
plane
- Test that proper error codes are returned for invalid plane
programming attempts.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
None of our hardware can support this today, but we'd like to be able to
write tests that check that the kernel returns the proper error code
when userspace tries it anyway.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Add support for universal planes. This involves revamping the existing
plane handling a bit to allow primary & cursor planes to come from the
DRM plane list, rather than always being manually added.
v2: Don't drop fixed ordering of internal plane list. Primary will
always be index 0, cursor will always be last in internal plane
list.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Add a new public API that will attempt a display commit, but will return
an error code upon failure rather than failing the IGT test. This is
intended to allow igt tests to verify that the expected error codes are
returned to userspace when invalid requests are issued.
Note that with non-atomic programming, a single commit operation has
several potential failure points internally; the first non-zero error
code encountered will be returned immediately and no further programming
will be performed by the commit. It is the caller's responsibility to
return to a sane state.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Add a new commit interface, igt_display_commit2(), that allows tests to
specify which programming API should be used to perform hardware
updates. COMMIT_LEGACY is the only option for now, but universal
and atomic interfaces will be added as additional options in the future.
igt_display_commit() remains unchanged for existing tests that wish to
place the hardware in a specific state, but that don't care which API is
used to achieve that state. The legacy API will be used by default for
now, but in the future we may decide to make the default API
configurable via an environment variable.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
The need to wait for a vblank after programming is due to the way we
actually program the hardware. Move need_wait_for_vblank out of the
pipe and into a local variable in preparation for future programming
styles (e.g., atomic pageflip) that will need different logic.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
The "need" flags on igt_pipe simply mirror the fb_changed field of the
primary/cursor planes. Drop them and just use fb_changed instead.
Signed-off-by: Matt Roper <matthew.d.roper@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
The command parser may be present, but not active, so check for PPGTT
before allowing this test to run.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Add a subtest to make sure the kernel doesn't leak the vma
pin_count for the last context on reset.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
A bug was recently introduced into the kernel that happened when the vma
was smaller than the object. Test that.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
igt-core.h/c provides some macros and initialisation
functions to support the tests but some of the single
tests do not use these. Modifying these tests to use
the igt_simple_main macro and igt_simple_init function
is the first step towards a consistent command line
across all tests.
Signed-off-by: Tim Gore <tim.gore@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
We need this to avoid hitting the slowpath and ending up with a
presumed_offset == -1. Regression reported by PRTS, bisected to
commit eb36fc993d7ae1988c80ba5b767989059c91d0ec
Author: Chris Wilson <chris@chris-wilson.co.uk>
AuthorDate: Mon Jun 16 10:49:16 2014 +0100
Commit: Chris Wilson <chris@chris-wilson.co.uk>
CommitDate: Mon Jun 16 10:51:02 2014 +0100
igt/gem_exec_big: Update to new igt_assert_eq
Use igt_assert_eq for better test output on failures.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
v2: igt_warn_on unexpected reloc offsets.
Cc: shuang.he@intel.com
Acked-by: Chris Wilson <chris@chris-wilson.co.uk> (on irc)
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
This reinstates the #includes for lib/debug.h since without that the
shader debugger fails to build. You need to manually enable it with
./autogen.sh --enable-shader-debugger and the fail only happens when
running a clean build with
$ git clean -dfx && ./autogen.sh --enable-shader-debugger && make
I have no idea what breaks, but this restores the build.
Reported by Ben Widawsky.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Actually just whitespace change to make sure the new built rules for
tests/*.txt pick up the changes properly everywhere.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
