Bug 79359 - [ALL IGT] igt/gem_evict_everything/forked-swapping-multifd-mempressure-normal costs long time to execute
Summary: [ALL IGT] igt/gem_evict_everything/forked-swapping-multifd-mempressure-normal...
Status: CLOSED FIXED
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/Intel (show other bugs)
Version: unspecified
Hardware: Other All
: high normal
Assignee: Intel GFX Bugs mailing list
QA Contact: Intel GFX Bugs mailing list
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2014-05-28 09:07 UTC by Guo Jinxian
Modified: 2017-08-14 08:33 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:

Attachments
dmesg (81.98 KB, text/plain)
2014-05-28 09:07 UTC, Guo Jinxian 		no flags Details
View All

Description Guo Jinxian 2014-05-28 09:07:18 UTC 
Created attachment 100022 [details]
dmesg

==System Environment==
--------------------------
Regression: No. 
The cases was disabled because Bug 69247

Non-working platforms: All

==kernel==
--------------------------
origin/drm-intel-nightly: 9f53d4f6f55aa0c037f299dbe2986eec9151be9b(fails)
    drm-intel-nightly: 2014y-05m-28d-00h-00m-47s integration manifest
origin/drm-intel-next-queued: 38d811545c61cf188beac860615a08fb32f73eb1(fails)
    drm/i915/vlv: add pll assertion when disabling DPIO common well   
origin/drm-intel-fixes: d23db88c3ab233daed18709e3a24d6c95344117f(fails)
    drm/i915: Prevent negative relocation deltas from wrapping

==Bug detailed description==
-----------------------------
igt/gem_evict_everything/forked-swapping-multifd-mempressure-normal costs long time to execute 

Output:
time ./gem_evict_everything --run-subtest forked-swapping-multifd-mempressure-normal
IGT-Version: 1.6-gff3c122 (x86_64) (Linux: 3.15.0-rc3_drm-intel-next-queued_38d811_20140528+ x86_64)
Subtest forked-swapping-multifd-mempressure-normal: SUCCESS

real    16m45.067s
user    0m24.266s
sys     0m45.563s

==Reproduce steps==
---------------------------- 
1. ./gem_evict_everything --run-subtest forked-swapping-multifd-mempressure-normal
Comment 1 Guo Jinxian 2014-06-03 03:56:41 UTC 
Cases below have the same issue on latest -next-queued(192155025197cc4765702a180904c3b62c152b7a)
igt/gem_evict_everything/forked-swapping-mempressure-interruptible
igt/gem_evict_everything/forked-swapping-mempressure-normal
igt/gem_evict_everything/forked-swapping-multifd-mempressure-interruptible
Comment 2 Guo Jinxian 2014-09-11 07:32:03 UTC 
This bug still able to reproduce on hsw on latest -nightly(99f444f7f8308ca93d56a3802a3e1a8abcad7f57)

[root@x-hsw24 tests]# ./gem_evict_everything --run-subtest forked-swapping-multifd-normal
IGT-Version: 1.8-g107151c (x86_64) (Linux: 3.17.0-rc4_drm-intel-nightly_99f444_20140910+ x86_64)
^C
[root@x-hsw24 tests]# time ./gem_evict_everything --run-subtest forked-swapping-multifd-normal
IGT-Version: 1.8-g107151c (x86_64) (Linux: 3.17.0-rc4_drm-intel-nightly_99f444_20140910+ x86_64)


^C^Z
[1]+  Stopped                 ./gem_evict_everything --run-subtest forked-swapping-multifd-normal

real    12m21.690s
user    0m0.000s
sys     0m0.000s
Comment 3 Chris Wilson 2014-12-21 19:51:02 UTC 
Is this fixed by Thomas's mlock patches?
Comment 4 Yi Sun 2014-12-25 01:31:34 UTC 
(In reply to Chris Wilson from comment #3)
> Is this fixed by Thomas's mlock patches?

The issue is gone on latest -nightly branch.

BTW, the Thomas's patch Chris mentioned is following one?
But it's not fixed by this patch by double checking with this patch.
Anyway the case is passed now, close it.

commit 42b02c284ed24871528df8f1b3eaad7fe1554fd9
Author: Thomas Wood <thomas.wood@intel.com>
Date:   Mon Dec 8 11:12:51 2014 +0000

    lib: add a function to lock memory into RAM

    Add a function to lock memory into RAM and use it in the
    gem_tiled_swapping test to reduce the amount of allocated memory
    required to force swapping. This also reduces the amount of time
    required for the test to complete, since the data set is smaller.


root@x-bytm02:/GFX/Test/Intel_gpu_tools/intel-gpu-tools/tests# time ./gem_evict_everything --run-subtest forked-swapping-multifd-mempressure-normal
IGT-Version: 1.9-geb799b2 (x86_64) (Linux: 3.18.0_drm-intel-nightly_4fa231_20141224+ x86_64)
Subtest forked-swapping-multifd-mempressure-normal: SUCCESS (83.659s)

real    1m24.458s
user    3m29.701s
sys     2m0.213s
Comment 5 Chris Wilson 2014-12-25 09:40:41 UTC 
(In reply to Yi Sun from comment #4)
> (In reply to Chris Wilson from comment #3)
> > Is this fixed by Thomas's mlock patches?
> 
> The issue is gone on latest -nightly branch.
> 
> BTW, the Thomas's patch Chris mentioned is following one?
> But it's not fixed by this patch by double checking with this patch.
> Anyway the case is passed now, close it.
> 
> commit 42b02c284ed24871528df8f1b3eaad7fe1554fd9
> Author: Thomas Wood <thomas.wood@intel.com>
> Date:   Mon Dec 8 11:12:51 2014 +0000
> 
>     lib: add a function to lock memory into RAM

That's the test I was thinking of. Could you run a reverse bisect to see what did actually fix it? For curiosity more than anything, so don't worry if it turns out to be too tricky.
Comment 6 Jari Tahvanainen 2017-08-14 08:33:00 UTC 
Moving old bug from Verified to Closed.

Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.