platform/upstream/intel-gpu-tools.git
10 years agotests/gem_ctx_bad_exec: Check the errno, too
Daniel Vetter [Tue, 5 Nov 2013 10:56:50 +0000 (11:56 +0100)]
tests/gem_ctx_bad_exec: Check the errno, too

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agogem_flink_race: Assure no pending requests before object counting
Oscar Mateo [Tue, 5 Nov 2013 10:57:31 +0000 (10:57 +0000)]
gem_flink_race: Assure no pending requests before object counting

Same thing that was done for prime_self_import.

v2: Move igt_drop_caches_set() call inside get_object_count() to make
it clearer why we want this.

Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoprime_self_import: Assure no pending requests before object counting
Oscar Mateo [Tue, 5 Nov 2013 10:56:24 +0000 (10:56 +0000)]
prime_self_import: Assure no pending requests before object counting

We don't want a previously used object to be freed in the middle of a
before/after object counting operation (or we would get a "-1 objects
leaked" message). We have seen this happening, e.g., when a context
from a previous run dies, but its backing object is alive waiting for
a retire_work to kick in.

v2: Use igt_debugfs facilities for drop cache.
v3: Move igt_drop_caches_set() call inside get_object_count() to make
it clearer why we want this.

Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Ben Widawsky <ben@bwidawsk.net>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoNEWS: Roll in updates.
Daniel Vetter [Sun, 3 Nov 2013 10:28:50 +0000 (11:28 +0100)]
NEWS: Roll in updates.

10 years agotests: establish core_ prefix
Daniel Vetter [Sat, 2 Nov 2013 11:42:21 +0000 (12:42 +0100)]
tests: establish core_ prefix

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Use kms_ prefix a bit more
Daniel Vetter [Sat, 2 Nov 2013 11:35:44 +0000 (12:35 +0100)]
tests: Use kms_ prefix a bit more

I was a bit on the fence about the basic pipe CRC test since that
doesn't really test kms, but debug infrastructure in debugfs.

Otoh running this one for a full kms testrun is always good, to make
sure that all the other (real) CRC based tests work sanely.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: establish pm_ prefix
Daniel Vetter [Sat, 2 Nov 2013 11:27:24 +0000 (12:27 +0100)]
tests: establish pm_ prefix

Imo power management, power consumption and performance are tightly
enough coupled that we can throw them all into one bin.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: estbalish drv_ prefix
Daniel Vetter [Sat, 2 Nov 2013 11:22:37 +0000 (12:22 +0100)]
tests: estbalish drv_ prefix

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Start to document naming conventions
Daniel Vetter [Sat, 2 Nov 2013 11:17:49 +0000 (12:17 +0100)]
tests: Start to document naming conventions

Just a start and we need more work here. Some follow-up patches will
clear up some of the historical confusion.

While at it rename the pc8 "basic" test to "rte".

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agoUpdate .gitignore a bit
Daniel Vetter [Sat, 2 Nov 2013 11:54:35 +0000 (12:54 +0100)]
Update .gitignore a bit

- Ignore build-aux/
- Cleanup ignores for assembler/

10 years agolib: drop return value from igt_drop_caches
Daniel Vetter [Mon, 4 Nov 2013 17:11:29 +0000 (18:11 +0100)]
lib: drop return value from igt_drop_caches

No one actually cares, everyone expects it to just work.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: Add igt_drop_caches_set()
Oscar Mateo [Mon, 4 Nov 2013 16:30:47 +0000 (16:30 +0000)]
lib: Add igt_drop_caches_set()

This is basically a "drop cache" interface to the igt_debugfs
facilities. Also, update existing users.

Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Fix "-thrashing" and "-thrash-inactive" distinction
Oscar Mateo [Mon, 4 Nov 2013 16:30:46 +0000 (16:30 +0000)]
tests: Fix "-thrashing" and "-thrash-inactive" distinction

A typo in the relocation tests made both sub-tests perform the
same action: drop *all* caches.

Signed-off-by: Oscar Mateo <oscar.mateo@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: igt_no_subtest
Daniel Vetter [Fri, 1 Nov 2013 21:18:20 +0000 (22:18 +0100)]
tests: igt_no_subtest

Yet another check for the library.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Add a Makefile comment about the expected failures in igt_ tests
Daniel Vetter [Fri, 1 Nov 2013 17:31:59 +0000 (18:31 +0100)]
tests: Add a Makefile comment about the expected failures in igt_ tests

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Add a test template
Daniel Vetter [Thu, 31 Oct 2013 16:05:28 +0000 (17:05 +0100)]
tests: Add a test template

Requested by Paulo.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: add igt_main macro
Daniel Vetter [Thu, 31 Oct 2013 15:23:26 +0000 (16:23 +0100)]
lib: add igt_main macro

In the past new testcases with subtest often forgot to add the call to
igt_exit at the end of their main() function. That is now caught with
a bit more obnoxious asserts, but it's still a nuissance.

This little igt_main macro takes care of that (and also of calling the
subtest machinery initialization code correctly).

If no one objects I'll roll this out for all the simple cases (i.e.
those tests that don't have additional argv parsing on top of the
subtest machinery).

v2: Roll it out across the board.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: make igt_install_exit_handler never fail
Daniel Vetter [Thu, 31 Oct 2013 16:16:33 +0000 (17:16 +0100)]
lib: make igt_install_exit_handler never fail

Most callers didn't bother checking, so just move the asserts into
the function itself.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: make igt_set_vt_graphics_mode never fail
Daniel Vetter [Thu, 31 Oct 2013 16:02:41 +0000 (17:02 +0100)]
lib: make igt_set_vt_graphics_mode never fail

All tests agree that this is a letal failure, so no point
to pass it back to callers.

Also add some igt_require calls for the drm_fd where I've
noticed that it's missing.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: consolidate pipe crc exit handler
Daniel Vetter [Thu, 31 Oct 2013 15:09:46 +0000 (16:09 +0100)]
lib: consolidate pipe crc exit handler

No need to sprinkle this all over:
- exit handlers will only be registered once
- they're always called when exiting, so no need to explictly call
  them.

This allows us to hide all the pipe crc cleanup in the library.

Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/kms_cursor_crc: rework to auto-skip connectors
Daniel Vetter [Thu, 31 Oct 2013 15:06:40 +0000 (16:06 +0100)]
tests/kms_cursor_crc: rework to auto-skip connectors

Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/debugfs_pipe_crc: set mode before creating CRC helper
Daniel Vetter [Thu, 31 Oct 2013 14:37:05 +0000 (15:37 +0100)]
tests/debugfs_pipe_crc: set mode before creating CRC helper

Otherwise the automagic skipping for DP ports on gmch platforms
won't work.

v2: We also need to just skip that connector, not the entire subtest.

kms_cursor_crc still needs to be updated.

Cc: Damien Lespiau <damien.lespiau@intel.com>
Cc: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: use "auto" pipe CRC source
Daniel Vetter [Thu, 31 Oct 2013 13:02:44 +0000 (14:02 +0100)]
tests: use "auto" pipe CRC source

Makes stuff work on DP ports on gmch platforms automatically.

Cc: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: add more self-tests around the igt_exit checks.
Daniel Vetter [Fri, 1 Nov 2013 17:29:33 +0000 (18:29 +0100)]
lib: add more self-tests around the igt_exit checks.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: fix igt_exit assert when only listing subtests.
Daniel Vetter [Fri, 1 Nov 2013 17:21:08 +0000 (18:21 +0100)]
lib: fix igt_exit assert when only listing subtests.

Ooops. Reported by Paulo. Also add a new testcase for make check to
make sure this actually works.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: Don't forget to close the pipe ctl fd in igt_pipe_crc_reset()
Damien Lespiau [Fri, 1 Nov 2013 16:25:10 +0000 (16:25 +0000)]
lib: Don't forget to close the pipe ctl fd in igt_pipe_crc_reset()

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agogitignore: Ignore recently introduced tools and tests
Damien Lespiau [Fri, 1 Nov 2013 15:47:42 +0000 (15:47 +0000)]
gitignore: Ignore recently introduced tools and tests

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agomodule_reload: remove snd_hda_intel
Paulo Zanoni [Mon, 21 Oct 2013 13:45:41 +0000 (11:45 -0200)]
module_reload: remove snd_hda_intel

The audio driver uses the power well provided by our driver, so on
Haswell we can't "rmmod i915" if we don't "rmmod snd_hda_intel" first.

The problem with removing snd_hda_intel is that we also need to kill
its users. On the specific machine I tested, the only user seem to be
alsactl, but on other machines this may change. IMHO we should leave
the "kill user space" step to whoever is running the script, but
Daniel asked me to put it here so we have a better chance of Just
Working on QA's machines.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=70336
Requested-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
10 years agoAdd VGA register read/write tools
Ville Syrjälä [Thu, 19 Sep 2013 15:14:54 +0000 (18:14 +0300)]
Add VGA register read/write tools

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agodebugfs_pipe_crc: skip on simulation
Ben Widawsky [Thu, 31 Oct 2013 21:16:41 +0000 (14:16 -0700)]
debugfs_pipe_crc: skip on simulation

As with other display tests, they're not working (and in this case,
cause hangs).

Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
10 years agotests/pc8: remove some printfs
Paulo Zanoni [Tue, 29 Oct 2013 16:02:43 +0000 (14:02 -0200)]
tests/pc8: remove some printfs

They pollute the output, which makes it harder to see which tests
failed or succeeded.

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
10 years agotests/pc8: clarify modeset_subtest arguments
Paulo Zanoni [Tue, 29 Oct 2013 13:14:50 +0000 (11:14 -0200)]
tests/pc8: clarify modeset_subtest arguments

My tiny little brain can't properly process 3 unnamed boolean
arguments.

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
10 years agotests/pc8: add modeset -stress-no-wait tests
Paulo Zanoni [Tue, 29 Oct 2013 13:00:45 +0000 (11:00 -0200)]
tests/pc8: add modeset -stress-no-wait tests

Used to trigger a racing condition between mode setting and
enabling/disabling PC8. The modeset-lpsp-stress-no-wait test should
fail on Kernels without the fix I just wrote today.

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
10 years agotests/pc8: add modeset-lpsp and modeset-non-lpsp
Paulo Zanoni [Mon, 28 Oct 2013 21:11:31 +0000 (19:11 -0200)]
tests/pc8: add modeset-lpsp and modeset-non-lpsp

The goal of these tests is to exercise the combination of power well
enabling/disabling and PC8 enabling/disabling. The modeset-non-lpsp
test exposes a silent bug on the current code that can just be
detected with the Runtime D3 patches. The Kernel patch that fixes the
bug will add a WARN to unmute it on the PC8-only case.

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
10 years agogem_suspend: Skip on simulation
Ben Widawsky [Wed, 30 Oct 2013 03:01:43 +0000 (20:01 -0700)]
gem_suspend: Skip on simulation

Simulation has proven flaky across both reset, and s3/4

Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
10 years agogem_close_race: Fix linker flags
Ben Widawsky [Wed, 30 Oct 2013 03:01:14 +0000 (20:01 -0700)]
gem_close_race: Fix linker flags

Seems the -lpthread was missed in
commit 8f771f3facae133bb72216fa05c74bc817920b6d
Author: Chris Wilson <chris@chris-wilson.co.uk>
Date:   Tue Oct 29 14:21:29 2013 +0000

    gem_close_race: Also test random closing of active fd

Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
10 years agogem_close_race: Also test random closing of active fd
Chris Wilson [Tue, 29 Oct 2013 14:21:29 +0000 (14:21 +0000)]
gem_close_race: Also test random closing of active fd

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
10 years agolib/drmtest: Scream harder when igt_exit isn't called for subtest tests
Daniel Vetter [Tue, 29 Oct 2013 10:29:00 +0000 (11:29 +0100)]
lib/drmtest: Scream harder when igt_exit isn't called for subtest tests

We really need this since otherwise the magic return value handling
for running testcases with piglit (or on QA's validation
infrastructure) doesn't work properly.

We need to be careful though to only install this check on success.

See also the previous commits to sprinkle igt_exit() calls over all
the tests that missed it.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Use igt_exit for tests with subtest
Daniel Vetter [Tue, 29 Oct 2013 10:15:43 +0000 (11:15 +0100)]
tests: Use igt_exit for tests with subtest

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agogem_close_race: Bump the workload
Chris Wilson [Tue, 29 Oct 2013 09:25:44 +0000 (09:25 +0000)]
gem_close_race: Bump the workload

Emit more work per client and many more clients in order to increase the
chance of racing i915_gem_release() and i915_gem_retire_requests()

10 years agogem_close_race: Tidy up call to execbuffer
Chris Wilson [Tue, 29 Oct 2013 08:38:30 +0000 (08:38 +0000)]
gem_close_race: Tidy up call to execbuffer

Pass the right pointer to the execlist would be a good start.

Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
10 years agotests/kms_cursor_crc: Use igt_exit
Daniel Vetter [Tue, 29 Oct 2013 06:47:33 +0000 (07:47 +0100)]
tests/kms_cursor_crc: Use igt_exit

Otherwise the exit codes are all bogus.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests: Convert the ctx test to use render nodes when possible
Damien Lespiau [Thu, 24 Oct 2013 14:27:12 +0000 (15:27 +0100)]
tests: Convert the ctx test to use render nodes when possible

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/gem_render_copy: Use render nodes
Damien Lespiau [Thu, 24 Oct 2013 14:21:59 +0000 (15:21 +0100)]
tests/gem_render_copy: Use render nodes

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add a drm_open_any_render() that will try to use render nodes
Damien Lespiau [Thu, 24 Oct 2013 14:19:32 +0000 (15:19 +0100)]
lib: Add a drm_open_any_render() that will try to use render nodes

I was fedup with having to run my tests as root and not being able to
use my usual setup for tests that only exercise the GT part of the GPU.

Render nodes to the rescue!

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Close non intel fds in drm_get_card()
Damien Lespiau [Thu, 24 Oct 2013 14:04:26 +0000 (15:04 +0100)]
lib: Close non intel fds in drm_get_card()

When going through card%u devices, close the ones that we were able to
open but weren't intel devices.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Remove stale comment
Damien Lespiau [Thu, 24 Oct 2013 13:08:17 +0000 (14:08 +0100)]
lib: Remove stale comment

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests: Add gem_close_race
Chris Wilson [Mon, 28 Oct 2013 09:10:46 +0000 (09:10 +0000)]
tests: Add gem_close_race

The intention is to exercise #70784. Yet the first challenge is make the
test suite happy.

10 years agogem_render_linear_blits: Remove aub dump support
Damien Lespiau [Tue, 22 Oct 2013 14:03:48 +0000 (15:03 +0100)]
gem_render_linear_blits: Remove aub dump support

It's much easier to follow the new gem_render_copy test and acquire a
aub dump from it.

Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/gem_render_copy: Add aub dump support
Damien Lespiau [Tue, 22 Oct 2013 13:57:57 +0000 (14:57 +0100)]
tests/gem_render_copy: Add aub dump support

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/gem_render_copy: Only dump pngs when the -d option is given
Damien Lespiau [Tue, 22 Oct 2013 13:26:38 +0000 (14:26 +0100)]
tests/gem_render_copy: Only dump pngs when the -d option is given

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/gem_render_copy: Add a simple render copy test
Damien Lespiau [Tue, 22 Oct 2013 13:20:10 +0000 (14:20 +0100)]
tests/gem_render_copy: Add a simple render copy test

The goal is here to both: demonstrate a simple usage of render copy with
the possibility to write pngs to visualize what it's doing and to
provide a test bed to port the render copy function to new
architectures.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add a function to dump a scratch buf into a png
Damien Lespiau [Tue, 22 Oct 2013 13:18:26 +0000 (14:18 +0100)]
lib: Add a function to dump a scratch buf into a png

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agobuild: include intel_l3_parity.h to its _SOURCES to fix make distcheck
Rodrigo Vivi [Tue, 22 Oct 2013 14:35:59 +0000 (12:35 -0200)]
build: include intel_l3_parity.h to its _SOURCES to fix make distcheck

and make sure intel_l3_parity.h will be included in tarball.

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
10 years agobuild: Fix small typo that was breaking make distcheck
Rodrigo Vivi [Tue, 22 Oct 2013 13:28:56 +0000 (11:28 -0200)]
build: Fix small typo that was breaking make distcheck

Signed-off-by: Rodrigo Vivi <rodrigo.vivi@gmail.com>
10 years agokms_flip: Add test for -ENOENT
Ville Syrjälä [Wed, 16 Oct 2013 20:07:34 +0000 (23:07 +0300)]
kms_flip: Add test for -ENOENT

Make sure we get -ENOENT when we pass a non-existing fb_id to
the page flip and setcrtc ioctls.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agokms_cursor_crc: Add a cursor test using CRCs
Ville Syrjälä [Fri, 18 Oct 2013 14:44:42 +0000 (17:44 +0300)]
kms_cursor_crc: Add a cursor test using CRCs

Use the display CRC support to validate cursor plane functionality.
The test will position the cursor plane either fully onscreen,
partially onscreen, or fully offscreen, using either a fully opaque
or fully transparent surface. In each case it then reads the PF CRC
and compares it with the CRC value obtained when the cursor plane
was disabled.

v2: Loop over all connectors and crtcs
    Use igt_wait_for_vblank
    Test more corner cases
    Add some progress indication ala. kms_flip
    Make it work for pf and pipe sources

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agolib: Check pipe source validity in igt_pipe_crc_new()
Ville Syrjälä [Mon, 21 Oct 2013 15:45:03 +0000 (18:45 +0300)]
lib: Check pipe source validity in igt_pipe_crc_new()

Have igt_pipe_crc_new() check whether the selected source is actually
supported.

v2: Make debugfs_pipe_crc.c not crash

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agolib: Allow pipe_crc_free(NULL)
Ville Syrjälä [Mon, 21 Oct 2013 16:37:33 +0000 (19:37 +0300)]
lib: Allow pipe_crc_free(NULL)

Prevent pipe_crc_free() from segfaulting on NULL ptr.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agolib/drmtest: Add kmstest_paint_color_alpha()
Ville Syrjälä [Fri, 18 Oct 2013 18:43:14 +0000 (21:43 +0300)]
lib/drmtest: Add kmstest_paint_color_alpha()

kmstest_paint_color_alpha() just like kmstest_paint_color() except you
also get to specify the alpha channel.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
10 years agointel_reg_dumper: fix "enable" at hsw_debug_lp_wm
Paulo Zanoni [Mon, 21 Oct 2013 15:19:57 +0000 (13:19 -0200)]
intel_reg_dumper: fix "enable" at hsw_debug_lp_wm

Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
10 years agogitignore: Ignore intel_opregion_decode
Damien Lespiau [Mon, 21 Oct 2013 14:55:52 +0000 (15:55 +0100)]
gitignore: Ignore intel_opregion_decode

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agointel_opregion_decode: decipher bclm table
Jani Nikula [Tue, 8 Oct 2013 18:18:14 +0000 (21:18 +0300)]
intel_opregion_decode: decipher bclm table

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
10 years agointel_opregion_decode: new tool for decoding graphics opregion
Jani Nikula [Tue, 8 Oct 2013 18:18:13 +0000 (21:18 +0300)]
intel_opregion_decode: new tool for decoding graphics opregion

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
10 years agointel_bios_reader: dump all sections, including unknown ones
Jani Nikula [Tue, 8 Oct 2013 18:15:28 +0000 (21:15 +0300)]
intel_bios_reader: dump all sections, including unknown ones

We still need to dump some of the known sections explicitly due to
dependencies on information extracted, such as LFP data pointers and
panel_type.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
10 years agointel_bios_reader: dump all edp device info structs
Jani Nikula [Tue, 8 Oct 2013 18:15:29 +0000 (21:15 +0300)]
intel_bios_reader: dump all edp device info structs

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
10 years agointel_bios_reader: prepare for dumping all sections
Jani Nikula [Tue, 8 Oct 2013 18:15:27 +0000 (21:15 +0300)]
intel_bios_reader: prepare for dumping all sections

Dump sections through a table based on the section id.

Hex dump the section. This works also for unknown sections.

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
10 years agointel_bios_reader: add size temp variable as a shorthand for finfo.st_size
Jani Nikula [Tue, 8 Oct 2013 18:15:26 +0000 (21:15 +0300)]
intel_bios_reader: add size temp variable as a shorthand for finfo.st_size

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
10 years agointel_error_decode: Fix X/Y fence for gen2/3
Chris Wilson [Mon, 21 Oct 2013 08:31:18 +0000 (09:31 +0100)]
intel_error_decode: Fix X/Y fence for gen2/3

10 years agotests/kms_flip: More leeway for dummy load tests
Daniel Vetter [Sun, 20 Oct 2013 18:37:21 +0000 (20:37 +0200)]
tests/kms_flip: More leeway for dummy load tests

At least on my old atom i945 they take forever. Also, metric tons of
pipe B underruns while running rcs-flip-vs-panning. No idea why that
happens.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agokmstest: don't pollute stderr for impossible output configs
Daniel Vetter [Sun, 20 Oct 2013 17:24:08 +0000 (19:24 +0200)]
kmstest: don't pollute stderr for impossible output configs

Otherwise on machines where the only output available is restricted to
one pipe we'll have tons of "warn" results for no gain in the piglit
runner. All tests that use the kmstest helpers already check the
return value and do something sensible (like skip the tests if there's
really no config available).

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/kms_flip: Unconfuse the framenumber step checking
Daniel Vetter [Sun, 20 Oct 2013 14:11:49 +0000 (16:11 +0200)]
tests/kms_flip: Unconfuse the framenumber step checking

We run flips and vblank waits in parallel, which means the step
for each is the maximum. If we'd switch to an interleaved scheme
(to be able to check the correlation) we'd need to add them instead.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/kms_flip: no absolute vblank wait on the first frame
Daniel Vetter [Sun, 20 Oct 2013 14:10:32 +0000 (16:10 +0200)]
tests/kms_flip: no absolute vblank wait on the first frame

We don't have a reference value for the vblank counter yet, so
will end up waiting for a long time for the vblank counter to
wrap around again.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotets/kms_flip: disable correlation check
Daniel Vetter [Sat, 19 Oct 2013 14:21:18 +0000 (16:21 +0200)]
tets/kms_flip: disable correlation check

It's broken since it doesn't take the order the events arrived into
account and so will fall over for longer seq_step values in the vblank
event: Since the flip completes right away, but the vblank later on
they won't have the same timestamp.

For now just give up, we could resurrect this by strictly alternating
between a flip and a vblank wait.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/debugfs_pipe_crc: new subtest to check CRC frame numbers
Daniel Vetter [Fri, 18 Oct 2013 16:43:30 +0000 (18:43 +0200)]
tests/debugfs_pipe_crc: new subtest to check CRC frame numbers

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agorendercopy: Add a way to dump an .aub file with the rendercopy bos
Damien Lespiau [Wed, 27 Feb 2013 14:51:29 +0000 (14:51 +0000)]
rendercopy: Add a way to dump an .aub file with the rendercopy bos

v2 (by Ben): Remove libdrm dependency since intel-gpu-tools now requires
a higher version anyway. Remove associated #ifdef ENABLE_AUB_DUMP

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
Signed-off-by: Ben Widawsky <benjamin.widawsky@intel.com>
10 years agolib: fix the assert in igt_stop_helper
Daniel Vetter [Fri, 18 Oct 2013 15:53:05 +0000 (17:53 +0200)]
lib: fix the assert in igt_stop_helper

So much for developing different patches on different machines and
then not retesting after rebasing. Reported by Ben on irc.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib: Don't wait for a vblank when enabling the CRCs
Damien Lespiau [Fri, 18 Oct 2013 13:29:20 +0000 (14:29 +0100)]
lib: Don't wait for a vblank when enabling the CRCs

This was a previous attempt to solve the first CRC being bogus. We know
wait for it and discard it at the end of the function.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/debugfs_pipe_crc: Test the read CRCs are not null
Damien Lespiau [Fri, 18 Oct 2013 11:27:21 +0000 (12:27 +0100)]
tests/debugfs_pipe_crc: Test the read CRCs are not null

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agotests/debugfs_pipe_crc: test all connectors
Daniel Vetter [Thu, 17 Oct 2013 12:18:35 +0000 (14:18 +0200)]
tests/debugfs_pipe_crc: test all connectors

Diff looks a bit ugly, but it's just due to the added indentation.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/debugfs_pipe_crc: test all pipes
Daniel Vetter [Thu, 17 Oct 2013 12:10:43 +0000 (14:10 +0200)]
tests/debugfs_pipe_crc: test all pipes

And skip them if there's no suitable connector.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/debugfs_pipe_crc: fix fb leak
Daniel Vetter [Thu, 17 Oct 2013 12:09:52 +0000 (14:09 +0200)]
tests/debugfs_pipe_crc: fix fb leak

Also remove the unused connector_t->mode.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/debugfs_pipe_crc: fall back to PIPE source
Daniel Vetter [Wed, 16 Oct 2013 20:49:24 +0000 (22:49 +0200)]
tests/debugfs_pipe_crc: fall back to PIPE source

With PLANE1 and PIPE CRC sources the test will work on all currently
shipping (and planed fwiw) platforms.

Also add all the other new sources for non-ivb/hsw chips.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agolib/drmtest: Check that helper processes have died correctly
Daniel Vetter [Wed, 16 Oct 2013 17:52:00 +0000 (19:52 +0200)]
lib/drmtest: Check that helper processes have died correctly

If the test gets stopped sometimes a helper process falls over.
We need to report this. Since we currently don't track helper
process to precisely we can't shut up the 2nd test failure messge.
This shouldn't happen anyway.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/ZZ_missed_irq: Fixup SOURCE_PATH handling
Daniel Vetter [Thu, 17 Oct 2013 08:48:29 +0000 (10:48 +0200)]
tests/ZZ_missed_irq: Fixup SOURCE_PATH handling

Copy the trick from ZZ_hangman of just cd'ing into the old working
directory in a subshell.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=70561
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agogem_ctx_bad_exec: skip vebox when appropriate
Ben Widawsky [Thu, 17 Oct 2013 05:01:34 +0000 (22:01 -0700)]
gem_ctx_bad_exec: skip vebox when appropriate

Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
10 years agoREADME: list some of the dependencies
Jani Nikula [Tue, 8 Oct 2013 18:19:06 +0000 (21:19 +0300)]
README: list some of the dependencies

Signed-off-by: Jani Nikula <jani.nikula@intel.com>
Signed-off-by: Ben Widawsky <benjamin.widawsky@intel.com>
10 years agotests/debugfs_pipe_crc: correctly skip on unsupported platforms
Daniel Vetter [Wed, 16 Oct 2013 12:12:49 +0000 (14:12 +0200)]
tests/debugfs_pipe_crc: correctly skip on unsupported platforms

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agodebugfs_pipe_crc: Let's check CRCs!
Damien Lespiau [Wed, 9 Oct 2013 10:47:43 +0000 (11:47 +0100)]
debugfs_pipe_crc: Let's check CRCs!

Let's add a new test that sets a mode, wait for a few vblanks (3) and
then make sure we read 3 identical CRCs.

Some subtests check for various parsing errors.

In the process, improve the debugfs helpers to deal with CRCs.

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add igt_wait_for_vblank() helper
Damien Lespiau [Fri, 11 Oct 2013 17:31:20 +0000 (18:31 +0100)]
lib: Add igt_wait_for_vblank() helper

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Make igt_debugfs_open() take the mode as argument
Damien Lespiau [Wed, 9 Oct 2013 22:53:15 +0000 (23:53 +0100)]
lib: Make igt_debugfs_open() take the mode as argument

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add a igt_display.h with a few enums and defines from the kernel
Damien Lespiau [Wed, 9 Oct 2013 16:54:11 +0000 (17:54 +0100)]
lib: Add a igt_display.h with a few enums and defines from the kernel

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add kmstest_paint_color()
Damien Lespiau [Wed, 9 Oct 2013 16:42:52 +0000 (17:42 +0100)]
lib: Add kmstest_paint_color()

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add a igt_assert_cmpint()
Damien Lespiau [Wed, 9 Oct 2013 12:42:13 +0000 (13:42 +0100)]
lib: Add a igt_assert_cmpint()

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add igt_debugfs_fopen()
Damien Lespiau [Wed, 9 Oct 2013 10:45:31 +0000 (11:45 +0100)]
lib: Add igt_debugfs_fopen()

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agolib: Add a small helper to open debugfs files
Damien Lespiau [Tue, 8 Oct 2013 22:39:33 +0000 (23:39 +0100)]
lib: Add a small helper to open debugfs files

Signed-off-by: Damien Lespiau <damien.lespiau@intel.com>
10 years agoDepend on libdrm_intel >= 2.4.47.
Matt Turner [Sat, 12 Oct 2013 00:34:13 +0000 (17:34 -0700)]
Depend on libdrm_intel >= 2.4.47.

Reviewed-by: Ben Widawsky <ben@bwidawsk.net>
10 years agoconfigure: Don't bail if libdrm_nouveau isn't available.
Matt Turner [Fri, 11 Oct 2013 03:54:25 +0000 (20:54 -0700)]
configure: Don't bail if libdrm_nouveau isn't available.

We were seriously *requiring* libdrm_nouveau unless explicitly disabled?

Acked-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Chad Versace <chad.versace@linux.intel.com>
10 years agotests/pc8: Readd verbose error message
Daniel Vetter [Mon, 14 Oct 2013 17:10:56 +0000 (19:10 +0200)]
tests/pc8: Readd verbose error message

I deemed them a bit redundant (assuming that developers are the only
ones that look at them anyway). But Paulo requested that I readd them
for the education of QA.

Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
10 years agotests/kms_flip: fix igt_assert
Daniel Vetter [Fri, 11 Oct 2013 18:53:40 +0000 (20:53 +0200)]
tests/kms_flip: fix igt_assert

I've blown the conversion in

commit 37e5b318c5dac290611099e2eec21df1e49fa71f
Author: Daniel Vetter <daniel.vetter@ffwll.ch>
Date:   Fri Oct 4 18:42:26 2013 +0200

    tests/kms_flip: use igt_assert more

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=70332
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>