platform/upstream/intel-gpu-tools.git
10 years agokms_rotation_crc: Remove useless comments
Damien Lespiau [Tue, 8 Jul 2014 12:59:42 +0000 (13:59 +0100)]
kms_rotation_crc: Remove useless comments

A typical example of what comments shouldn't be:

 case DRM_PLANE_TYPE_PRIMARY: /* primary */

Well, yes!, it's written just there, PRIMARY!

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Factor out common primary/sprite code in prepare_crtc()
Damien Lespiau [Tue, 8 Jul 2014 11:59:03 +0000 (12:59 +0100)]
kms_rotation_crc: Factor out common primary/sprite code in prepare_crtc()

This results in less code, always a good thing. Also, we only really
need one reference CRC.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Factor out the square drawing function
Damien Lespiau [Tue, 8 Jul 2014 11:49:03 +0000 (12:49 +0100)]
kms_rotation_crc: Factor out the square drawing function

Making function to the similar things is very common in programming.
Let's do it once again.

Cairo being a drawing library, it can be used to do the rotation!

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Fix style issue: single statement conditionals
Damien Lespiau [Tue, 8 Jul 2014 11:28:41 +0000 (12:28 +0100)]
kms_rotation_crc: Fix style issue: single statement conditionals

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Fix style issue: '{' at the end of lines
Damien Lespiau [Tue, 8 Jul 2014 11:26:38 +0000 (12:26 +0100)]
kms_rotation_crc: Fix style issue: '{' at the end of lines

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Require universal planes for the testing primary rotation
Damien Lespiau [Tue, 8 Jul 2014 11:18:47 +0000 (12:18 +0100)]
kms_rotation_crc: Require universal planes for the testing primary rotation

Otherwise the test will fail instead of just skipping.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Test the validity of the output first
Damien Lespiau [Tue, 8 Jul 2014 11:10:44 +0000 (12:10 +0100)]
kms_rotation_crc: Test the validity of the output first

So we don't need code to unwind what we just did.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Update the copyright to have this year as well
Damien Lespiau [Tue, 8 Jul 2014 11:08:20 +0000 (12:08 +0100)]
kms_rotation_crc: Update the copyright to have this year as well

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Align a few wrapped lines to the opening brace
Damien Lespiau [Tue, 8 Jul 2014 11:02:42 +0000 (12:02 +0100)]
kms_rotation_crc: Align a few wrapped lines to the opening brace

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Make more functions static
Damien Lespiau [Tue, 8 Jul 2014 10:58:43 +0000 (11:58 +0100)]
kms_rotation_crc: Make more functions static

More of the same. This time no need to move code around, just adding
static.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_rotation_crc: Make check_plane_type() static
Damien Lespiau [Tue, 8 Jul 2014 10:55:20 +0000 (11:55 +0100)]
kms_rotation_crc: Make check_plane_type() static

Clearly, someone tried to solve the following warning:
  kms_rotation_crc.c:189:6: warning: no previous prototype for ‘check_plane_type’ [-Wmissing-prototypes]

Without really understanding what was the warning about. Make
check_plane_type() static and move it before its user to get rid of the forward
declaration.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/kms_rotation_crc: IGT for 180 degree HW rotation
Sonika Jindal [Wed, 18 Jun 2014 08:57:27 +0000 (14:27 +0530)]
tests/kms_rotation_crc: IGT for 180 degree HW rotation

Testcase for 180 degree HW rotation

Cc: sagar.a.kamble@intel.com
Signed-off-by: Sonika Jindal <sonika.jindal@intel.com>
10 years agoigt_kms: Introduce a for_each_pipe() macro
Damien Lespiau [Tue, 8 Jul 2014 17:43:22 +0000 (18:43 +0100)]
igt_kms: Introduce a for_each_pipe() macro

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agoigt_kms: Add support for setting plane rotation
Damien Lespiau [Tue, 8 Jul 2014 15:02:05 +0000 (16:02 +0100)]
igt_kms: Add support for setting plane rotation

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agoigt_kms: Add a way to query of the plane supports rotation
Damien Lespiau [Tue, 8 Jul 2014 14:23:05 +0000 (15:23 +0100)]
igt_kms: Add a way to query of the plane supports rotation

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agoigt_kms: Provide a get_plane_property() shorthand
Damien Lespiau [Tue, 8 Jul 2014 13:56:54 +0000 (14:56 +0100)]
igt_kms: Provide a get_plane_property() shorthand

So one doesn't have to write the plane type all the time.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agoigt_kms: Factor out a generic get_property() out of get_drm_plane_type()
Damien Lespiau [Tue, 8 Jul 2014 13:34:16 +0000 (14:34 +0100)]
igt_kms: Factor out a generic get_property() out of get_drm_plane_type()

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib/igt_core: Don't log when listing subtests
Daniel Vetter [Fri, 11 Jul 2014 08:24:30 +0000 (10:24 +0200)]
lib/igt_core: Don't log when listing subtests

I've noticed some spam in the userptr list ...

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoREADME: update the test run instructions
Thomas Wood [Fri, 11 Jul 2014 09:47:43 +0000 (10:47 +0100)]
README: update the test run instructions

Describe using the new run-tests.sh script and also update the Piglit
instructions now that using a symlink is no longer supported.

Signed-off-by: Thomas Wood <thomas.wood@intel.com>
10 years agoscripts: add a script to help run tests with Piglit
Thomas Wood [Tue, 1 Jul 2014 10:35:16 +0000 (11:35 +0100)]
scripts: add a script to help run tests with Piglit

Add a script to facilitate running the tests with Piglit by providing
simplified options for listing, filtering and creating summaries of test
runs.

Signed-off-by: Thomas Wood <thomas.wood@intel.com>
10 years agolib: allow the edid to be reset
Thomas Wood [Wed, 9 Jul 2014 09:38:05 +0000 (10:38 +0100)]
lib: allow the edid to be reset

Signed-off-by: Thomas Wood <thomas.wood@intel.com>
10 years agotests: enable extra connectors in kms_flip and kms_pipe_crc_basic
Thomas Wood [Wed, 18 Jun 2014 13:28:43 +0000 (14:28 +0100)]
tests: enable extra connectors in kms_flip and kms_pipe_crc_basic

Signed-off-by: Thomas Wood <thomas.wood@intel.com>
10 years agolib: add igt_enable_connectors and igt_reset_connectors
Thomas Wood [Wed, 28 May 2014 13:06:01 +0000 (14:06 +0100)]
lib: add igt_enable_connectors and igt_reset_connectors

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.

10 years agolib: add the ability to set an EDID data block on a connector
Thomas Wood [Tue, 27 May 2014 13:47:25 +0000 (14:47 +0100)]
lib: add the ability to set an EDID data block on a connector

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>
10 years agolib: add function to change connector states
Thomas Wood [Fri, 16 May 2014 16:22:46 +0000 (17:22 +0100)]
lib: add function to change connector states

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>
10 years agoigt/gem_userptr_blits: Verify that userptr bo work on unshared memory
Chris Wilson [Fri, 11 Jul 2014 09:38:32 +0000 (10:38 +0100)]
igt/gem_userptr_blits: Verify that userptr bo work on unshared memory

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>
10 years agoigt/gem_userptr_blits: Shared memory allocations
Chris Wilson [Fri, 11 Jul 2014 09:16:54 +0000 (10:16 +0100)]
igt/gem_userptr_blits: Shared memory allocations

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>
10 years agoWait for any pid in order to reap failure quicker
Chris Wilson [Fri, 11 Jul 2014 09:15:19 +0000 (10:15 +0100)]
Wait for any pid in order to reap failure quicker

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>
10 years agoNEWS: Updates
Daniel Vetter [Thu, 10 Jul 2014 18:19:18 +0000 (20:19 +0200)]
NEWS: Updates

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Run igt.cocci
Daniel Vetter [Thu, 10 Jul 2014 18:04:19 +0000 (20:04 +0200)]
tests: Run igt.cocci

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>
10 years agodrv_module_reload: Unbind the right console driver
Daniel Vetter [Thu, 10 Jul 2014 17:46:46 +0000 (19:46 +0200)]
drv_module_reload: Unbind the right console driver

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>
10 years agodrv_module_reload: Don't declare success when failing
Damien Lespiau [Thu, 10 Jul 2014 13:03:37 +0000 (14:03 +0100)]
drv_module_reload: Don't declare success when failing

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>
10 years agocore: Apply the same treatment to the in errno message in __igt_fail_assert()
Damien Lespiau [Wed, 9 Jul 2014 10:40:16 +0000 (11:40 +0100)]
core: Apply the same treatment to the in errno message in __igt_fail_assert()

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>
10 years agocore: Only display the errno message if errno is set
Damien Lespiau [Wed, 9 Jul 2014 10:36:27 +0000 (11:36 +0100)]
core: Only display the errno message if errno is set

Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agocore: Put the requirement failure messages together
Damien Lespiau [Wed, 9 Jul 2014 10:32:22 +0000 (11:32 +0100)]
core: Put the requirement failure messages together

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>
10 years agolib/gen6_render: removed duplicate defines
Mika Kuoppala [Wed, 18 Jun 2014 11:34:04 +0000 (14:34 +0300)]
lib/gen6_render: removed duplicate defines

Textually the same so no harm was done and no warnings
from compiler either.

Signed-off-by: Mika Kuoppala <mika.kuoppala@intel.com>
10 years agotests/gem_render_copy_redux needs legacy drm nodes
Daniel Vetter [Wed, 9 Jul 2014 18:45:54 +0000 (20:45 +0200)]
tests/gem_render_copy_redux needs legacy drm nodes

... 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>
10 years agointel-gpu-tools: Dont build kms_univeral_plane if no cairo
Tim Gore [Tue, 8 Jul 2014 09:04:28 +0000 (10:04 +0100)]
intel-gpu-tools: Dont build kms_univeral_plane if no cairo

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>
10 years agokms_plane: Specify the pipe when grabbing reference CRCs
Damien Lespiau [Wed, 9 Jul 2014 10:05:06 +0000 (11:05 +0100)]
kms_plane: Specify the pipe when grabbing reference CRCs

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>
10 years agokms_plane: Add a debug message when grabbing the CRC of a colored fb
Damien Lespiau [Mon, 7 Jul 2014 13:41:15 +0000 (14:41 +0100)]
kms_plane: Add a debug message when grabbing the CRC of a colored fb

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_plane: Add panning test for primary plane
Yi Sun [Fri, 23 May 2014 00:28:47 +0000 (08:28 +0800)]
kms_plane: Add panning test for primary plane

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>
10 years agokms_plane: Make the gathering of reference CRCs test agnostic
Damien Lespiau [Mon, 7 Jul 2014 12:30:41 +0000 (13:30 +0100)]
kms_plane: Make the gathering of reference CRCs test agnostic

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>
10 years agopipe_crc: Make collect_crc() ensure the CRC looks somewhat valid
Damien Lespiau [Mon, 7 Jul 2014 14:17:56 +0000 (15:17 +0100)]
pipe_crc: Make collect_crc() ensure the CRC looks somewhat valid

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agopipe_crc: Warn if the CRC values is 0xffffffff
Damien Lespiau [Mon, 7 Jul 2014 14:16:51 +0000 (15:16 +0100)]
pipe_crc: Warn if the CRC values is 0xffffffff

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>
10 years agolib: Reset 'position_changed' after a drmModeSetCrtc()
Damien Lespiau [Mon, 7 Jul 2014 14:08:33 +0000 (15:08 +0100)]
lib: Reset 'position_changed' after a drmModeSetCrtc()

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>
10 years agokms_pipe_crc_basic: Add a bit a debugging output
Damien Lespiau [Wed, 28 May 2014 17:59:09 +0000 (18:59 +0100)]
kms_pipe_crc_basic: Add a bit a debugging output

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agokms_pipe_crc_basic: Make the number of CRCs a parameter
Damien Lespiau [Wed, 28 May 2014 16:55:06 +0000 (17:55 +0100)]
kms_pipe_crc_basic: Make the number of CRCs a parameter

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>
10 years agokms_pipe_crc_basic: Cycle between 2 differently colored buffer
Damien Lespiau [Wed, 28 May 2014 16:40:05 +0000 (17:40 +0100)]
kms_pipe_crc_basic: Cycle between 2 differently colored buffer

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>
10 years agokms_pipe_crc_basic: Split the main test function a bit more
Damien Lespiau [Wed, 28 May 2014 16:33:32 +0000 (17:33 +0100)]
kms_pipe_crc_basic: Split the main test function a bit more

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>
10 years agolib/kms: Provide universal plane #define's
Matt Roper [Mon, 7 Jul 2014 16:08:29 +0000 (09:08 -0700)]
lib/kms: Provide universal plane #define's

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>
10 years agotests/gem_exec_parse: use gem_uses_aliasing_ppgtt
Daniel Vetter [Mon, 7 Jul 2014 15:54:30 +0000 (17:54 +0200)]
tests/gem_exec_parse: use gem_uses_aliasing_ppgtt

Suggested by Brad Volking.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/kms_psr_sink_crc: Fix delayed gtt/cpu write tests
Daniel Vetter [Tue, 17 Jun 2014 21:47:49 +0000 (23:47 +0200)]
tests/kms_psr_sink_crc: Fix delayed gtt/cpu write tests

- 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>
10 years agokms_universal_plane: Don't assert outside of fixtures/subtests
Damien Lespiau [Thu, 3 Jul 2014 14:50:53 +0000 (15:50 +0100)]
kms_universal_plane: Don't assert outside of fixtures/subtests

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>
10 years agotests/kms_cursor_crc: Call drmModeMoveCursor() directly
Matt Roper [Mon, 30 Jun 2014 23:44:30 +0000 (16:44 -0700)]
tests/kms_cursor_crc: Call drmModeMoveCursor() directly

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>
10 years agotests/kms_cursor_crc: Combine data_t and test_data_t
Matt Roper [Mon, 30 Jun 2014 23:44:29 +0000 (16:44 -0700)]
tests/kms_cursor_crc: Combine data_t and test_data_t

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>
10 years agotests/kms_universal_plane: Universal plane testing (v6)
Matt Roper [Mon, 30 Jun 2014 23:44:28 +0000 (16:44 -0700)]
tests/kms_universal_plane: Universal plane testing (v6)

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>
10 years agolib/kms: Drop primary plane windowing test from igt_plane_set_position()
Matt Roper [Mon, 30 Jun 2014 23:44:27 +0000 (16:44 -0700)]
lib/kms: Drop primary plane windowing test from igt_plane_set_position()

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>
10 years agolib/kms: Add universal plane support (v2)
Matt Roper [Mon, 30 Jun 2014 23:44:26 +0000 (16:44 -0700)]
lib/kms: Add universal plane support (v2)

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>
10 years agolib/kms: Add igt_display_try_commit2()
Matt Roper [Mon, 30 Jun 2014 23:44:25 +0000 (16:44 -0700)]
lib/kms: Add igt_display_try_commit2()

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>
10 years agolib/kms: Add igt_display_commit2()
Matt Roper [Mon, 30 Jun 2014 23:44:24 +0000 (16:44 -0700)]
lib/kms: Add igt_display_commit2()

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>
10 years agolib/kms: Track need for vblank wait locally
Matt Roper [Mon, 30 Jun 2014 23:44:23 +0000 (16:44 -0700)]
lib/kms: Track need for vblank wait locally

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>
10 years agolib/kms: Drop igt_pipe->need_set_{crtc, cursor}
Matt Roper [Mon, 30 Jun 2014 23:44:22 +0000 (16:44 -0700)]
lib/kms: Drop igt_pipe->need_set_{crtc, cursor}

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>
10 years agogem_exec_parse: require PPGTT as well
Jesse Barnes [Thu, 26 Jun 2014 15:47:08 +0000 (08:47 -0700)]
gem_exec_parse: require PPGTT as well

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>
10 years agotests/gem_ctx_exec: Add reset-pin-leak subtest
Ville Syrjälä [Wed, 18 Jun 2014 18:14:11 +0000 (21:14 +0300)]
tests/gem_ctx_exec: Add reset-pin-leak subtest

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>
10 years agogem_mmap_gtt: Test mmaping less than the full object
Chris Wilson [Sat, 21 Jun 2014 14:39:09 +0000 (15:39 +0100)]
gem_mmap_gtt: Test mmaping less than the full object

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>
10 years agogem_fence_upload: Improve the contended messages
Chris Wilson [Sat, 21 Jun 2014 14:38:31 +0000 (15:38 +0100)]
gem_fence_upload: Improve the contended messages

In the contended upload case, we did not distinguish between linear and
tiled uploads.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
10 years agointel-gpu-tools: add igt_core init func calls to some tests
Tim Gore [Fri, 20 Jun 2014 10:28:12 +0000 (11:28 +0100)]
intel-gpu-tools: add igt_core init func calls to some tests

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>
10 years agotests/gem_exec_big: Re-add gem_sync
Daniel Vetter [Thu, 19 Jun 2014 08:13:12 +0000 (10:13 +0200)]
tests/gem_exec_big: Re-add gem_sync

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>
10 years agotests/gem_mmap: Add short mmap test
Daniel Vetter [Tue, 17 Jun 2014 21:06:59 +0000 (23:06 +0200)]
tests/gem_mmap: Add short mmap test

Blows up with the current full-object prefault code.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoParitially revert "lib: add missing includes to headers"
Daniel Vetter [Tue, 17 Jun 2014 19:22:05 +0000 (21:22 +0200)]
Paritially revert "lib: add missing includes to headers"

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>
10 years agotests: Move gem_userptr_blits to the right Makefile target
Daniel Vetter [Tue, 17 Jun 2014 16:52:08 +0000 (18:52 +0200)]
tests: Move gem_userptr_blits to the right Makefile target

We need to be able to check for this somehow ...

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/Makefile.sources: Remove empty last line
Daniel Vetter [Wed, 30 Apr 2014 16:15:36 +0000 (18:15 +0200)]
tests/Makefile.sources: Remove empty last line

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>
10 years agoFixed the review issues for pm_rc6_residency IGT case
Wendy Wang [Mon, 9 Jun 2014 08:36:47 +0000 (16:36 +0800)]
Fixed the review issues for pm_rc6_residency IGT case

Why need add rc6_residency_counter subtest case:
RC6 feature support residency counter,from power consumption aspect,
the counter closer to 1,the better.If the counter is < 0.9, the residency
is not good and will impact power consumption value, if the counter is  > 1,
sysfs file is inaccurate.

Attach the test result message:
root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rc6_residency
IGT-Version: 1.6-g9a70e29 (x86_64) (Linux: 3.15.0-rc7_drm-intel-nightly_0a37b5_20140604+ x86_64)
Subtest rc6-residency-check: SUCCESS
This machine doesn't support rc6pp
This machine doesn't support rc6p
The residency counter : 0.987000
This machine entry rc6 state.
Subtest rc6-residency-counter: SUCCESS

root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rc6_residency --run-subtest rc6-residency-counter
IGT-Version: 1.6-g9a70e29 (x86_64) (Linux: 3.15.0-rc7_drm-intel-nightly_0a37b5_20140604+ x86_64)
This machine doesn't support rc6pp
This machine doesn't support rc6p
The residency counter : 0.987000
This machine entry rc6 state.
Subtest rc6-residency-counter: SUCCESS

root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rc6_residency --run-subtest rc6-residency-check
IGT-Version: 1.6-g9a70e29 (x86_64) (Linux: 3.15.0-rc7_drm-intel-nightly_0a37b5_20140604+ x86_64)
Subtest rc6-residency-check: SUCCESS

root@x-bdw05:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# ./pm_rc6_residency --list
rc6-residency-check
rc6-residency-counter

Run as non-root
[haha@x-pk home]$ ./pm_rc6_residency
IGT-Version: 1.6-g18d2130 (x86_64) (Linux: 3.13.0-rc3_drm-intel-nightly_639e4d_20131210+ x86_64)
No intel gpu found
Subtest rc6-residency-check: SKIP
Subtest rc6-residency-counter: SKIP

Run on non-intel platform
[root@x-pk5 home]# ./pm_rc6_residency
IGT-Version: 1.6-g18d2130 (x86_64) (Linux: 3.13.0-rc3_drm-intel-nightly_639e4d_20131210+ x86_64)
Test requirement not met in function read_rc6_residency, file pm_rc6_residency.c:77:
Last errno: 2, No such file or directory
Test requirement: (!(file))
Subtest rc6-residency-check: SKIP
Subtest rc6-residency-counter: SKIP

Signed-off-by: Wendy Wang <wendy.wang@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoRevert "Revert "Add rc6_residency_counter subtest""
Daniel Vetter [Mon, 16 Jun 2014 21:54:43 +0000 (23:54 +0200)]
Revert "Revert "Add rc6_residency_counter subtest""

This reverts commit 35554a1bcaaea55c1cfa88c0176c58d2fb3b8013.

Apparently people are too unhappy with this.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: add missing includes to headers
Thomas Wood [Mon, 16 Jun 2014 15:21:35 +0000 (16:21 +0100)]
lib: add missing includes to headers

Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: ensure igt_display_init clears the memory for the display struct
Thomas Wood [Mon, 16 Jun 2014 15:12:21 +0000 (16:12 +0100)]
lib: ensure igt_display_init clears the memory for the display struct

Add the call to memset that was accidentally removed in:

commit 1e9e1baba389fe498be12390ceeeacb1d141a5cf
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Thu Mar 13 17:20:05 2014 +0100

    lib/igt_kms: rip out custom verbose loggin support

    Instead just piggy-pack on top of igt_log.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Thomas Wood <thomas.wood@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agointel-gpu-tools: remove the --cmd option from gem_seqno_wrap
Tim Gore [Mon, 16 Jun 2014 08:38:09 +0000 (09:38 +0100)]
intel-gpu-tools: remove the --cmd option from gem_seqno_wrap

gem_seqno_wrap was not being built on Android because it uses
wordexp which is not in Bionic.
After discussion with Mika Kuoppala (the test author) it seems
that wordexp was used to implement the --cmd option that was
really only intended for use during development of the test and
is no longer needed. So I have removed support for this option
and enabled this test for Android.

Signed-off-by: Tim Gore <tim.gore@intel.com>
Reviewed-by: Mika Kuoppala <mika.kuoppala@intel.com>
[danvet: Resolve conflicts with cocci run.]
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoigt/gem_exec_big: Update to new igt_assert_eq
Chris Wilson [Mon, 16 Jun 2014 09:49:16 +0000 (10:49 +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>
10 years agoigt/gem_exec_parse: Convert to igt_assert_eq
Chris Wilson [Mon, 16 Jun 2014 08:01:52 +0000 (09:01 +0100)]
igt/gem_exec_parse: Convert to igt_assert_eq

For better test output on failure.

References: https://bugs.freedesktop.org/show_bug.cgi?id=80083
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
10 years agotools/intel_display_poller: Add a new tool that will poll various display registers
Ville Syrjälä [Tue, 29 Oct 2013 16:17:39 +0000 (18:17 +0200)]
tools/intel_display_poller: Add a new tool that will poll various display registers

intel_poller can be used to poll various display registers
(IIR,scanline/pixel/flip/frame counter, live address, etc.).

It can be used to determine eg. at which scanline or pixel count certain
events occur.

v2: s/intel_poller/intel_display_poller/

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agotools: Add intel_iosf_sb_{read,write} tools
Ville Syrjälä [Tue, 10 Jun 2014 18:28:10 +0000 (21:28 +0300)]
tools: Add intel_iosf_sb_{read,write} tools

Add generic tools to poke at IOSF sideband. The user needs to
manually specify SB port as well as the register.

TODO: Maybe add symbolic names for the units? Would avoid having
to trawl the docs for the magic hex value.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agolib/igt_debufs: Add IGT_NO_FORCEWAKE environment variable
Ville Syrjälä [Wed, 11 Jun 2014 16:21:27 +0000 (19:21 +0300)]
lib/igt_debufs: Add IGT_NO_FORCEWAKE environment variable

If IGT_NO_FORCEWAKE is set, skip the forcewake open. Useful when you
want to poke at register without otherwise disturbing the GPU.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agolib/igt_debugfs: Don't fail if debugfs is already mounted
Ville Syrjälä [Wed, 11 Jun 2014 16:20:19 +0000 (19:20 +0300)]
lib/igt_debugfs: Don't fail if debugfs is already mounted

Remove the igt_assert() from the debugfs mount. It will fail if debugfs
is already mounted. With the assert in place it's very annying to use
igt without i915 loaded (eg. to dump BIOS configured registers).

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agotests: run igt.cocci
Daniel Vetter [Fri, 13 Jun 2014 16:27:59 +0000 (18:27 +0200)]
tests: run igt.cocci

Re-run with correct igt_fail rules. Again manually fixup missing
includes for igt_core.h.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoNEWS: Mention igt.cocci
Daniel Vetter [Fri, 13 Jun 2014 14:23:10 +0000 (16:23 +0200)]
NEWS: Mention igt.cocci

And also pimp the spatch file itself with usage hints.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: Introduce igt_fail_on/_f
Daniel Vetter [Fri, 13 Jun 2014 16:01:33 +0000 (18:01 +0200)]
lib: Introduce igt_fail_on/_f

I've yet again totally screwed things up (this time automated with
cocci even, but not yet pushed luckily). So finally add a new version
for easier conversion and adjust the cocci script.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib/igt.cocci: Also add rule to use igt_warn_on_f
Daniel Vetter [Fri, 13 Jun 2014 13:45:30 +0000 (15:45 +0200)]
lib/igt.cocci: Also add rule to use igt_warn_on_f

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib/igt.cocci: Convert abort() to igt_fail
Daniel Vetter [Fri, 13 Jun 2014 13:45:12 +0000 (15:45 +0200)]
lib/igt.cocci: Convert abort() to igt_fail

abort should only be used for internal library checks - using
abort() we get a "crash" result, using igt_fail we get "fail"
in piglit.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib/igt.cocci: Conversion to igt logging
Daniel Vetter [Fri, 13 Jun 2014 13:41:28 +0000 (15:41 +0200)]
lib/igt.cocci: Conversion to igt logging

Also update old hunks to match on igt logging instead of fprintf.

v2: Don't forget about perror.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Don't use stderr for informational messages
Daniel Vetter [Fri, 13 Jun 2014 13:32:50 +0000 (15:32 +0200)]
tests: Don't use stderr for informational messages

These should go to stdout instead. The next patch will clean this up
with cocci, so no change from fprintf(stdout, to printf( here.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/pm_psr_sink_crc: Fix longjmp fun
Daniel Vetter [Fri, 13 Jun 2014 13:19:09 +0000 (15:19 +0200)]
tests/pm_psr_sink_crc: Fix longjmp fun

igt_fixture and igt_subtests use longjmp/setjmp internally, which
means local variables at the same stack frame are at risk. Best
practice is to move them out right in front of the igt_main block.

It would be awesome if someone could come up with a cocci patch to
auto-fix this, but unfortunately my attempts failed.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoRevert "tests: Run igt.cocci over tests"
Daniel Vetter [Fri, 13 Jun 2014 16:03:50 +0000 (18:03 +0200)]
Revert "tests: Run igt.cocci over tests"

This reverts commit 6903ab04e5f9048e3932eb3225e94b6a228681ba.

The igt_assert conversion rule is broken and doesn't invert the check
as it should.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoigt/gem_fence_upload: Exercise contention on the fault handler
Chris Wilson [Fri, 13 Jun 2014 14:46:33 +0000 (15:46 +0100)]
igt/gem_fence_upload: Exercise contention on the fault handler

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
10 years agotests: Run igt.cocci over tests
Daniel Vetter [Fri, 13 Jun 2014 09:06:18 +0000 (11:06 +0200)]
tests: Run igt.cocci over tests

Cocci is awesome

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: add igt.cocci
Daniel Vetter [Fri, 13 Jun 2014 09:16:28 +0000 (11:16 +0200)]
lib: add igt.cocci

Small start but useful to collect refactorings/simplifications for
common igt patterns. Please add more if you stumble over some so that
we can occasionally run this to clean up the tests.

I haven't figured out yet how to do the assert_cmpint->assert_eq
transformation.

Run this with

spatch --sp-file lib/igt.cocci --in-place tests/*.c

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib/igt_core: Add igt_assert_eq
Daniel Vetter [Fri, 13 Jun 2014 08:46:55 +0000 (10:46 +0200)]
lib/igt_core: Add igt_assert_eq

Suggested by Chris Wilson. Not yet rolled out since I'm trying to use
cocci for this.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/prime_self_import: Use igt_assert_cmpint
Daniel Vetter [Fri, 13 Jun 2014 08:23:01 +0000 (10:23 +0200)]
tests/prime_self_import: Use igt_assert_cmpint

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/kms_psr_sink_crc: Skip properly
Daniel Vetter [Fri, 13 Jun 2014 08:18:24 +0000 (10:18 +0200)]
tests/kms_psr_sink_crc: Skip properly

Aside: The test has way too many bool return values that are then
always checked with igt_assert. Imo cleaner to switch to a more
declarative approach and shovel the igt_assert/require into those
functions instead.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79962
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoigt/gem_fence_upload: Throw some threads into the mix
Chris Wilson [Thu, 12 Jun 2014 16:01:02 +0000 (17:01 +0100)]
igt/gem_fence_upload: Throw some threads into the mix

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
10 years agoigt/gem_mmap_gtt: Exercise concurrent pagefaulting
Chris Wilson [Thu, 12 Jun 2014 10:50:01 +0000 (11:50 +0100)]
igt/gem_mmap_gtt: Exercise concurrent pagefaulting

This should hit the BUG inside remap_pfn_range in

commit c5158fabeaf53ed2c614c3333aaa4b3cce80f500
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Jun 10 12:14:41 2014 +0100

[   27.767634] kernel BUG at mm/memory.c:2315!
[   27.767655] invalid opcode: 0000 [#1] SMP
[   27.767679] Modules linked in: cpufreq_userspace cpufreq_powersave cpufreq_stats cpufreq_conservative binfmt_misc nfs lockd fscache sunrpc hid_generic usbhid hid x86_pkg_temp_thermal microcode i2c_i801 lpc_ich mfd_core battery acpi_cpufreq evdev processor ac loop ehci_pci xhci_hcd ehci_hcd sr_mod usbcore cdrom usb_common fan thermal
[   27.767872] CPU: 3 PID: 912 Comm: gem_mmap_gtt Not tainted 3.15.0-rc8+ #953
[   27.767903] Hardware name: Intel Corporation Shark Bay Client platform/Flathead Creek Crb, BIOS HSWLPTU1.86C.0109.R03.1301282055 01/28/2013
[   27.767956] task: ffff880448415010 ti: ffff88044d22c000 task.ti: ffff88044d22c000
[   27.767988] RIP: 0010:[<ffffffff81130734>]  [<ffffffff81130734>] remap_pfn_range+0x2a4/0x400
[   27.768033] RSP: 0000:ffff88044d22fc28  EFLAGS: 00010282
[   27.768057] RAX: 0000000000020002 RBX: 00000000000a3b57 RCX: ffff880448b73fe8
[   27.768088] RDX: 0000000000000002 RSI: ffff880000000000 RDI: ffffea000efe8158
[   27.768119] RBP: ffff88044d22fcd8 R08: 00007fc7b57fe000 R09: 00007fc7b57fe000
[   27.768150] R10: 00000000000001fd R11: 0000000000000a9a R12: ffffea000efe8128
[   27.768180] R13: 0000000000000001 R14: 00007fc7b57fd000 R15: 800000000000002f
[   27.768212] FS:  00007fc7977fc700(0000) GS:ffff88045e380000(0000) knlGS:0000000000000000
[   27.768246] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[   27.768272] CR2: 00007fc7b67fd000 CR3: 000000044866c000 CR4: 00000000001407e0
[   27.768303] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[   27.768333] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[   27.768363] Stack:
[   27.768374]  ffff880448bbb6f8 00000000ffffffff 00007fc7b57fdfff 00007fc7b57fdfff
[   27.768414]  00007fc7b57fe000 ffff88044866c7f8 ffff8804496b08f0 00007fc7b57fdfff
[   27.768454]  fffffff8038ee35a 0000000000001000 ffff88044c208180 00007fc7b57fe000
[   27.768494] Call Trace:
[   27.768511]  [<ffffffff81365277>] i915_gem_fault+0x337/0x340
[   27.768538]  [<ffffffff8112d3c4>] __do_fault+0x34/0x70
[   27.768565]  [<ffffffff8109088e>] ? wake_up_process+0x1e/0x40
[   27.768592]  [<ffffffff8113015c>] do_shared_fault.isra.96+0x2c/0x1f0
[   27.768623]  [<ffffffff81502405>] ? rwsem_down_read_failed+0xe5/0x130
[   27.768654]  [<ffffffff810a2519>] ? __rwsem_do_wake+0x129/0x160
[   27.768682]  [<ffffffff81131085>] handle_mm_fault+0x2b5/0xb80
[   27.768712]  [<ffffffff81270a64>] ? call_rwsem_down_read_failed+0x14/0x30
[   27.768745]  [<ffffffff81036c67>] __do_page_fault+0x167/0x4c0
[   27.768774]  [<ffffffff8109a540>] ? pick_next_task_fair+0x700/0x870
[   27.768804]  [<ffffffff814ff23b>] ? __schedule+0x27b/0x860
[   27.768831]  [<ffffffff81036fec>] do_page_fault+0xc/0x10
[   27.768857]  [<ffffffff815034a2>] page_fault+0x22/0x30
[   27.768881] Code: 4d 85 ed 49 0f 44 d7 80 cc 02 49 81 c6 00 10 00 00 48 83 c3 01 48 83 c1 08 48 09 d0 48 89 41 f8 4d 39 f0 74 32 48 83 39 00 74 c4 <0f> 0b 66 2e 0f 1f 84 00 00 00 00 00 48 39 37 75 63 48 8b 45 c8
[   27.769081] RIP  [<ffffffff81130734>] remap_pfn_range+0x2a4/0x400
[   27.769113]  RSP <ffff88044d22fc28>

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>