tests/initial_state: Add a test to capture the state of the GPU
PATCH 1/1 tests/initial_state: Add a test to capture the state of the GPU
When running testlist for CI iot would be good to know if the state of the GPU is as expected. This adds a subtest to check if the GPU is wedged.
--- tests/Makefile.sources | 1 + tests/initial_state.c | 52 +++++++++++++++++++++++++++++++++++ tests/intel-ci/extended.testlist | 1 + tests/intel-ci/fast-feedback.testlist | 1 + 4 files changed, 55 insertions(+) create mode 100644 tests/initial_state.c
tests/Makefile.sources +1 -0
@@ -152,6 +152,7 @@ TESTS_progs_M = \ vgem_basic \ vgem_slow \ meta_test \ + initial_state \ $(NULL) if HAVE_CHAMELIUM
tests/initial_state.c new file +52 -0
52 lines not on this page · raw message
tests/intel-ci/extended.testlist +1 -0
@@ -1,3 +1,4 @@ +igt@initial_state@is_wedged igt@core_auth@many-magics igt@core_getclient igt@core_get_client_auth@master-drop
tests/intel-ci/fast-feedback.testlist +1 -0
@@ -1,3 +1,4 @@ +igt@initial_state@is_wedged igt@core_auth@basic-auth igt@core_prop_blob@basic igt@drv_getparams_basic@basic-eu-total
Signature · 6 lines
2.11.0 _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
The commit name "Add a test to capture the state of the GPU" should likely be renamed to "Add a test that checks the current state of the driver".
Quoted 1 line · Marta Lofstedt
When running testlist for CI iot would be good to know if
iot -> it
Quoted 2 lines
the state of the GPU is as expected. This adds a subtest to check if the GPU is wedged.
This is not mandatory, but how about adding:
"When coupled with piglit's abort-on feature, this makes the piglit results contain the reason for the abort when the driver's initial state is unacceptable, improving debuggability"
With the commit title and the typo fixed, the patch is:
Quoted 95 lines
Signed-off-by Marta Lofstedt <[email protected]>--- tests/Makefile.sources | 1 + tests/initial_state.c | 52 +++++++++++++++++++++++++++++++++++ tests/intel-ci/extended.testlist | 1 + tests/intel-ci/fast-feedback.testlist | 1 + 4 files changed, 55 insertions(+)create mode 100644
tests/initial_state.ctests/Makefile.sources +1
-0@@ -152,6 +152,7 @@ TESTS_progs_M = \ vgem_basic \ vgem_slow \ meta_test \ + initial_state \ $(NULL) if HAVE_CHAMELIUMtests/initial_state.c new file +52
-0@@ -0,0 +1,52 @@ +/* +* Copyright © 2017 Intel Corporation +* +* 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. +* +*/ + +#include <stdio.h> +#include <string.h> +#include "igt.h" + +/* +* The purpose of this test is to capture the +* state of the GPU before running a testlist. +* +* Note, this test currently only checks if +* i915 is wedged. But, it could be extended to report +* on other bad initial states. +*/ + +static void is_wedged (void ) +{ +char buf[128 ]; +int fd =drm_open_driver (DRIVER_INTEL ); + +igt_debugfs_read (fd,"i915_wedged" , buf); +igt_assert (!strstr (buf,"1" )); +} + +igt_main +{ + + igt_subtest("is_wedged") +is_wedged (); +}tests/intel-ci/extended.testlist +1
-0@@ -1,3 +1,4 @@ +igt@initial_state@is_wedged igt@core_auth@many-magics igt@core_getclient igt@core_get_client_auth@master-droptests/intel-ci/fast-feedback.testlist +0
-0
tests/intel-ci/fast-feedback.testlist line 4
Quoted 4 lines
tests/intel-ci/fast-feedback.testlist +1
-0@@ -1,3 +1,4 @@ +igt@initial_state@is_wedged igt@core_auth@basic-auth igt@core_prop_blob@basic igt@drv_getparams_basic@basic-eu-total
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 83 lines · Martin Peres
The commit name "Add a test to capture the state of the GPU" should likely be renamed to "Add a test that checks the current state of the driver".
Quoted 1 line · Marta Lofstedt
When running testlist for CI iot would be good to know if
iot -> it
Quoted 2 lines
the state of the GPU is as expected. This adds a subtest to check if the GPU is wedged.
This is not mandatory, but how about adding:
"When coupled with piglit's abort-on feature, this makes the piglit results contain the reason for the abort when the driver's initial state is unacceptable, improving debuggability"
With the commit title and the typo fixed, the patch is:
Acked-by Martin Peres <[email protected]>Quoted 62 lines
Signed-off-by Marta Lofstedt <[email protected]>--- tests/Makefile.sources | 1 + tests/initial_state.c | 52 +++++++++++++++++++++++++++++++++++ tests/intel-ci/extended.testlist | 1 + tests/intel-ci/fast-feedback.testlist | 1 + 4 files changed, 55 insertions(+) create mode 100644 tests/initial_state.ctests/Makefile.sources +1
-0@@ -152,6 +152,7 @@ TESTS_progs_M = \ vgem_basic \ vgem_slow \ meta_test \ + initial_state \ $(NULL) if HAVE_CHAMELIUMtests/initial_state.c new file +35
-0@@ -0,0 +1,52 @@ +/* +* Copyright © 2017 Intel Corporation +* +* 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. +* +*/ + +#include <stdio.h> +#include <string.h> +#include "igt.h" + +/* +* The purpose of this test is to capture the +* state of the GPU before running a testlist. +* +* Note, this test currently only checks if +* i915 is wedged. But, it could be extended to report +* on other bad initial states.
Quoted 8 lines
tests/initial_state.c line 43
Quoted 8 lines
tests/initial_state.c new file +8
-0+*/ + +static void is_wedged (void ) +{ +char buf[128 ]; +int fd =drm_open_driver (DRIVER_INTEL ); + +igt_debugfs_read (fd,"i915_wedged" , buf);
The ABI for testing whether the device is wedged before use is gem_throttle(). This is encapsulated into igt_require_gem() and any test that depends upon execution on the GPU should be checking for GEM first.
This test should just be igt_gem_require(drm_open_driver(DRIVER_INTEL)); -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 5 lines · Chris Wilson
The ABI for testing whether the device is wedged before use is
gem_throttle(). This is encapsulated intoigt_require_gem()and any test that depends upon execution on the GPU should be checking for GEM first.This test should just be igt_gem_require(drm_open_driver(DRIVER_INTEL));
Well, how do you expect people to spot your nifty tools and refactorings if they don't show up on the m-l, don't get acks or reviews, and don't come with api-docs? At least neither gmail nor google did find that patch submitted anywhere.
Marta, can you pls review Chris' patch:
commit 9518cb59abe35143f15abac2b7ffdb99820ef53c
Author: Chris Wilson <[email protected]> Date: Wed Feb 22 15:24:54 2017 +0000
igt: Start marking up GEM tests that require an alive GPU to function
Best would be to check that all gem tests really use this, and then also add abi docs for this new helper.
Also, since we have igt_require_gem() already, what's the value of this additional testcase? If we want to reboot machines if the gpu died, then we need to have such a test at a higher level, in piglit. What we could do is add a special exit code to igt that indicates a runtime problem (and tells piglit to reboot), essentially PLEASE_REBOOT instead of SKIP (which is what igt_require_gem() results in).
-Daniel
Signature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 8 lines · Daniel Vetter
Quoted 5 lines · Chris Wilson
The ABI for testing whether the device is wedged before use is
gem_throttle(). This is encapsulated intoigt_require_gem()and any test that depends upon execution on the GPU should be checking for GEM first.This test should just be igt_gem_require(drm_open_driver(DRIVER_INTEL));
Well, how do you expect people to spot your nifty tools and refactorings
The ABI was established in
commit 537e73f3f935b917f2f5f9b51499cb29d65e3889
Author: Chris Wilson <[email protected]> Date: Fri Sep 24 17:37:41 2010 +0100
Disable dri2 after forcing fallbacks
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning of testing, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped. Also, we are planning to soon deploy a piglit.conf solution where testing will be terminated on wedged, so I agree that my test isn't really needed.
/Marta
Quoted 35 lines
-----Original Message-----
From: Daniel Vetter [mailto:[email protected]] On Behalf Of Daniel Vetter
Sent: Monday, May 15, 2017 5:01 PM
To: Chris Wilson <[email protected]>; Martin Peres <[email protected]>; Lofstedt, Marta <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 3 lines · Chris Wilson
The ABI for testing whether the device is wedged before use is
gem_throttle(). This is encapsulated intoigt_require_gem()and any test that depends upon execution on the GPU should be checking for GEMfirst.
Quoted 3 lines
This test should just be
igt_gem_require(drm_open_driver(DRIVER_INTEL));Well, how do you expect people to spot your nifty tools and refactorings if they don't show up on the m-l, don't get acks or reviews, and don't come with api-docs? At least neither gmail nor google did find that patch submitted anywhere.
Marta, can you pls review Chris' patch:
commit
9518cb59abe35143f15abac2b7ffdb99820ef53cAuthor: Chris Wilson <[email protected]> Date: Wed Feb 22 15:24:54 2017 +0000igt: Start marking up GEM tests that require an alive GPU to function
Best would be to check that all gem tests really use this, and then also add abi docs for this new helper.
Quoted 11 lines
Also, since we have
igt_require_gem()already, what's the value of this additional testcase? If we want to reboot machines if the gpu died, then we need to have such a test at a higher level, in piglit. What we could do is add a special exit code to igt that indicates a runtime problem (and tells piglit to reboot), essentiallyPLEASE_REBOOTinstead of SKIP (which is whatigt_require_gem()results in).
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 3 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning of testing, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped. Also, we are planning to soon deploy a piglit.conf solution where testing will be terminated on wedged, so I agree that my test isn't really needed.
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate session).
Whilst that may be out of scope for BAT, it should not be completely beyond us for robustness testing.
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 17 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 2 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning oftesting, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 8 lines
Quoted 1 line
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really needed.
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate session).
It doesn't matter if it is broken or not, if we are terminally wedged the rest of the result may be silly. Look for example at CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CI testing should be stopped.
Quoted 7 lines
Whilst that may be out of scope for BAT, it should not be completely beyond us for robustness testing.
-Chris--
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 20 lines · Marta
Quoted 20 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 2 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning oftesting, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 1 line
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really needed.
Quoted 7 lines
Quoted 5 lines
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate session).
It doesn't matter if it is broken or not, if we are terminally wedged the rest of the result may be silly. Look for example at
CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CI testing should be stopped.
The machine didn't hang, it was remotely killed because the run timed out. -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 49 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 12:04 PM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 22 lines · Marta
Quoted 22 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 3 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning oftesting, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 2 lines
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really needed.
Quoted 7 lines
Quoted 5 lines
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate session).
It doesn't matter if it is broken or not, if we are terminally wedged the rest
of the result may be silly. Look for example at
CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CI testing should be stopped.
Quoted 1 line
The machine didn't hang, it was remotely killed because the run timed out.
How do you know that?
Quoted 4 lines
-Chris
--
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 49 lines · Marta
Quoted 49 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 12:04 PM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 22 lines · Marta
Quoted 22 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 3 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning oftesting, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 2 lines
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really needed.
Quoted 7 lines
Quoted 5 lines
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate session).
It doesn't matter if it is broken or not, if we are terminally wedged the rest
of the result may be silly. Look for example at
CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CI testing should be stopped.
Quoted 2 lines
Quoted 1 line
The machine didn't hang, it was remotely killed because the run timed out.
How do you know that?
The dmesg is a stream of flip timeouts until we run out of total BAT runtime (12 minutes + some startup slack).
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 65 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 12:48 PM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 27 lines · Marta
Quoted 27 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 12:04 PM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 15 lines · Marta
Quoted 15 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 3 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning ofQuoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
testing, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 2 lines
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really
needed.
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate
session).
Quoted 10 lines
Quoted 10 lines
Quoted 3 lines
It doesn't matter if it is broken or not, if we are terminally wedged the rest
of the result may be silly. Look for example at
CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CItesting should be stopped.
Quoted 1 line
Quoted 1 line
Quoted 7 lines
Quoted 1 line
Quoted 1 line
The machine didn't hang, it was remotely killed because the run timed
out.
Quoted 1 line
How do you know that?
The dmesg is a stream of flip timeouts until we run out of total BAT runtime (12 minutes + some startup slack).
-Chris
Then look at CI_DRM_2602, wedged at igt@gem_busy@basic-hang-default, after a lot of skipping, we get incomplete result for another test, this time gem_exec_reloc@basic-gtt-cpu-noreloc
So, gem_exec_reloc@basic-cpu-gtt-noreloc and gem_exec_reloc@basic-gtt-cpu-noreloc are falsely getting blamed and my conclusion is that this is due to the permanent wedging started at gem_busy@basic-hang-default. So, to avoid bug reports for gem_exec_reloc@basic-cpu-gtt-noreloc and gem_exec_reloc@basic-gtt-cpu- noreloc the suggestion is to stop testing after we are terminally wedged.
Quoted 3 lines
--
Chris Wilson, Intel Open Source Technology Centre
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 69 lines · Marta
Quoted 69 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 12:48 PM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 27 lines · Marta
Quoted 27 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 12:04 PM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 15 lines · Marta
Quoted 15 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 3 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning ofQuoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
testing, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 2 lines
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really
needed.
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Quoted 6 lines
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate
session).
Quoted 10 lines
Quoted 10 lines
Quoted 3 lines
It doesn't matter if it is broken or not, if we are terminally wedged the rest
of the result may be silly. Look for example at
CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CItesting should be stopped.
Quoted 2 lines
Quoted 2 lines
The machine didn't hang, it was remotely killed because the run timed
out.
Quoted 1 line
How do you know that?
Quoted 7 lines
Quoted 3 lines
The dmesg is a stream of flip timeouts until we run out of total BAT runtime (12 minutes + some startup slack).
-ChrisThen look at
CI_DRM_2602, wedged at igt@gem_busy@basic-hang-default, after a lot of skipping, we get incomplete result for another test, this time gem_exec_reloc@basic-gtt-cpu-norelocSo, gem_exec_reloc@basic-cpu-gtt-noreloc and gem_exec_reloc@basic-gtt-cpu-noreloc are falsely getting blamed and my conclusion is that this is due to the permanent wedging started at gem_busy@basic-hang-default. So, to avoid bug reports for gem_exec_reloc@basic-cpu-gtt-noreloc and gem_exec_reloc@basic-gtt-cpu- noreloc the suggestion is to stop testing after we are terminally wedged.
The conclusion is still wrong though, it doesn't derive from the wedged state itself but from another bug that can be triggered by a recovered gpu hang. The issue is that post-processing isn't yet smart enough to determine the real cause and picks on the victim. We are always going to have that problem in some form, the question is what can be done to make the process smart to avoid false positives or rather just say this is a general bug and be careful not to be precise on the blame. -Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx
Quoted 20 lines · Marta
Quoted 20 lines
-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, May 16, 2017 11:21 AM
To: Lofstedt, Marta <[email protected]>
Cc: Daniel Vetter <[email protected]>; Martin Peres <[email protected]>; [email protected]
Subject: Re: [Intel-gfx] [PATCH i-g-t] tests/initial_state: Add a test to capture the state of the GPUQuoted 2 lines · Marta
I hereby pull-out this patch.
The idea of it was to know if we were already wedged at the beginning oftesting, that would give us information on how to interpret silly results; such that test starting to get skipped and/or we got dmesg-warns/incomplete on tests that usually should be skipped.
Quoted 1 line
Also, we are planning to soon deploy a piglit.conf solution where testing
will be terminated on wedged, so I agree that my test isn't really needed.
Quoted 7 lines
Quoted 5 lines
Not everything is broken by wedged; internally we just use that as an indicator that GEM is hosed. KMS should still work, we must still be able to drive the displays to show the error and keep the servers alive until the data is saved (and hopefully gracefully degrade that we don't have to interrupt their immediate session).
It doesn't matter if it is broken or not, if we are terminally wedged the rest of the result may be silly. Look for example at
CI_DRM_2612, the fi-elk-e7500 is wedged at igt@gem_busy@basic-hang-default, then all test are skipped until gem_exec_reloc@basic-cpu-gtt-noreloc where the machine hangs, but it is a gem test so it should have been skipped, right. My conclusion from seeing this pattern multiple times is that after terminally wedged, silly things can happen, i.e. we can't trust the results, and since we don't want silly bugs, the CI testing should be stopped.
Oh, and note it was running so slowly because of a KMS bug that is quite common.
-Chris
Signature · 5 lines
Chris Wilson, Intel Open Source Technology Centre _______________________________________________ Intel-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/intel-gfx