The command parser's batch_len optimization causes the parser to
reject this batch as not having an MI_BATCH_BUFFER_END because
the length was not set correctly.
Signed-off-by: Brad Volkin <bradley.d.volkin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
The size of the batch buffer passed to the kernel is significantly
larger than the size of the batch buffer passed to the function. A
proposed optimization as part of the batch copy kernel series is to
use batch_len for the copy and parse operations, which leads to a
false "batch without MI_BATCH_BUFFER_END" failure for this test.
To fix this, modify the test to set batch_start_offset and batch_len
such that they define the range of actual commands in the batch,
including a few of the surrounding nops for alignment purposes.
v2: update batch_start_offset as well
Signed-off-by: Brad Volkin <bradley.d.volkin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
TESTS_scripts was accidentally removed from EXTRA_DIST in commit 685e577
(Move library selftests to lib/tests).
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
to see if one can fool hangcheck by keeping non hanging
ring busy
v2: break from loop instead of return (Daniel Vetter)
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
The optopt variable is not set if an invalid long option is used, so
check the current option character instead.
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
This test has a few checks that batch buffer addresses in the error
state match the expected address for the userspace supplied batch.
But the batch buffer copy piece of the command parser means that
the logged addresses are actually _supposed_ to be different. So
skip just those checks.
Signed-off-by: Brad Volkin <bradley.d.volkin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
commit c45216b26d223ffd7f68c5794870214445a64b5c
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Tue Nov 4 07:30:57 2014 +0000
igt/gem_tiled_wb: Exercise CPU mmaps with swizzling
brought along a few interlopers.
This exercises the the extended get_tiling ioctl in order to determine
proper swizzling for direct access to objects through WB.
Userspace cannot handle bit17 swizzling through wc mmaps (because bit17
requires swizzling based on the actual physical address of the page -
which is unknown to userspace) and so we need an extended get_tiling
ioctl to report the actual as well as the logical swizzling on an
object.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Fix distcheck issues introduced by commit 685e577 (Move library
selftests to lib/tests).
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
This seems to have been accidentally disabled in
commit 982f7eb238a0898c456e0574dee7c4507738d75f
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date: Fri Aug 29 15:19:57 2014 +0100
Prepare for 64bit relocation addresses
Apparently no one noticed.
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
This is a regression from
4306538d1d3f60877866c39c9ca953cc5e541dae is the first bad commit
commit 4306538d1d3f60877866c39c9ca953cc5e541dae
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
AuthorDate: Thu Oct 2 11:18:20 2014 +0200
Commit: Daniel Vetter <daniel.vetter@ffwll.ch>
CommitDate: Thu Oct 2 11:34:55 2014 +0200
tests: Sprinkle missing igt_exit() where needed.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=85582
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Again they're not really igt testcases so are in the way of
running spatch unconditionally. Move them someplace else.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
They're now igt tests, and so if you blindly run lib/igt.cocci with
spatch on tests/*c they get mangled. Move them away, but still keep
them as noinst targets.
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
We were having a problem where the system-suspend test was passing,
but then the next test - gem-execbuf-stress - was failing because of
bugs caused by the suspend subtest. So add a single test that emulates
the same problem, and another test that just sets a mode after resume.
This way, we should be reproducing the problem even if you don't run
the tests in the order they are defined.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Make sure DSPSURF will change during the panning operation
in flip-vs-panning-vs-hang.
This will now test agains bugs between the kernel's mmio vs.
CS flip race handling and GPU resets. If the kernel is buggy
if will fail to notice that the panning operation changed the
base address before the GPU reset had a chance to deal with the
pending page flips, and so the flip would never complete due to
DSPSURFLIVE not matching the expected value.
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Driver now returns the constant values that were set once
on ring initialization time. Extend the simple interface
check to all gens (i915_wa_registers needs to be present).
Due to delay when render context is visible on cpu side,
wait gpu to execute something on default context to get rid
of undeterministic behaviour by sometimes getting all ones.
Cc: Arun Siluvery <arun.siluvery@linux.intel.com>
Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
Without LLC, we cannot assume coherency between system RAM and the GTT.
If we try to use the GTT with snooped memory, the machine may lock up,
so the kernel prevents us from doing so, causing the test to fail.
However, we still do want to verify that we can access the userptr
through a dmabuf, so just check that we can make a copy from it using
the GPU and ignore the GTT cross-checking.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=85354
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If we don't enable audio runtime PM, the audio driver won't release
its reference, the refcount won't ever become zero, so we will never
actually runtime suspend. So move this code from pm_rpm.c to
igt_aux.c, so kms_flip - and any other IGT test case using RPM - can
benefit from it.
Previously, if you ran pm_rpm before running the other tests - or if
you just didn't have snd_hda_intel loaded - you wouldn't notice this
bug.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78893
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
libva makes extensive use of chained batch buffers. The batch
buffer copy portion of the command parser has the potential to
break chained batches, so add a simple test to make sure that
doesn't happen.
Signed-off-by: Brad Volkin <bradley.d.volkin@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
When I look at cursor_size_ok() (from the Kernel's intel_display.c), I
see that only 845g and i865g support non-square cursors, so SKIP the
tests on HSW/BDW instead of failing them.
This problem happened because support for non-square cursors was being
developed for HSW/BDW, but it ended up not being merged due to a small
problem and priority changes. Let's make those tests SKIP until we
have proper support on the Kernel: after that, we can revert this
patch.
v2: Improve commit message.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84268
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Compare the number of modes available when the edid is reset with the
number before the edid was overridden.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=82230
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Ville suggested that we should use the same semantics as C arrays to
reduce the number of those pesky +1/-1 in the allocation code.
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
v2: rename "connector" to "vga_connector" to improve the skip message
(Chris Wilson)
Suggested-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Since commit 5782eca (lib/igt_core.c: disable lowmemorykiller during
tests), igt_exit needs to be called before the test exits.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84771
Cc: Tim Gore <tim.gore@intel.com>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
sysfs interface for rc6 has been changed to only expose deep RC6 file to the platorms
that really support them, aka SNB and IVB.
So let's split this test and only test RC6p and RC6pp on these platforms.
Reference: https://bugs.freedesktop.org/show_bug.cgi?id=84886
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
A few weeks ago we were hitting this assertion, and the _eq macro is
more useful when you're debugging.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
warning: ignoring return value of ‘write’, declared with attribute
warn_unused_result [-Wunused-result]
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
With the current Kernel, it takes more than 10 minutes to run
modeset-stress-extra-wait on BYT with eDP, and we don't really want
IGT subtests that take more than 10 minutes to run: QA reports them as
bugs.
Since, as far as I remember, any of these RPM race condition bugs we
had could always be triggered with just 4-5 rounds of testing, let's
tune down the number of rounds we test:
Before:
rounds = (--quick option used) ? 10 : 50;
After:
rounds = (--quick option used) ? 10 : 40;
rounds = (wait_flags & WAIT_EXTRA) ? rounds / 2 : rounds;
This should make all subtests be way above the 10m limit.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=84897
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
This reverts
commit d92fbc23138b1014e8574daf29dbb06b8c81aa7a.
Author: Rodrigo Vivi <rodrigo.vivi@intel.com>
Date: Tue Sep 9 12:25:47 2014 -0400
tests/kms_sink_crc_basic: Wait 2 vblanks before grabing the new crc.
Sink CRC is fixed on kernel to wait as many vblanks as needed. It was fixed b
commit ad9dc91b6e21266bfc6f466db4b95e10211f31ee
Author: Rodrigo Vivi <rodrigo.vivi@gmail.com>
Date: Tue Sep 16 19:18:12 2014 -0400
drm/i915: Fix Sink CRC
This this fix in place we don't need this extra 2 vblanks on test case itself
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
