Re: [Piglit] [PATCH] winsys-framework Default to showing window
Does anyone else want to review this before it goes in? Thanks, Alex -Original Message- From: Marek Olšák [mailto:mar...@gmail.com] Sent: Friday, August 21, 2015 5:09 AM To: Alexander Goins Cc: Ilia Mirkin; piglit@lists.freedesktop.org Subject: Re: [Piglit] [PATCH] winsys-framework Default to showing window Hi Alex, The tests which don't specifically test the behavior of the default framebuffer should use -fbo. I think this is done by marking a test as "concurrent" in all.py. A lot of tests don't do that, because nobody had the time to update all.py and check if it doesn't break them and if it's okay to do that with regard to the purpose of the tests. (i.e. do they specifically test the default framebuffer or something else?) Marek On Sat, Aug 15, 2015 at 1:27 AM, Alexander Goinswrote: > From NVIDIA's side, I don't think it would be that big of a problem to just > keep running our tests with PIGLIT_FORCE_WINDOW=1 or -fbo. However, my > reasoning is that Piglit should default to the 'correct' behavior according > to the OpenGL spec. While the NVIDIA driver is the only one I know of that > actually enforces pixel ownership, any driver theoretically could do so while > still adhering to the spec. > > I agree that having a window pop up for every test is annoying, but using the > default framebuffer without mapping it is undefined behavior. The 'correct' > way to do offscreen rendering would be using FBOs which Piglit supports with > the -fbo argument. Is there a reason why this can't/shouldn't be used for > running Piglit offscreen? I originally removed the ability to render > offscreen with the default framebuffer altogether, but I figured some users > might still desire this behavior so instead decided to leave it in as a > non-default option. > > Thanks, > Alex > > -Original Message- > From: ibmir...@gmail.com [mailto:ibmir...@gmail.com] On Behalf Of Ilia > Mirkin > Sent: Friday, August 14, 2015 8:07 AM > To: Alexander Goins > Cc: piglit@lists.freedesktop.org > Subject: Re: [Piglit] [PATCH] winsys-framework Default to showing > window > > Interesting. PIGLIT_FORCE_WINDOW was introduced in this change: > > commit b797406fb850c5e3c9fdaaebae36127b5e5f8954 > Author: Paul Berry > Date: Tue Jun 11 16:34:32 2013 -0700 > > Add PIGLIT_FORCE_WINDOW environment variable. > > Setting PIGLIT_FORCE_WINDOW=1 causes all piglit windowsystem tests > (those that weren't invoked with the "-fbo" flag) to show the window > prior to calling piglit_display(). Setting PIGLIT_FORCE_WINDOW=0 (the > default) preserves piglit's existing behaviour, which is to only show > the window when the test is run in "manual" mode (with no "-auto" > argument supplied). > > Setting PIGLIT_FORCE_WINDOW=1 slows down the piglit run, and produces > more annoying windows on the screen, but it has been observed to > produce more repeatable results when testing with the nVidia > proprietary driver for Linux. > > Reviewed-by: Chad Versace > > Which explicitly mentions the nvidia blob driver thing. Having more windows > popping up during piglit runs is going to be annoying... and slowing things > down will also be unwelcome. Is there anything else that can be done here? > > -ilia > > On Thu, Aug 13, 2015 at 1:49 PM, agoins wrote: >> winsys-framework incorrectly assumed that it will always have >> ownership of the pixels in its own buffers. If using the default >> framebuffer (i.e., when Piglit is not running in FBO mode,) ownership >> will not be acquired unless the window is mapped. >> >> While this is not typically a problem because buffers are separate, >> NVIDIA's Unified Back Buffer feature makes all windows share a back >> buffer while still conforming to the OpenGL spec. If Piglit does not >> establish pixel ownership, its output will be clobbered by other windows. >> >> While this problem could be easily fixed by specifying >> PIGLIT_FORCE_WINDOW=1 or -fbo, the current default takes advantage of >> undefined behavior in the OpenGL spec. A better solution would be to >> replace PIGLIT_FORCE_WINDOW with PIGLIT_NO_WINDOW, a flag that allows >> tests to be run without mapping a window or using an FBO. The default >> behavior, the, would be to map a window. Really, though, if users >> want to test with offscreen rendering, they should use FBOs with flag -fbo. >> >> More information on pixel ownership here: >> https://www.opengl.org/documentation/specs/version1.1/glspec1.1/node9 >> 4 >> .html >> --- >> .../piglit-framework-gl/piglit_winsys_framework.c| 20 >> ++-- >> 1 file changed, 10 insertions(+), 10 deletions(-) >> >> diff --git a/tests/util/piglit-framework-gl/piglit_winsys_framework.c >> b/tests/util/piglit-framework-gl/piglit_winsys_framework.c >> index c80e972..fcba91e 100644 >> ---
Re: [Piglit] issue with waffle
Ubuntu 14.04 is quite old. Which version of mesa are you running, it seems like either 1) you've built a version too old to support SKL, or 2) you're not actually using the mesa you've built. Can you give use the version of mesa you've built, and also are you setting LIBLG_DRIVERS_PATH and LD_LIBRARY_PATH? Dylan signature.asc Description: PGP signature ___ Piglit mailing list Piglit@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/piglit
[Piglit] issue with waffle
Hi : I am trying to run piglit on SKL-Y with ubuntu 14.04.03 but when ran waffle always got the following error : cd waffle --> [ok] cmake . --> [ok] make--> [ok] make check --> [ok] make check-func --> [fail] [ 50%] Built target gl_basic_test Scanning dependencies of target gl_basic_test_run test: pass: gl_basic.glx_init libGL error: pci id for fd 4: 8086:191e, driver (null) i965_dri.so does not support the 0x191e PCI ID. libGL error: failed to create dri screen libGL error: failed to load driver: i965 test: pass: gl_basic.glx_gl_rgb libGL error: pci id for fd 4: 8086:191e, driver (null) i965_dri.so does not support the 0x191e PCI ID. libGL error: failed to create dri screen libGL error: failed to load driver: i965 test: pass: gl_basic.glx_gl_rgba libGL error: pci id for fd 4: 8086:191e, driver (null) i965_dri.so does not support the 0x191e PCI ID. I build mesa from the source with the following parameters : ./autogen.sh --prefix=/opt/X11R7 --libdir=/opt/X11R7/lib --with-dri-driverdir=/opt/X11R7/lib/dri --with-dri-drivers="i915 i965 swrast" --enable-gles1 --enable-gles2 --enable-shared-glapi --with-egl-platforms=x11,drm,wayland --enable-texture-float --enable-gbm --enable-glx-tls --enable-dri3 --without-gallium-drivers My configuration is : Platform : SKL-Y Ubuntu 14.04.03 Primary display : eDP 3200x1800 I share with you a part of dmesg output rm:drm_mode_debug_printmodeline] Modeline 37:"3200x1800" 48 361310 3200 3248 3280 3680 1800 1802 1807 2045 0x40 0xa [ 42.340235] [drm:drm_mode_getconnector] [CONNECTOR:35:?] [ 42.340332] [drm:drm_mode_getconnector] [CONNECTOR:44:?] [ 42.340334] [drm:drm_helper_probe_single_connector_modes_merge_bits] [CONNECTOR:44:DP-1] [ 42.340335] [drm:intel_dp_detect] [CONNECTOR:44:DP-1] [ 42.340336] [drm:intel_display_power_get] enabling power well 2 [ 42.340338] [drm:skl_disable_dc6] Disabling DC6 [ 42.340351] [drm:skl_set_power_well] Enabling power well 2 [ 42.340356] [drm:intel_display_power_get] enabling DDI B power well [ 42.340358] [drm:skl_set_power_well] Enabling DDI B power well [ 42.340375] [drm:intel_display_power_put] disabling DDI B power well [ 42.340377] [drm:skl_set_power_well] Disabling DDI B power well [ 42.340378] [drm:intel_display_power_put] disabling power well 2 [ 42.340381] [drm:skl_set_power_well] Disabling power well 2 [ 42.340384] [drm:skl_enable_dc6] Enabling DC6 [ 42.340386] [drm:drm_helper_probe_single_connector_modes_merge_bits] [CONNECTOR:44:DP-1] disconnected [ 42.340390] [drm:drm_mode_getconnector] [CONNECTOR:48:?] [ 42.340391] [drm:drm_helper_probe_single_connector_modes_merge_bits] [CONNECTOR:48:HDMI-A-1] [ 42.340392] [drm:intel_hdmi_detect] [CONNECTOR:48:HDMI-A-1] [ 42.340394] [drm:intel_display_power_get] enabling power well 2 [ 42.340396] [drm:skl_disable_dc6] Disabling DC6 [ 42.340408] [drm:skl_set_power_well] Enabling power well 2 [ 42.340412] [drm:intel_display_power_get] enabling DDI B power well [ 42.340414] [drm:skl_set_power_well] Enabling DDI B power well [ 42.340701] [drm:gmbus_xfer] GMBUS [i915 gmbus dpb] NAK for addr: 0050 w(1) [ 42.340702] [drm:gmbus_xfer] GMBUS [i915 gmbus dpb] NAK on first message, retry [ 42.340983] [drm:gmbus_xfer] GMBUS [i915 gmbus dpb] NAK for addr: 0050 w(1) [ 42.340984] [drm:drm_do_probe_ddc_edid] drm: skipping non-existent adapter i915 gmbus dpb [ 42.340986] [drm:intel_display_power_put] disabling DDI B power well [ 42.340987] [drm:skl_set_power_well] Disabling DDI B power well [ 42.340988] [drm:intel_display_power_put] disabling power well 2 [ 42.340990] [drm:skl_set_power_well] Disabling power well 2 [ 42.340991] [drm:skl_enable_dc6] Enabling DC6 Other useful information : $ lsmod | grep i915 i915 1105920 3 drm_kms_helper126976 1 i915 drm 356352 4 i915,drm_kms_helper i2c_algo_bit 16384 1 i915 video 36864 1 i915 $ glxinfo | grep VGA libGL: screen 0 does not appear to be DRI3 capable libGL: pci id for fd 4: 8086:191e, driver i965 libGL: OpenDriver: trying /opt/X11R7/lib/dri/tls/i965_dri.so i965_dri.so does not support the 0x191e PCI ID. libGL error: failed to create dri screen libGL error: failed to load driver: i965 libGL: OpenDriver: trying /opt/X11R7/lib/dri/tls/swrast_dri.so Can you help me to resolve this issue ? Thanks in advance [cid:image001.png@01CF8BAC.3B4C5DD0] Humberto Perez OTC Linux Graphics OpenSource Technology Center iNet 8-645-2968 Before printing this email, consider your Environmental Responsibility! ___ Piglit mailing list Piglit@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/piglit
Re: [Piglit] [PATCH] cl: add clEnqueueMigrateMemObjects
On Thursday 17 September 2015 19:30:41 Jan Vesely wrote: > On Sat, 2015-09-12 at 14:53 +0200, Serge Martin wrote: > > On Thursday 13 August 2015 23:14:23 Serge Martin wrote: > > > > > > Please note that I will made those changes : > > > > s/Migrating the buffer./Migrating the buffer/ > > s/Migrating the buffer with flag./Migrating the buffer with flag/ > > > > in order to remove the extra dot printed in case of error for the > > normal case. > > > > With this, can I get a R-b ? > > > > > --- > > > > > > tests/cl.py| 1 + > > > tests/cl/api/CMakeLists.cl.txt | 1 + > > > tests/cl/api/enqueue-migrate-mem-objects.c | 227 > > > > > > + > > > > > > 3 files changed, 229 insertions(+) > > > create mode 100644 tests/cl/api/enqueue-migrate-mem-objects.c > > > > > > diff --git a/tests/cl.py b/tests/cl.py > > > index 4668ddc..07f5fd4 100644 > > > --- a/tests/cl.py > > > +++ b/tests/cl.py > > > @@ -58,6 +58,7 @@ with profile.group_manager(PiglitCLTest, 'api') > > > > > > as g: > > > g(['cl-api-enqueue-copy-buffer-rect'], > > > > > > 'clEnqueueCopyBufferRect') > > > > > > g(['cl-api-enqueue-read_write-buffer'], > > > > > >'clEnqueueReadBuffer and clEnqueueWriteBuffer') > > > > > > +g(['enqueue-migrate-mem-objects'], > > > 'clEnqueueMigrateMemObjects') > > > > > > g(['cl-api-get-mem-object-info'], 'clGetMemObjectInfo') > > > g(['cl-api-get-image-info'], 'clGetImageInfo') > > > g(['cl-api-retain_release-mem-object'], > > > > > > diff --git a/tests/cl/api/CMakeLists.cl.txt > > > b/tests/cl/api/CMakeLists.cl.txt > > > index b598528..939f1f3 100644 > > > --- a/tests/cl/api/CMakeLists.cl.txt > > > +++ b/tests/cl/api/CMakeLists.cl.txt > > > @@ -21,6 +21,7 @@ piglit_cl_add_api_test (enqueue-copy-buffer > > > enqueue-copy-buffer.c) > > > > > > piglit_cl_add_api_test (enqueue-map-buffer enqueue-map-buffer.c) > > > piglit_cl_add_api_test (enqueue-copy-buffer-rect enqueue-copy > > > > > > -buffer-rect.c) > > > > > > piglit_cl_add_api_test (enqueue-read_write-buffer enqueue > > > > > > -read_write-buffer.c) > > > +piglit_cl_add_api_test (enqueue-migrate-mem-objects enqueue > > > -migrate-mem-objects.c) > > > > > > piglit_cl_add_api_test (retain_release-mem-object retain_release > > > > > > -mem-object.c) > > > > > > piglit_cl_add_api_test (get-mem-object-info get-mem-object-info.c) > > > piglit_cl_add_api_test (get-image-info get-image-info.c) > > > > > > diff --git a/tests/cl/api/enqueue-migrate-mem-objects.c > > > b/tests/cl/api/enqueue-migrate-mem-objects.c > > > new file mode 100644 > > > index 000..eafadce > > > --- /dev/null > > > +++ b/tests/cl/api/enqueue-migrate-mem-objects.c > > > @@ -0,0 +1,227 @@ > > > +/* > > > + * Copyright © 2015 Serge Martin (EdB)> > > + * > > > + * Permission is hereby granted, free of charge, to any person > > > obtaining a > > > + * copy of this software and associated documentation files (the > > > "Software"), > > > + * to deal in the Software without restriction, including without > > > limitation > > > + * the rights to use, copy, modify, merge, publish, distribute, > > > sublicense, > > > + * and/or sell copies of the Software, and to permit persons to > > > whom the > > > + * Software is furnished to do so, subject to the following > > > conditions: > > > + * > > > + * The above copyright notice and this permission notice > > > (including the next > > > + * paragraph) shall be included in all copies or substantial > > > portions of the > > > + * Software. > > > + * > > > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, > > > EXPRESS OR > > > + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF > > > MERCHANTABILITY, > > > + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO > > > EVENT SHALL > > > + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, > > > DAMAGES OR OTHER > > > + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, > > > ARISING > > > + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR > > > OTHER > > > + * DEALINGS IN THE SOFTWARE. > > > + */ > > > + > > > +/** > > > + * @file enqueue-migrate-mem-objects.c > > > + * > > > + * Test API function: > > > + * > > > + * cl_int > > > + * clEnqueueMigrateMemObjects(cl_command_queue command_queue, > > > + * cl_uint num_mem_objects, > > > + * const cl_mem *mem_objects, > > > + * cl_mem_migration_flags flags, > > > + * cl_uint num_events_in_wait_list, > > > + * const cl_event *event_wait_list, > > > + * cl_event *event) > > > + */ > > > + > > > +#include "piglit-framework-cl-api.h" > > > + > > > + > > > +PIGLIT_CL_API_TEST_CONFIG_BEGIN > > > + > > > + config.name = "clEnqueueMigrateMemObjects"; > > > +