Paulo Zanoni [Wed, 24 Aug 2011 17:57:31 +0000 (14:57 -0300)]
gem_vmap_blits: fix compilation if I915_PARAM_HAS_VMAP is not defined
Previously, "make check" failed because the main() function was not
defined.
Signed-off-by: Paulo Zanoni <paulo.r.zanoni@intel.com>
Tested-By: Eugeni Dodonov <eugeni.dodonov@intel.com>
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Eugeni Dodonov [Mon, 5 Sep 2011 23:04:17 +0000 (20:04 -0300)]
intel_gpu_top: access hardware before running profiling command
Without this patch, intel_gpu_top will try to get access to the hardware
after running the profiling command in the background. In case such access
fails, the main process quits, but the background one will continue
running.
So let's move this test up, so it is done before we fork.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 22:59:40 +0000 (19:59 -0300)]
intel_gpu_tools: describe -e option in usage screen
This patch adds a short description of '-e' option to the usage screen.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 22:50:52 +0000 (19:50 -0300)]
intel_gpu_top: adopt to kernel coding style
This patch fixes adopts the code to the kernel coding_style.
Eugeni Dodonov [Mon, 5 Sep 2011 22:41:24 +0000 (19:41 -0300)]
intel_gpu_top: support profiling user-specified commands
This patch adds support for running intel_gpu_top to profile specific
commands. The required command will be carried out in separate process,
and main intel_gpu_top will leave when the child process will exit.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 22:39:58 +0000 (19:39 -0300)]
intel_gpu_top: support non-interactive mode
This patch adds support for non-interactive mode, invoked by running with
'-o output' switch. In this case, no interactive output is being
performed, but the execution statistics are being saved into the output
file.
The output file is generated in both human and gnuplot-readable format.
Unlike interactive mode, where non-supported pipes and non-active
registers are skipped, the content of such pipes and registers is recorded
into the log file to simplify parsing and standardize the list of columns.
Also, unlike interactive mode, the registers are not sorted according to
the usage - this way, their variation over time can be analysed offline.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 22:35:44 +0000 (19:35 -0300)]
This patch initializes the last_stats[] for registers prior to starting
the monitoring itself. This way, the first measure will already contain
the difference from the previous value instead of non-initialized value.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 22:33:40 +0000 (19:33 -0300)]
Revert "intel_gpu_top: initialize monitoring statistics at startup"
This reverts commit
431fe7803d37f344275fdaceb57c1fbaeee8541c.
Will be replaced with more fine-grained commits.
Eugeni Dodonov [Mon, 5 Sep 2011 19:33:04 +0000 (16:33 -0300)]
intel_gpu_top: initialize monitoring statistics at startup
This patch initializes the last_stats[] for registers prior to starting
the monitoring itself. This way, the first measure will already contain
the difference from the previous value instead of non-initialized value.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 19:29:39 +0000 (16:29 -0300)]
intel_gpu_tool: initial support for non-screen output
This patch adds initial support for non-stdio output, to be used for
non-interactive monitoring.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 19:23:11 +0000 (16:23 -0300)]
intel_gpu_top: suport command line parameters and variable samples per
second
This patch adds support for getopt, and adds two default parameters to it:
-h to show usage notes; and -s to allow user to define number of samples
to acquire per second.
Manpage documentation is also adjusted accordingly.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Eugeni Dodonov [Mon, 5 Sep 2011 18:31:17 +0000 (15:31 -0300)]
intel_gpu_top: account for time spent in syscalls
This allows intel_gpu_top to properly account for time spent inside system
calls. Effectively, with previous implementation, intel_gpu_top could
spent longer than 1s between consecutive measures. This attempts to
minimize the extra time spent while polling for data.
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
Jesse Barnes [Fri, 12 Aug 2011 19:51:34 +0000 (12:51 -0700)]
tests: add store dword tests
Add a couple of simple store dword tests to test memory coherence.
gem_storedw_loop simply executes a batch that continually stores an
incremented value to a target buffer object, checking the results after
each batch completes.
gem_storedw_batches_loop does the same thing, but creates a new command
batch buffer for each iteration, which can exercise the buffer creation
code. This test is based on one from Andrzej Kacprowski from Intel.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Hai Lan [Fri, 5 Aug 2011 04:47:25 +0000 (00:47 -0400)]
testdisplay: for tiled mode test, the height should be aligned
Signed-off-by: Hai Lan <hai.lan@intel.com>
Eric Anholt [Mon, 1 Aug 2011 17:39:51 +0000 (10:39 -0700)]
Add missing copyright info to intel_decode.c
Ben Widawsky [Thu, 28 Jul 2011 21:58:24 +0000 (14:58 -0700)]
tools: readme addition for shader debugging
Ben Widawsky [Thu, 28 Jul 2011 22:07:28 +0000 (15:07 -0700)]
tools/range handing: forgot the register map file :(
Ben Widawsky [Fri, 24 Jun 2011 18:42:03 +0000 (11:42 -0700)]
intel-gpu-tools/debugging: shader debugger
high level summary of the files:
* debug_rdata - get current state from debug registers. Helpful
when developing the debugger, and could serve some purpose in the
future.
* eudb - the debugger itself
* eviction_macro - generate the proper macro to flush the EU render
cache until I get control flow working
* pre_cpp - an evaluating c preprocesser like thing, to be used before
cpp
* sr - the system routine, exception handler which runs on the EU
* test - a very basic test system routine
* debug.h
Ben Widawsky [Sun, 17 Jul 2011 19:43:12 +0000 (12:43 -0700)]
intel-gpu-tools/debugging: add important debug regs
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Ben Widawsky [Tue, 14 Jun 2011 18:00:30 +0000 (11:00 -0700)]
intel-gpu-tools/forcewaked: sample forcewake app
This app is required for debug features which seem to (undocumented)
reset themselves if/when the GT goes to sleep.
It is also useful for those doing general debugging or profiling from
userspace when they do not wish to have the GT sleep.
v2: made forcewaked more daemon-like
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Ben Widawsky [Thu, 28 Jul 2011 20:42:45 +0000 (13:42 -0700)]
intel-gpu-tools/range handling: register range handling
Hooks to allow safe accesses from userspace. Can revert to old behavior
by using unsafe access.
Ben Widawsky [Thu, 28 Jul 2011 20:40:19 +0000 (13:40 -0700)]
forcewake: Add mmio code to do proper forcewake stuff for gen6
Chris Wilson [Wed, 20 Jul 2011 19:40:13 +0000 (20:40 +0100)]
Add intel_backlight
A trivial tool to adjust the backlight registers (gen5+ only atm).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Wed, 20 Jul 2011 19:39:56 +0000 (20:39 +0100)]
intel_decode: Handle a few gen7 blocks
Daniel Vetter [Sun, 17 Jul 2011 21:33:27 +0000 (23:33 +0200)]
gem_stress: ducttaping mandatory
In other news: We've been missing a unmapping_mapping_range somewhere
in the kernel. But lazy me never came around to digging up the real
cause.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Hai Lan [Tue, 31 May 2011 21:13:30 +0000 (17:13 -0400)]
testdisplay: Test the first mode if there is no preferred mode.
For a TV device, there might be no preferred mode. In this case,
we can test the first mode.
Signed-off-by: Hai Lan <hai.lan@intel.com>
Chris Wilson [Fri, 8 Jul 2011 14:01:39 +0000 (15:01 +0100)]
tests/gem_vmap_blits: Fix typo
Why, oh why, do these only become obvious after pushing upstream?
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Fri, 8 Jul 2011 13:55:36 +0000 (14:55 +0100)]
intel_reg_dumper: Slightly tidy the reporting of pipe configs
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Fri, 8 Jul 2011 13:54:32 +0000 (14:54 +0100)]
Add a test case for coherency of vmaps
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Wed, 25 May 2011 13:49:40 +0000 (14:49 +0100)]
gem_stress: Assert we have no pending ops when using the CPU
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Wed, 25 May 2011 13:40:38 +0000 (14:40 +0100)]
gem_gtt_speed: Add option for variable buffer size
Chris Wilson [Sat, 2 Jul 2011 07:59:32 +0000 (08:59 +0100)]
gem_stress: Add an option to test handling of signals
As signals cause the syscalls to be interrupted, we often need to clean
up partial state before returning to userspace. Often a source of
unamusing bugs, so encourage gem_stress to provoke them.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Thu, 23 Jun 2011 13:04:35 +0000 (14:04 +0100)]
tests: Add an exercise for zero-length buffers
Triggers an OOPS with dmar enabled currently.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Mon, 20 Jun 2011 12:53:12 +0000 (13:53 +0100)]
tests: Add a simple exercise of GTT mmaps
Test copying between 2 mappings and reading/writing from and to.
References: https://bugs.freedesktop.org/show_bug.cgi?id=38115
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Daniel Vetter [Mon, 13 Jun 2011 20:02:52 +0000 (22:02 +0200)]
gem_stress: some stats about incoherencies
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 13 Jun 2011 18:14:53 +0000 (20:14 +0200)]
gem_stress: option to disable render copyfunc check
Too noisy - which is an another way of saying too broken :(
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 13 Jun 2011 18:09:38 +0000 (20:09 +0200)]
gem_stress: option for tile_size
I want to know how large these corruptions can get!
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 13 Jun 2011 11:18:35 +0000 (13:18 +0200)]
gem_stress: option to set tiles per buffer
gem_stress -p1 is much more evil than gem_stress -c1, it also manages
to tear appart untiled workloads!
Now duct-taping over it still works (--apply-duct-tape) ... hm.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 12 Jun 2011 22:22:01 +0000 (00:22 +0200)]
gem_stress: option to continue despite failures
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Tue, 7 Jun 2011 20:30:03 +0000 (22:30 +0200)]
gem_stress: option to override the number of rounds to complete
gem_stress -c 1
... sometimes takes a while to hit a problem.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson [Wed, 8 Jun 2011 11:13:46 +0000 (12:13 +0100)]
scripts: Add throttle.py
Parses a trace.dat and works out how long each throttle was and how many
batches retired within that period (and their average duration).
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Jesse Barnes [Tue, 7 Jun 2011 20:32:37 +0000 (13:32 -0700)]
testdisplay: automatically enable TEST_PLANES if possible
Just look for the ADDFB2 ioctl and enable the new code if it exists.
Jesse Barnes [Tue, 7 Jun 2011 20:28:19 +0000 (13:28 -0700)]
testdisplay: fix error output if test pattern draw fails
Jesse Barnes [Tue, 7 Jun 2011 20:27:35 +0000 (13:27 -0700)]
testdisplay: add tiled buffer test support
Useful for testing tiled vs linear framebuffers.
Jesse Barnes [Tue, 7 Jun 2011 20:17:46 +0000 (13:17 -0700)]
testdisplay: add overlay plane test option
Enumerate available extra planes and try to use one on each CRTC we enable.
Lan, Hai [Tue, 7 Jun 2011 02:16:26 +0000 (19:16 -0700)]
testdisplay: fix possible_crtcs check so LVDS works
When using testdisplay on GM965 and Pineview with LVDS, it will fail to
set a mode because the first unused crtc can't be used for LVDS. So
check the possible_crtcs to make sure the crtc can be used.
Signed-off-by: Hai Lan <hai.lan@intel.com>
Daniel Vetter [Mon, 6 Jun 2011 18:16:35 +0000 (20:16 +0200)]
gem_stress: avoid division by 0
Without this height would be 16 with the minimal buffer size.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 6 Jun 2011 18:15:44 +0000 (20:15 +0200)]
gem_stress: disable fence storm when only using untiled buffers
This explains why gem_stress -u worked beforehand - the rendercpy
was not actually used!
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 6 Jun 2011 16:59:54 +0000 (18:59 +0200)]
gem_stress: round max_dimension down to the next pot
Creates funny rounding problems otherwise.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson [Sun, 5 Jun 2011 20:53:49 +0000 (21:53 +0100)]
gem_stress: trim buffer sizes to fit into 3D pipeline
Remember the 3D pipeline is much more restricted than the BLT engine,
and we were feeding it buffers much larger than either the
render engine or the sampler could manager.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 20:35:34 +0000 (21:35 +0100)]
gem_stress: cpu maps force tiling to be disabled, so fix the printf
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 20:34:07 +0000 (21:34 +0100)]
gem_stress: Add extra passes to intial render copyfunc verification
...and allow for cpu maps!
Chris Wilson [Sun, 5 Jun 2011 18:10:14 +0000 (19:10 +0100)]
test/gen3_mixed_blits: Remember that the BLT engine cannot handle Y-tiling
...even through a fence that can.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 17:05:28 +0000 (18:05 +0100)]
test/gen3_mixed_blits: Alternately use fence regs for the render copies
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 16:45:28 +0000 (17:45 +0100)]
test/gen3_mixed_blits: Acutally use fences for BLT
Gah, in my excitement of reproducing the failure reported by
gem_stress, I missed using fenced relocs for the BLT.
Fortunately, it doesn't affect the presence of the error.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 17:13:18 +0000 (18:13 +0100)]
decode: Fix segmentation fault
The decode for A8 is incomplete and died with a segfault. First hide the
segfault, latter fix the pretty printing.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 16:23:09 +0000 (17:23 +0100)]
test: Add more gen3 render blitters
... and we have a winner: gen3_mixed_blits reproduces the issue Daniel
Vetter originally found. It seems clear that we have some incoherence
between the RENDER and BLT units on gen3 that no amount of MI_FLUSH can
hide. Hmmm....
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 5 Jun 2011 10:20:34 +0000 (11:20 +0100)]
test: Exercise gen3 render copies
gem_stress is unhappy with tiled render copies on gen3. This is a simple
little test to ensure that a set of pure copies with a working set
larger than the aperture are handled correctly.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sat, 4 Jun 2011 20:41:11 +0000 (21:41 +0100)]
tests: Add a fenced execbuffer thrash test
Exercise a nasty corner-case in the reservation logic for the fence
accounting.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Sun Yi [Tue, 31 May 2011 01:20:30 +0000 (18:20 -0700)]
testdisplay: print pixel clock in mode debug output
Print the pixel clock of each mode.
Chris Wilson [Wed, 25 May 2011 12:42:29 +0000 (13:42 +0100)]
Update gitignore
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Wed, 25 May 2011 12:41:28 +0000 (13:41 +0100)]
gem_stress: '-b' option to disable BLT copyfunc
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Chris Wilson [Sun, 22 May 2011 09:34:12 +0000 (10:34 +0100)]
gem_stress: Add render copyfunc for SandyBridge
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Eric Anholt [Fri, 6 May 2011 19:15:50 +0000 (12:15 -0700)]
Add Ivybridge support to intel_gpu_dump and the BLT tests.
Jesse Barnes [Fri, 13 May 2011 00:49:07 +0000 (17:49 -0700)]
testdisplay: add 30bpp support (#ifdef'd out until the next cairo release)
Jesse Barnes [Wed, 11 May 2011 00:21:12 +0000 (17:21 -0700)]
Add Ivybridge device IDs
Makes the reg dumper work better.
Jesse Barnes [Wed, 4 May 2011 21:55:32 +0000 (14:55 -0700)]
testdisplay: don't read hotplug events from buffer before libudev
Otherwise we'll just block rather than trying to update the display.
Signed-off-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Daniel Vetter [Wed, 27 Apr 2011 20:13:29 +0000 (22:13 +0200)]
gem_stress: option to use cpu mappings
This things just won't die (libva!!!), so add an option to test them.
_Not_ meant to test snoopable mappings.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Hai Lan [Tue, 26 Apr 2011 03:03:39 +0000 (11:03 +0800)]
Fix a bug of testing preferred mode
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Jesse Barnes [Tue, 19 Apr 2011 18:50:16 +0000 (11:50 -0700)]
testdisplay: add non-24 bit depth support
Testing 16bpp is useful, and 30bpp will be added once cairo supports it.
Jesse Barnes [Mon, 18 Apr 2011 23:48:27 +0000 (16:48 -0700)]
testdisplay: paint color gradients
Helps identify dithering problems.
Daniel Vetter [Tue, 12 Apr 2011 20:26:42 +0000 (22:26 +0200)]
gem_stress: option to only use x-tiling
This seems to be another trick to massively improve correctness
of the render blit. At least on my i945.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Ben Widawsky [Mon, 11 Apr 2011 15:47:11 +0000 (08:47 -0700)]
Add C99 requirement to autoconf.
snprintf is part of the C99 standard and is used in many places in the
tools.
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Daniel Vetter [Sat, 9 Apr 2011 18:15:10 +0000 (20:15 +0200)]
Disable gem tests when libdrm is not available
Quickly checked on ppc by Cyril Brulebois.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Tue, 5 Apr 2011 19:14:54 +0000 (21:14 +0200)]
intel_decode: drop unused code
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson [Tue, 5 Apr 2011 07:52:52 +0000 (08:52 +0100)]
decode: Correct a couple of mistakes in gen3 IMMEDIATE_STATE_1
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Daniel Vetter [Mon, 4 Apr 2011 20:10:21 +0000 (22:10 +0200)]
gem_stress: option to disable tiling
Of all the things I've tried, this seems to be the only thing
to fix tile corruptions reliably on gen2&gen3 (safe for outright
disabling the render copy).
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 4 Apr 2011 19:24:58 +0000 (21:24 +0200)]
gem_stress: clean up fence_storm
That little snippet creapt in and magically made render copy work -
by essentially disabling it.
Restore order, everything incoherent again.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 3 Apr 2011 11:10:07 +0000 (13:10 +0200)]
gem_stress: extract emit_blt
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 31 Mar 2011 21:30:17 +0000 (23:30 +0200)]
gem_stress: gen2 render copy
Sometimes acts up like gen3 on my i855gm. Sometimes works flawless.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Thu, 31 Mar 2011 21:29:25 +0000 (23:29 +0200)]
gem_stress: split out gen3 render copy
Header conflicts with other headers for different generations.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Wed, 30 Mar 2011 19:46:04 +0000 (21:46 +0200)]
gem_stress: option to disable render copy
It seems to be randomly broken, every boot in a slightly different
way on my i945gme. Works quite well on my Q35. So add an option to
disable it till this is resolved.
Well, more testing seems to suggest that I've been hunting ghosts.
Or maybe not and it works now simply because it's a different day.
Anyway, leave this in for future testing.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 27 Mar 2011 19:33:29 +0000 (21:33 +0200)]
gem_stress: render copy on gen3
Headers copied over from xf86-video-intel, code built after the Xrender
support.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 27 Mar 2011 14:17:54 +0000 (16:17 +0200)]
lib: execbuffer2 support
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 28 Mar 2011 19:42:26 +0000 (21:42 +0200)]
gem_stress: fix set_tiling
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Tue, 29 Mar 2011 17:23:35 +0000 (19:23 +0200)]
gem_stress: ad-hoc tile tracing
Signed-off-by: root <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 27 Mar 2011 13:37:26 +0000 (15:37 +0200)]
gem_stress: optin to set num_buffers
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 27 Mar 2011 13:25:12 +0000 (15:25 +0200)]
gem_stress: variable gpu busy load
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 27 Mar 2011 12:26:43 +0000 (14:26 +0200)]
gem_stress: getoptification
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sun, 27 Mar 2011 12:02:49 +0000 (14:02 +0200)]
gem_stress: dump complete failed tile
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Fri, 25 Mar 2011 21:09:28 +0000 (22:09 +0100)]
Move gen5 blitter work-around to lib
Suggested by Chris Wilson to prevent utterly strange gpu hangs.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Fri, 25 Mar 2011 20:54:02 +0000 (21:54 +0100)]
gem_stress: limit gtt_size to 256
gem_stress maps all buffers, so more only results in trashing (which
should be handled with an option).
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Fri, 25 Mar 2011 20:53:07 +0000 (21:53 +0100)]
gem_stress: implement gen5 blitter work-around
ddx and mesa assume that this is issued after every blit command.
Breaking that invariant results in a dying gpu.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Fri, 25 Mar 2011 19:42:57 +0000 (20:42 +0100)]
gem_stress: fix blitter_copy on gen4+
Utter fail at copy&pasting.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Forest Bond [Tue, 22 Mar 2011 14:31:43 +0000 (10:31 -0400)]
intel_bios_dumper: handle rom_size == 0
Some versions of libpciaccess on Linux set rom_size to 0 for VGA
devices. While this behavior may change, intel_bios_dumper should
handle this situation to be compatible with current versions of the
library.
This fixes segmentation faults on affected systems.
Signed-off-by: Forest Bond <forest.bond@rapidrollout.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Daniel Vetter [Mon, 21 Mar 2011 21:13:18 +0000 (22:13 +0100)]
gem_stress: submit batch if there's only one fence left
... for a copy might use up to two fences!
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Chris Wilson [Mon, 21 Mar 2011 07:58:51 +0000 (07:58 +0000)]
stress: Distinguish between busy_bo and scratch_bo creation sizes
The busy bo has a fixed size (1024x256, 32bpp) whereas the scratch bo
may need to vary their size to exercise different features of fence
allocation.
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
Daniel Vetter [Sun, 20 Mar 2011 20:12:20 +0000 (21:12 +0100)]
gem_stress: pwrite/pread support
And a few other things:
- inline checking when copying tiles with the cpu, fails _much_ faster.
- bo size seems to have a tremendous effect, put on the TODO.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Sat, 19 Mar 2011 20:34:55 +0000 (21:34 +0100)]
Add gem stress test
Not where I want it to be, yet. And not as good as I've hoped in detecting
broken kernels.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Daniel Vetter [Mon, 14 Mar 2011 12:03:48 +0000 (13:03 +0100)]
Decode map state on gen2
At least the important stuff.
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>