gpu: introduce DISABLE_TEXTURE_STORAGE workaround for mesa

gpu: introduce DISABLE_TEXTURE_STORAGE workaround for mesa

Without the workaround, TextureStorageTest.CorrectPixels crashes on mesa drivers.

BUG=521904
Committed: https://crrev.com/360cdb678007b63bfebe7644ce2eac546c99f3af
Cr-Commit-Position: refs/heads/master@{#346665}

[dshwang, 2015-08-19 12:46:59]

dongseong.hwang@intel.com changed reviewers:
+ piman@chromium.org, sievers@chromium.org

[dshwang, 2015-08-19 12:46:59]

could you review?

After mesa bug is fixed in https://bugs.freedesktop.org/show_bug.cgi?id=91673,
I'll revert this CL.

[piman, 2015-08-19 21:07:21]

https://codereview.chromium.org/1301053002/diff/1/gpu/command_buffer/tests/gl...
File gpu/command_buffer/tests/gl_texture_storage_unittest.cc (right):

https://codereview.chromium.org/1301053002/diff/1/gpu/command_buffer/tests/gl...
gpu/command_buffer/tests/gl_texture_storage_unittest.cc:53: // texture bound to
FBO, remove this workaround. crbug.com/521904
Is this something that should be fixed / worked around on the service side? It
sounds like a client could cause a GPU process crash (and then maybe worse?)
with this pattern?

[dshwang, 2015-08-20 07:47:47]

https://codereview.chromium.org/1301053002/diff/1/gpu/command_buffer/tests/gl...
File gpu/command_buffer/tests/gl_texture_storage_unittest.cc (right):

https://codereview.chromium.org/1301053002/diff/1/gpu/command_buffer/tests/gl...
gpu/command_buffer/tests/gl_texture_storage_unittest.cc:53: // texture bound to
FBO, remove this workaround. crbug.com/521904
On 2015/08/19 21:07:21, piman (slow to review) wrote:
> Is this something that should be fixed / worked around on the service side? It
> sounds like a client could cause a GPU process crash (and then maybe worse?)
> with this pattern?

right, client can kill gpu process via this pattern.
let me add workaround as you mentioned

[dshwang, 2015-08-20 15:41:24]

I introduce DISABLE_TEXTURE_STORAGE workaround to prevent client for crashing
gpu process.

Under Linux Mesa 10.5.2 (GLX), it disable texture storage.
ChromeOS Mesa (EGL) 10.6 doesn't crash, so disable only linux.

This bug is assigned and work in progress in mesa project. so I think it will be
fixed soon.
https://bugs.freedesktop.org/show_bug.cgi?id=91673

https://codereview.chromium.org/1301053002/diff/1/gpu/command_buffer/tests/gl...
File gpu/command_buffer/tests/gl_texture_storage_unittest.cc (right):

https://codereview.chromium.org/1301053002/diff/1/gpu/command_buffer/tests/gl...
gpu/command_buffer/tests/gl_texture_storage_unittest.cc:53: // texture bound to
FBO, remove this workaround. crbug.com/521904
On 2015/08/20 07:47:46, dshwang wrote:
> On 2015/08/19 21:07:21, piman (slow to review) wrote:
> > Is this something that should be fixed / worked around on the service side?
It
> > sounds like a client could cause a GPU process crash (and then maybe worse?)
> > with this pattern?
> 
> right, client can kill gpu process via this pattern.
> let me add workaround as you mentioned

Done.

[dshwang, 2015-08-21 10:30:08]

piman, could you review again?

[piman, 2015-08-22 00:11:41]

https://codereview.chromium.org/1301053002/diff/40001/gpu/config/gpu_driver_b...
File gpu/config/gpu_driver_bug_workaround_type.h (right):

https://codereview.chromium.org/1301053002/diff/40001/gpu/config/gpu_driver_b...
gpu/config/gpu_driver_bug_workaround_type.h:50: disable_texture_storage)        
                   \
You added this workaround type, but you're not using it anywhere... So I don't
think this does anything.

[dshwang, 2015-08-26 15:20:14]

sorry fog lagging update.
piman, sievers, could you review again?

https://codereview.chromium.org/1301053002/diff/40001/gpu/config/gpu_driver_b...
File gpu/config/gpu_driver_bug_workaround_type.h (right):

https://codereview.chromium.org/1301053002/diff/40001/gpu/config/gpu_driver_b...
gpu/config/gpu_driver_bug_workaround_type.h:50: disable_texture_storage)        
                   \
On 2015/08/22 00:11:41, piman (OOO back 2015-08-31) wrote:
> You added this workaround type, but you're not using it anywhere... So I don't
> think this does anything.

you're right. the patch set missed feature_info.cc change
I update it in new patch set.

[piman, 2015-08-31 23:36:16]

https://codereview.chromium.org/1301053002/diff/80001/gpu/config/gpu_driver_b...
File gpu/config/gpu_driver_bug_list_json.cc (right):

https://codereview.chromium.org/1301053002/diff/80001/gpu/config/gpu_driver_b...
gpu/config/gpu_driver_bug_list_json.cc:1495: "value": "10.5.2"
Should this be < 10.6 ? According to the mesa bug it's fixed in 10.6, but there
are other 10.5 releases after 10.5.2.

[dshwang, 2015-09-01 13:50:13]

piman, thx for reviewing. could you review again?

https://codereview.chromium.org/1301053002/diff/80001/gpu/config/gpu_driver_b...
File gpu/config/gpu_driver_bug_list_json.cc (right):

https://codereview.chromium.org/1301053002/diff/80001/gpu/config/gpu_driver_b...
gpu/config/gpu_driver_bug_list_json.cc:1495: "value": "10.5.2"
On 2015/08/31 23:36:16, piman (OOO back 2015-08-31) wrote:
> Should this be < 10.6 ? According to the mesa bug it's fixed in 10.6, but
there
> are other 10.5 releases after 10.5.2.

correct. sorry for naive code. Done.

[piman, 2015-09-01 14:49:57]

lgtm

[dshwang, 2015-09-01 15:29:04]

On 2015/09/01 14:49:57, piman (OOO back 2015-08-31) wrote:
> lgtm

thx for rapid reviewing!

[dshwang, 2015-09-01 15:29:09]

The CQ bit was checked by dongseong.hwang@intel.com

[commit-bot: I haz the power, 2015-09-01 15:29:32]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1301053002/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1301053002/100001

[commit-bot: I haz the power, 2015-09-01 15:58:15]

Committed patchset #6 (id:100001)

[commit-bot: I haz the power, 2015-09-01 15:58:52]

Patchset 6 (id:??) landed as
https://crrev.com/360cdb678007b63bfebe7644ce2eac546c99f3af
Cr-Commit-Position: refs/heads/master@{#346665}

[Nico, 2015-09-01 21:26:30]

thakis@chromium.org changed reviewers:
+ thakis@chromium.org

[Nico, 2015-09-01 21:26:31]

A bunch of tests started failing on this (non-main waterfall) bot after this
change:
http://build.chromium.org/p/chromium.fyi/builders/CrWinClang64%28dll%29%20tes...
Do you see why that could happen?

[Nico, 2015-09-01 21:28:29]

On 2015/09/01 21:26:31, Nico wrote:
> A bunch of tests started failing on this (non-main waterfall) bot after this
> change:
>
http://build.chromium.org/p/chromium.fyi/builders/CrWinClang64%28dll%29%20tes...
> Do you see why that could happen?

(If not: The bot has incremental build issues, so maybe it's just due to that.
I'm mostly asking in case it's something obvious to you. If not, don't worry
about it.)

[dshwang, 2015-09-02 08:19:13]

On 2015/09/01 21:28:29, Nico wrote:
> On 2015/09/01 21:26:31, Nico wrote:
> > A bunch of tests started failing on this (non-main waterfall) bot after this
> > change:
> >
>
http://build.chromium.org/p/chromium.fyi/builders/CrWinClang64%28dll%29%20tes...
> > Do you see why that could happen?
> 
> (If not: The bot has incremental build issues, so maybe it's just due to that.
> I'm mostly asking in case it's something obvious to you. If not, don't worry
> about it.)

I don't think it can be related to the failure, because many platforms are fine
without immutable texture supports already.
For some reason, the failure has recovered since Sep 01 22:02.