We might as well verify that we have a semblance of all being in order
by making sure the DCO frequency is within the expected bounds.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
I had various problems (infinite loops, unable to compute dividers for
certain frequencies) after implementing a BSpec update. Much easier to
debug that in userspace.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Move function CPU mmap test of large bo to gem_mmap, and include a
page-by-page copy between two huge objects (as we have had many bugs
triggering pagefault-of-doom for full apertures before).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
For all those IGT tests that need an easy way to draw rectangles on
buffers using different methods. Current planned users: FBC and PSR
CRC tests.
There is also a tests/kms_draw_crc program to check if the library is
sane.
v2: - Move the test from lib/tests to tests/ (Daniel).
- Add igt_require() to filter out the swizzling/tiling methods we
don't support (Daniel).
- Simplify reloc handling on the BLT case (Daniel).
- Document enum igt_draw_method (Daniel).
- Document igt_draw_get_method_name() (Paulo).
v3: - Add IGT_DRAW_MMAP_WC (Chris).
- Implement the other trivial swizzling methods (Chris).
- Remove the gem_sync() calls (Chris).
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Commit 47f6b1305cc3752f318a555b932e194e1500c1d8 completely broke this
test due to the fread() assertion. When we're reading the debugfs file
we really don't care about how many bytes we read because the number
is not constant and we just use strstr() later. Change the assertion
to make it check for at least 1 byte read, just to make sure no one
changes that again.
Regression introduced by:
commit 47f6b1305cc3752f318a555b932e194e1500c1d8
Author: Thomas Wood <thomas.wood@intel.com>
Date: Wed Mar 25 16:42:57 2015 +0000
igt.cocci: check the return values of various functions
Cc: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Using imported objects should not leak i915 vmas (and vms).
In practice this simulates Xorg importing fbcon and leaking (or not) one vma
per Xorg startup cycle.
v2: use low-level ioctl wrappers and bo offset to check the leak (Chris)
v3: use the flinked bo as batch (Chris)
v4: add check on offset, remove unneeded assignments (Chris)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniele Ceraolo Spurio <daniele.ceraolospurio@intel.com> (v2+)
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
While it is nice to have shorter names for the most-accessed
variables, it makes the code more difficult to read since it's not
clear to the code reader whether that "gem_handle" is from some FB or
something else. The reader also has to audit the code to see if, for
example, the value of data->handle[0] stays consistent with
data->fb[0].gem_handle all the tame or if at some point the value is
replaced with something else. So remove the redundant information,
making it explicit that we're using the gem handles and FB IDs of the
framebuffers all the time.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
The code has a common pattern of "wait 300ms, then check if FBC is
enabled". Most of the time FBC is enabled in either 50ms or 0ms, so
introduce wait_for_fbc_enabled(), which can return much earlier if FBC
is actually enabled before the 300ms timeout.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Just a little helper for code that needs to wait for a certain
condition to happen. It has the nice advantage that it can survive the
signal helper.
Despite the callers added in this patch, there is another that will go
in a separate patch, and another in a new IGT test file that I plan to
push later.
v2: Check COND again before returning in case we hit the timeout.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Global variable names should reflect the fact that they are indeed
global, and at the very least they should not be as short as just
"mmio". Rename mmio to igt_global_mmio.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
igfx_get_mmio() uses the global mmio variable by accident. Use a local
variable instead.
The intention is to rename the global variable later on, so shadowing it
here does not matter.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
This adds tests/drm_auth.c which tests for drmGetMagic() and
drmAuthMagic() deficiencies.
Signed-off-by: David Herrmann <dh.herrmann@gmail.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Exercise lite-restore (re-submit a context that is currently running),
by queueing several small batchbuffers.
This test helps to validate WaIdleLiteRestore.
Signed-off-by: Michel Thierry <michel.thierry@intel.com>
The call to low_mem_killer_disable(true) was being done
from within function oom_adjust_for_doom. However,
oom_adjust_for_doom gets called from 3 places. We only
want the call to low_mem_killer_disable(true) to happen
during common_init, so call it from here instead of from
oom_adjust_for_doom.
v2:Thomas Wood pointed out that the initial call to disable
the low_mem_killer does not get made when we are just
listing subtests; so I have qualified the call from the
exit handler, which re-enables the low_mem_killer, with
if (!igt_only_list_subtests()).
For belt and braces I have also made low_mem_killer_disable
idempotent, so multiple calls to disable or re-enable are
safe.
Signed-off-by: Tim Gore <tim.gore@intel.com>
[Thomas: small coding style fix]
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Test used to call prepare_crtc twice in the plane loop and leaked two
framebuffers per [subtest]x[pipe]x[plane].
What the loops really wants to do, instead of second invocation of
prepare_crtc, is to just turn on the display with the unrotated fb to
verify that the plane property has been restored by the VT transition
from previous to graphics mode.
To enable that factor out code which does that from prepare_crtc into
commit_crtc and call it instead.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Produce the intel_reg man page from rst using rst2man. Also facilitate
writing any man page in reStructured text, as long as rst2man is
available.
v2: configure check for rst2man, credits to Thomas Wood for that.
Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Rendering into Y and Yf tiled frame buffers with Cairo was losing the
previous content ie. was starting from black. This is different than the
behaviour with linear and X tiled so make it the same by blitting the
initial content when creating the rendering context.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
On gen8, we should check that the full 64bit relocation value is
correct, and we should be sure to poison the relocation offset between
runs.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Now that there is PAGE_SIZE define, use it.
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Saves a good amount of code duplication by supporting expected
failures from the main loop.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Now that size is calculated in a single place and correct geometry passed in,
paint squares does not need to concern itself with it.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
There can only be one, either a plane or a cursor, in each subtest so there
is no need for two framebuffer varilables and also some codepaths can be
unified.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
It is just there to light up the display using the full modeset. Also renamed it
from fb_full to fb_modeset to be more descriptive.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Sonika Jindal <sonika.jindal@intel.com>
Reviewed-by: Sonika Jindal <sonika.jindal@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
We frequently check for device capabilities, for which we can safely
assume that there is but one on a system and so cache the first query
value and return it for all future queries. The benefit is to reduce
dmesg debug spam which helps when either bringing up a test or trying to
track down why a test fails.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
