Bug 706

Summary: JOGL crashes under Windows 7 (32 bits & 64 bits) with ATI Mobility Radeon HD 6370
Product: [JogAmp] Jogl Reporter: robinhaf
Component: windowsAssignee: Sven Gothel <sgothel>
Status: RESOLVED FIXED    
Severity: critical CC: gouessej, guillaume.lurenbaum, wwalker3
Priority: P1    
Version: 2   
Hardware: pc_all   
OS: windows   
Type: --- SCM Refs:
75b3d37a3b15c071b49609921244bcb62d329fa9 36dc4f5ab6957a4078842c488afb51df2fdc0630
Workaround: ---
Attachments: log made by eclipse
log from eclipse again
Test
Test_dbg
OpenGL Viewer
Debug log
Test from etc folder
Debug log
Test_dbg from etc folder

Description robinhaf 2013-03-15 21:58:37 CET
Created attachment 431 [details]
log made by eclipse

log from test.bak
Error: Could not find or load main class com.jogamp.newt.opengl.GLWindow


Even if I tried demo here:
http://jogamp.org/jogl-demos/www/
The window appeared just for a few miliseconds a and then nothing is happening.
Problem is in all browsers.

I have opengl v4.0.

Tried Java 7u17 and 6u41 (both x86 and x64)
Tried Eclipse x86 and x64
Comment 1 Sven Gothel 2013-03-20 08:42:32 CET
http://jogamp.org/wiki/index.php/Jogl_FAQ#Bugreports_.26_Testing

Pls follow the procedure, which I have updated lately, 
i.e. using the latest aggregated build:
  http://jogamp.org/wiki/index.php/Downloading_and_installing_JOGL#Downloading_the_latest_aggregated_autobuild

I assume you use Windows and an AMD GPU ?

The bug report's test.log and test_dbg.log
will help us determining your system .. and what went wrong, 
hopefully.
Comment 2 Sven Gothel 2013-03-20 10:31:09 CET
Please also update your Windows driver, i.e. the AMD/ATI Catalyst.

Searching the net for 'Windows AMD crash atio6axx.dll DrvPresentBuffers'
gives plenty of results ..

https://ixquick.com/do/search?q=Windows+AMD+crash+wglCreateContextAttribsARB+atio6axx.dll+DrvPresentBuffers

Generic:
  http://www.townswiki.com/wiki/Known_Issues_and_Workarounds#ATI_Driver_Issue
  http://devgurus.amd.com/thread/159238

JOGL:
  http://forum.salemthegame.com/viewtopic.php?f=10&t=5018
  http://pastebin.com/vHbWxDt1

Whatever we do, it shall not result in a crash by simply calling:
  wglCreateContextAttribsARB(..)

This is clearly a Windows/AMD driver issue, which we sadly do not reproduce
on our accessible Windows systems here .. hmm.

Sorry .. I know it sucks .. we wrote a workaround for a multithreading 
NV/Windows driver bug and Mesa8 already, just to be able to suite such machines.
+++

Again, please provide the test.log and test_dbg.log - this is very important
to be able to reproduce ..

Further more, please add detailed system information, like
  - Windows version
  - AMD / OpenGL driver
  - CPU type 
  - .. whatever more you think might be important.

Test w/ 
  http://jogamp.org/deployment/archive/master/gluegen_652-joal_413-jogl_936-jocl_761/

+++

Will keep this bug open ..

+++
Comment 3 Sven Gothel 2013-03-20 10:31:50 CET
-> normal, b/c this is highly likely a AMD/Windows driver bug, see my last comment
Comment 4 robinhaf 2013-03-25 00:51:40 CET
I tried all things on Wednesday, but I forgot about it, so I am posting it now.

I have Win 7 Home, x64, SP1 (latest updates)
GPU ATI Mobility Radeon HD 6370
CPU Intel Core i5 M 480 @ 2.67GHz
Java 7u17 (latest) tried again both x86 and x64.
OpenGL v4.0 (screen from Opengl Extensions Viewer as attachement)

I downloaded this version of jogl:
gluegen_652-joal_413-jogl_936-jocl_761/  from	20-Mar-2013 10:01.

Test and Test_dbg as attachement.

GPU driver updated today, now I have version from 30 April 2011 (newest). I had to download it from HP web. I wasn't able to install drivew from AMD. It said, I have incompatible hardware even if I use autodetect for installation.


I was already searching the Internet all day before I post the problem here.

Anyway, thank for your help. It seems it isn't going to work.
Comment 5 robinhaf 2013-03-25 00:52:50 CET
Created attachment 439 [details]
log from eclipse again
Comment 6 robinhaf 2013-03-25 00:53:21 CET
Created attachment 440 [details]
Test
Comment 7 robinhaf 2013-03-25 00:53:38 CET
Created attachment 441 [details]
Test_dbg
Comment 8 robinhaf 2013-03-25 00:55:27 CET
Created attachment 442 [details]
OpenGL Viewer
Comment 9 Sven Gothel 2013-03-25 01:51:59 CET
(In reply to comment #4)
> I tried all things on Wednesday, but I forgot about it, so I am posting it
> now.
> 
> I have Win 7 Home, x64, SP1 (latest updates)
> GPU ATI Mobility Radeon HD 6370
> CPU Intel Core i5 M 480 @ 2.67GHz
> Java 7u17 (latest) tried again both x86 and x64.
> OpenGL v4.0 (screen from Opengl Extensions Viewer as attachement)
> 
> I downloaded this version of jogl:
> gluegen_652-joal_413-jogl_936-jocl_761/  from	20-Mar-2013 10:01.
> 
> Test and Test_dbg as attachement.
For whatever reason, the JARs were not being picked up.

I have tested the scripts, and they run fine
when invoked as advertised, i.e. "etc\test_dbg.bat"

> 
> GPU driver updated today, now I have version from 30 April 2011 (newest). I
> had to download it from HP web. I wasn't able to install drivew from AMD. It
> said, I have incompatible hardware even if I use autodetect for installation.
> 
> 
> I was already searching the Internet all day before I post the problem here.
> 
> Anyway, thank for your help. It seems it isn't going to work.

If anyhow possible, can you pls run what runs (eclipse - if you don't get the test_dbg.bat
files to work - which is sad), which DEBUG properties ?

"-Djogamp.debug=all" "-Dnativewindow.debug=all" "-Djogl.debug=all" "-Dnewt.debug=all"

That could give us a clue maybe why the crash happens, i.e. invalid drawable or shared context handle.

However, it appears that JOGL's own initialization has passed already,
since the crash happens with your GLCanvas AWT initialization.

Have you tried to run on of our applets on the test-applet page ?

The log file w/ above DEBUG flags would help alot, I would not disregard the issue,
i.e. I would love to fix it if course.

Maybe Wade has a bit more details .. adding him to CC!
Comment 10 Sven Gothel 2013-03-25 01:53:20 CET
Adding Wade and Julien for review ..
Comment 11 Wade Walker 2013-03-25 03:44:29 CET
I looked at the "logs from Eclipse", but they just seemed to be JVM stack traces reporting an illegal access exception in the driver code -- I was assuming you wanted me to look as the "Eclipse guy" :)

Maybe the reporter of this bus is running test_dbg.bat from the wrong directory? If you try to run from within the etc directory for example, it won't find the JARs -- as Sven mentioned, you have to invoke it from above by etc\test_dbg.bat. The log from this should be helpful.
Comment 12 Julien Gouesse 2013-03-25 10:17:56 CET
(In reply to comment #4)
> I tried all things on Wednesday, but I forgot about it, so I am posting it
> now.
> 
> I have Win 7 Home, x64, SP1 (latest updates)
> GPU ATI Mobility Radeon HD 6370
> CPU Intel Core i5 M 480 @ 2.67GHz
> Java 7u17 (latest) tried again both x86 and x64.
> OpenGL v4.0 (screen from Opengl Extensions Viewer as attachement)
> 
> I downloaded this version of jogl:
> gluegen_652-joal_413-jogl_936-jocl_761/  from	20-Mar-2013 10:01.
> 
> Test and Test_dbg as attachement.
> 
> GPU driver updated today, now I have version from 30 April 2011 (newest). I
> had to download it from HP web. I wasn't able to install drivew from AMD. It
> said, I have incompatible hardware even if I use autodetect for installation.
> 
> 
> I was already searching the Internet all day before I post the problem here.
> 
> Anyway, thank for your help. It seems it isn't going to work.

Hi

I renamed this bug report as it wasn't accurate enough. I agree with Sven, this crash is probably caused by a driver bug but if you follow his advises, we might get some information which would be helpful to find a workaround. This kind of crash occurs with any Java binding for the OpenGL API, even Minecraft is concerned. Please go to "Debug Configurations..." -> tab "Arguments" -> textfield entitled "VM Arguments" and copy/paste this:
-Djogamp.debug=all -Dnativewindow.debug=all -Djogl.debug=all -Dnewt.debug=all

Several manufacturers encourage planned obsolescence to sell more GPUs, it's a shame, I understand your frustration. My main desktop computer uses an ATI Radeon 9250 Pro and ATI provides no driver for it under GNU Linux. I had some problems with ATI Radeon Xpress 200 several years ago, it wrongly claimed to support VBOs, it was a real nightmare and I was not alone in this case:
http://fixthe200m.wordpress.com/2006/08/07/research-on-the-xpress-200m/
http://dri.freedesktop.org/wiki/Radeon200M?highlight=%28CategoryHardwareChipset%29

I don't know if your problem comes from that but avoid using display lists.
Comment 13 Julien Gouesse 2013-03-25 10:20:01 CET
@robinhaf, upgrade your BIOS (if possible) so that you can install the latest version of the driver from ATI.
Comment 14 Julien Gouesse 2013-03-25 10:21:52 CET
Updating the driver fixes the bug as you can see there with a game using JOGL 2.0:
http://forum.salemthegame.com/viewtopic.php?t=4778&p=55800
Comment 15 Julien Gouesse 2013-03-27 14:30:32 CET
@robinhaf Please can you post the logs? We would like to implement a workaround but we don't have any similar hardware.
Comment 16 robinhaf 2013-03-27 20:44:51 CET
(In reply to comment #15)
> @robinhaf Please can you post the logs? We would like to implement a
> workaround but we don't have any similar hardware.

Sorry, I didn't have time earlier

I downloaded version from 27-Mar-2013 08:01 from https://jogamp.org/deployment/archive/master/gluegen_652-joal_417-jogl_943-jocl_767/archive/jogamp-all-platforms.7z

I extract the archive, then I go to the folder ...\jogamp-all-platforms\etc and run file test_dbg.bat. Is this right? There is still only 1 line of text in the log.

> Have you tried to run on of our applets on the test-applet page ?

I tried demos here: http://jogamp.org/jogl-demos/www/ and here: http://jogamp.org/deployment/jogamp-next/jogl-test-applets.html. It isn't working.

> -Djogamp.debug=all -Dnativewindow.debug=all -Djogl.debug=all -Dnewt.debug=all

I tried it. Eclipse gives me a huge log to console, so I post it whole as an attachement (debug log.txt)

> upgrade your BIOS (if possible) so that you can install the latest version of the driver from ATI.

Sorry, but I can not afford such a risk. I have version from 6.12.2010 and there is a new version from 23.2.2011.
Comment 17 robinhaf 2013-03-27 20:46:40 CET
Created attachment 443 [details]
Debug log

log from eclipse running with debug configuration:
-Djogamp.debug=all -Dnativewindow.debug=all -Djogl.debug=all -Dnewt.debug=all
Comment 18 Sven Gothel 2013-04-05 01:49:48 CEST
(In reply to comment #16)
> 
> I downloaded version from 27-Mar-2013 08:01 from
> https://jogamp.org/deployment/archive/master/gluegen_652-joal_417-jogl_943-
> jocl_767/archive/jogamp-all-platforms.7z
> 
> I extract the archive, then I go to the folder ...\jogamp-all-platforms\etc
> and run file test_dbg.bat. Is this right? There is still only 1 line of text
> in the log.

as the Wiki says [and Wade and myself tried to point out],
you have to CD to '...\jogamp-all-platforms',
i.e. w/o the etc folder!
Then execute: 'etc\test_dbg.bat'

I see how this can be confusing ..

> 
> > Have you tried to run on of our applets on the test-applet page ?
> 
> I tried demos here: http://jogamp.org/jogl-demos/www/ and here:
> http://jogamp.org/deployment/jogamp-next/jogl-test-applets.html. It isn't
> working.
> 
> > -Djogamp.debug=all -Dnativewindow.debug=all -Djogl.debug=all -Dnewt.debug=all
> 
> I tried it. Eclipse gives me a huge log to console, so I post it whole as an
> attachement (debug log.txt)
> 
thx
Comment 19 Sven Gothel 2013-04-05 02:00:01 CEST
(In reply to comment #17)
> Created attachment 443 [details]
> Debug log
> 
> log from eclipse running with debug configuration:
> -Djogamp.debug=all -Dnativewindow.debug=all -Djogl.debug=all -Dnewt.debug=all

Your log file is truncated :(,
however .. it gives a little bit of information usable.

AWT-EventQueue-0 - WindowWGLContext.createContextARBImpl: 4.0 (Compatibility profile, arb, FBO, hardware) - @creation, handle 0xc70111f9, share 0x0, direct true, wglCreateContextAttribsARB: 0x69065ec0
#

All this seems to be valid, almost looks like that
'wglCreateContextAttribsARB' simply does not work.

If the test_dbg.log or the one from your eclipse run
would be complete - we might could see weather the 
GL vendor /driver is at least valid. But since you don't 
use a dual GPU (Intel/AMD) config .. this should not be the issue ?!

Has your catalyst driver an option to _disable_ threading optimizations ?
If so .. maybe worth a try ..

We would need the full test_dbg.log and your full use case logs.
And even then .. w/o having our [remote] hands on such machine, 
I am afraid I don't see a way to fix this issue .. hmm.
Comment 20 Sven Gothel 2013-04-05 05:20:00 CEST
See Bug 520 - Probably related, if not a duplicate!
Comment 21 Sven Gothel 2013-04-06 21:04:48 CEST
If anybody is able to direct me to get my physical hands on such a buggy 
Windows machine w/ AMD GPU and such driver [incl. purchasing one] - I would appreciate it.

I am just afraid that if I buy a machine similar, the bug will not be exposed
- since all our Windows machines (XP and 7) work well w/ 
AMD desktop and mobile GPUs.

Thanks in advanced for an exact purchase hint.

Of course - you could borrow me such machine for a week as well, 
if possible.
Comment 22 Sven Gothel 2013-04-07 00:47:26 CEST
(In reply to comment #4)
> I tried all things on Wednesday, but I forgot about it, so I am posting it
> now.
> 
> I have Win 7 Home, x64, SP1 (latest updates)
> GPU ATI Mobility Radeon HD 6370
> CPU Intel Core i5 M 480 @ 2.67GHz
> Java 7u17 (latest) tried again both x86 and x64.
> OpenGL v4.0 (screen from Opengl Extensions Viewer as attachement)
> 
> I downloaded this version of jogl:
> gluegen_652-joal_413-jogl_936-jocl_761/  from	20-Mar-2013 10:01.
> 
> Test and Test_dbg as attachement.
> 
> GPU driver updated today, now I have version from 30 April 2011 (newest). I
> had to download it from HP web. I wasn't able to install drivew from AMD. It
> said, I have incompatible hardware even if I use autodetect for installation.
> 
> 
> I was already searching the Internet all day before I post the problem here.
> 
> Anyway, thank for your help. It seems it isn't going to work.

Is that yours ?
  HP ProBook 4720s (XX802EA)
  http://www.hardwarest.com/product/HP-ProBook-4720s-%28XX802EA%29-167737.html
Comment 23 robinhaf 2013-04-08 22:22:30 CEST
> as the Wiki says [and Wade and myself tried to point out],
> you have to CD to '...\jogamp-all-platforms',
> i.e. w/o the etc folder!
> Then execute: 'etc\test_dbg.bat'

I finally run it. Test and test_dbg.
(Test 2, Test_dbg 2)


> Your log file is truncated :(,
> however .. it gives a little bit of information usable.

Sorry, I didn't noticed. I disabled the limit console output in eclipse, so it should now be complete. (Debug log 2)


> Has your catalyst driver an option to _disable_ threading optimizations ?
> If so .. maybe worth a try ..

I didn't find anything like this.


> Is that yours ?
>   HP ProBook 4720s (XX802EA)
>  http://www.hardwarest.com/product/HP-ProBook-4720s-%28XX802EA%29-167737.html

No, I have HP ProBook 4720s (XX838EA).
http://www.hardwarest.com/product/HP-ProBook-4720s-(XX838EA)-167738.html
Comment 24 robinhaf 2013-04-08 22:25:55 CEST
Created attachment 453 [details]
Test from etc folder
Comment 25 robinhaf 2013-04-08 22:29:37 CEST
"The file you are trying to attach is 3307 kilobytes (KB) in size. Attachments cannot be more than 2000 KB."

The other 2 attachements has both more than 3 MB. Where should I post it?
Comment 26 Julien Gouesse 2013-04-08 22:41:00 CEST
(In reply to comment #25)
> "The file you are trying to attach is 3307 kilobytes (KB) in size.
> Attachments cannot be more than 2000 KB."
> 
> The other 2 attachements has both more than 3 MB. Where should I post it?

In the worst case, send them to us by email please. Thank you for your patience and your precious help.
Comment 27 Sven Gothel 2013-04-08 23:41:29 CEST
(In reply to comment #25)
> "The file you are trying to attach is 3307 kilobytes (KB) in size.
> Attachments cannot be more than 2000 KB."
> 
> The other 2 attachements has both more than 3 MB. Where should I post it?

pls compress them w/ 7z .. should be very efficient on text files.
Comment 28 Sven Gothel 2013-04-08 23:43:07 CEST
(In reply to comment #23)
> I finally run it. Test and test_dbg.
> (Test 2, Test_dbg 2)
Hurray!

> 
> No, I have HP ProBook 4720s (XX838EA).
> http://www.hardwarest.com/product/HP-ProBook-4720s-(XX838EA)-167738.html

Thank you - I will then try to get my hands on one of those .. hmm,
probably a used one from somewhere, since they are no more sold I guess.
Comment 29 robinhaf 2013-04-09 10:50:24 CEST
Created attachment 454 [details]
Debug log
Comment 30 robinhaf 2013-04-09 10:51:28 CEST
Created attachment 455 [details]
Test_dbg from etc folder
Comment 31 Julien Gouesse 2013-04-16 10:40:54 CEST
This crash is probably reproducible with these chips too:
ATI Radeon HD 4550
ATI Radeon Mobility 3650 HD
ATI Radeon HD 5870 extreme edition
AMD Radeon HD 7800 Series (even with the driver 13.3b)
Comment 32 Sven Gothel 2013-04-24 18:52:19 CEST
*** Bug 520 has been marked as a duplicate of this bug. ***
Comment 33 Sven Gothel 2013-04-24 18:53:21 CEST
confirmed w/ local machine .. working on it today.
Comment 34 Sven Gothel 2013-04-24 22:12:53 CEST
Win7 ATI mobile driver simply crashes when trying to create a GL 4 ctx, 
while GL 4 works w/ ARB creation functions.

+++

Versions we receive:

GL_VENDOR: ATI Technologies Inc.
GL_RENDERER: ATI Mobility Radeon HD 6370
GL_VERSION: 3.3.9910 Compatibility Profile Context

+++

ATI Driver versions (sorry .. German Win7 installation):

Treiber-Paketversion	8.743.2.5-110315a-115833C-HP	
Anbieter	ATI Technologies Inc.	
2D-Treiberversion	8.01.01.1038	
2D-Treiberpfad	/REGISTRY/MACHINE/SYSTEM/ControlSet001/Control/CLASS/{4D36E968-E325-11CE-BFC1-08002BE10318}/0000	
Direct3D-Version	8.14.10.0761	
OpenGL-Version	6.14.10.9910	
Catalyst Control Center-Version	2011.0316.116.298

+++

We need a Quirk for these drivers !

Questions: 
 - What are the driver versions, such Quirk is needed,
   i.e. GL profile support range per driver ?

 - How to properly detected the driver versions ?
Comment 35 Sven Gothel 2013-04-24 22:20:10 CEST
(In reply to comment #34)
> +++
> 
> Versions we receive:
> 
> GL_VENDOR: ATI Technologies Inc.
> GL_RENDERER: ATI Mobility Radeon HD 6370
> GL_VERSION: 3.3.9910 Compatibility Profile Context
> 
> +++
> 
> ATI Driver versions (sorry .. German Win7 installation):
> 
> Treiber-Paketversion	8.743.2.5-110315a-115833C-HP	
> Anbieter	ATI Technologies Inc.	
> 2D-Treiberversion	8.01.01.1038	
> 2D-Treiberpfad
> /REGISTRY/MACHINE/SYSTEM/ControlSet001/Control/CLASS/{4D36E968-E325-11CE-
> BFC1-08002BE10318}/0000	
> Direct3D-Version	8.14.10.0761	
> OpenGL-Version	6.14.10.9910	
> Catalyst Control Center-Version	2011.0316.116.298
> 
> +++

Robinhaf's driver is 

ATI Driver Version: 8.743.2.5000

GL_VENDOR      ATI Technologies Inc.
GL_RENDERER    ATI Mobility Radeon HD 6370
GL_VERSION     4.0.9910 Compatibility Profile Context
Comment 36 Julien Gouesse 2013-04-25 01:39:54 CEST
I assume it is not trivial but this quirk is not needed in some cases, it depends both on the driver version and on the chip. For example, in the case of the ATI Mobility Radeon HD 6370, the latest driver really fixes this bug but on some other chips I listed in a previous comment, it doesn't work.
Comment 37 Sven Gothel 2013-04-25 05:30:41 CEST
I figured the quirk is to make a pre-existing context (shared or an old-GL-ctx) current,
then call wglCreateContextAttribsARB(..). At least for my test machine.

Well, the costs for this quirk is limited, 
i.e. only one extra: makeCurrent/release at GL context creation.

Most simple fix would be to enable it for all ATI drivers ..
Comment 38 Sven Gothel 2013-04-25 05:39:18 CEST
(In reply to comment #31)
> This crash is probably reproducible with these chips too:

- ATI Radeon HD 4550 [2]
- ATI Radeon Mobility 3650 HD [2]
- ATI mobility Radeon HD 5470 [1] (Bug 520)
- ATI Radeon HD 5870 extreme edition [2]
- ATI Mobility Radeon HD 6370 [1] (Bug 706)
- AMD Radeon HD 7800 Series (even with the driver 13.3b) [2]

[1] confirmed
[2] ???

+++

I don't know how you gathered the other devices.

I agree, that the bugs are probably gone w/ new drivers.
However, you also mention this might be not true.

However, if we would validate device driver version, we would need to 
go through hoops, i.e. Device Driver SDK [3] - plus we would need to
have a rough equation what works and what not.

Considering my comment 37 regarding the costs, 
I will push a quick and dirty fix enabling the workaround for all Windows ATI driver.
We can refine this .. when we have more knowledge, but it might be better 
to test the workaround itself on said machines - also your list.

[3] http://msdn.microsoft.com/en-us/library/ff553567.aspx#ddk_setupdi_device_information_functions_dg]
Comment 39 Sven Gothel 2013-04-25 06:10:10 CEST
fixed: 75b3d37a3b15c071b49609921244bcb62d329fa9

Please re-open if bug persists.

We may open a new bug depending on this one
to refine the behavior as discussed.
Comment 40 Sven Gothel 2013-04-25 06:32:52 CEST
ran unit test on machine as described in comment 34:

Results of ../build-win64//test/results

number of junit classes
    193    1544   56480

number of passed junit classes - failures
    191    1528   55913

number of passed junit classes - errors
    191    1528   55920

number of failed junit classes - failures
      2      16     567

number of failed junit classes - errors
      2      16     560

failed junit classes - failures
../build-win64//test/results/TEST-com.jogamp.opengl.test.junit.newt.event.TestNewtKeyCodeModifiersAWT.xml:<testsuite errors="0" failures="2" hostname="jogamp06" name="com.jogamp.opengl.test.junit.newt.event.TestNewtKeyCodeModifiersAWT" tests="3" time="5.645" timestamp="2013-04-25T04:26:08">
../build-win64//test/results/TEST-com.jogamp.opengl.test.junit.newt.event.TestNewtKeyCodesAWT.xml:<testsuite errors="0" failures="2" hostname="jogamp06" name="com.jogamp.opengl.test.junit.newt.event.TestNewtKeyCodesAWT" tests="3" time="21.9" timestamp="2013-04-25T04:26:15">

(German Keyboard - Expected!)

failed junit classes - errors
../build-win64//test/results/TEST-com.jogamp.opengl.test.junit.jogl.caps.TestMultisampleES1NEWT.xml:<testsuite errors="1" failures="0" hostname="jogamp06" name="com.jogamp.opengl.test.junit.jogl.caps.TestMultisampleES1NEWT" tests="8" time="4.274" timestamp="2013-04-25T04:10:56">
../build-win64//test/results/TEST-com.jogamp.opengl.test.junit.jogl.caps.TestMultisampleES2NEWT.xml:<testsuite errors="1" failures="0" hostname="jogamp06" name="com.jogamp.opengl.test.junit.jogl.caps.TestMultisampleES2NEWT" tests="6" time="3.713" timestamp="2013-04-25T04:11:01">

(Not expected .. but well ..)
Comment 43 Sven Gothel 2013-04-25 10:41:39 CEST
Updated drivers: 
  'amd_catalyst_13.5_mobility_beta2.exe' from
  <http://support.amd.com/us/gpudownload/windows/Pages/radeonmob_win7-64.aspx>

worked on my machine (comment 34):
  HP ProBook 4720s (XX838EA)
  ATI Mobility Radeon HD 6370

Old driver (buggy):
   GL Version     4.0 (hardware) - 4.0.9910 Compatibility Profile Context [GL 4.0.0, vendor null]  
   GL_VENDOR      ATI Technologies Inc.
   GL_RENDERER    ATI Mobility Radeon HD 6370
   GL_VERSION     4.0.9910 Compatibility Profile Context

New driver (fixed, i,.e old JOGL rc11 is working):
  GL Version     4.2 (hardware) - 4.2.12198 Compatibility Profile Context 12.102.3.0 [GL 4.2.0, vendor 12.102.3 (12.102.3.0)]
  GL_VENDOR      ATI Technologies Inc.
  GL_RENDERER    AMD Radeon HD 6300M Series
  GL_VERSION     4.2.12198 Compatibility Profile Context 12.102.3.0

Also notable our old AMD desktop driver form our desktop win64-amd jenkins node:

  GL Version     4.2 (hardware) - 4.2.11931 Compatibility Profile Context [GL 4.2.0, vendor null]
  GL_VENDOR      ATI Technologies Inc.
  GL_RENDERER    ATI Radeon HD 5500 Series
  GL_VERSION     4.2.11931 Compatibility Profile Context

+++

Dropping the quirk could be done using some heuristics:

 - New driver provides the vendor driver version at the end of GL_VERSION.
   Note: NV/Mesa does the same. 
   Hence if( null != ctx.getGLVendorVersionNumber() ) -> no-quirk required. 
   See ' vendor 12.102.3 (12.102.3.0)' above.
   Guess this would be a safe bet.

-  No 'Mobility' [and/or 'ATI'] in GL_RENDERER,
   but don't know whether all other old GPUs and drivers use 'Mobility'
Comment 44 robinhaf 2013-04-25 11:05:01 CEST
(In reply to comment #41)
> Pls test:
>  
> http://jogamp.org/deployment/archive/master/gluegen_666-joal_432-jogl_981-
> jocl_788/

This works

(In reply to comment #42)
> http://jogamp.org/deployment/archive/master/gluegen_666-joal_432-jogl_981-
> jocl_788-signed/jogl-test-applets.html
> 
> signed .. you may test w/ applets

This works too
Comment 45 Julien Gouesse 2013-04-25 12:50:42 CEST
(In reply to comment #44)
> (In reply to comment #41)
> > Pls test:
> >  
> > http://jogamp.org/deployment/archive/master/gluegen_666-joal_432-jogl_981-
> > jocl_788/
> 
> This works
> 
> (In reply to comment #42)
> > http://jogamp.org/deployment/archive/master/gluegen_666-joal_432-jogl_981-
> > jocl_788-signed/jogl-test-applets.html
> > 
> > signed .. you may test w/ applets
> 
> This works too

This is an excellent piece of news, I'm very glad for you and Sven did a great job. I'm almost sure this case is not properly handled in other similar APIs.
Comment 46 Sven Gothel 2013-04-26 06:40:03 CEST
http://jogamp.org/git/?p=jogl.git;a=commit;h=36dc4f5ab6957a4078842c488afb51df2fdc0630

Move WindowsWGLDrawableFactory Quirks to GLRendererQuirks 
(NeedCurrCtx4ARBPixFmtQueries, NeedCurrCtx4ARBCreateContext)
  - For better maintenance, move the mentioned quirks from the windows factory to our common place,
    being detected within GLContextImpl after each context creation.

Confine NeedCurrCtx4ARBCreateContext to [Windows, ATI, driver < 12.102.3)
  - Before we added this quirk if [Windows, ATI],
    however, we have hopes that the new drivers will suffice for all
    as tested successful on my test machine 
   (AMD Radeon HD 6300M Series,  amd_catalyst_13.5_mobility_beta2).

+++

Old Driver:
  GL Version     4.0 (hardware) - 4.0.9910 Compatibility Profile Context [GL 4.0.0, vendor 0.0.0 (n/a)]
  Quirks         [NoDoubleBufferedBitmap, NeedCurrCtx4ARBCreateContext]
  GL_VENDOR      ATI Technologies Inc.
  GL_RENDERER    ATI Mobility Radeon HD 6370
  GL_VERSION     4.0.9910 Compatibility Profile Context

New Driver:
  GL Version     4.2 (hardware) - 4.2.12198 Compatibility Profile Context 12.102.3.0 
                        [GL 4.2.0, vendor 12.102.3 (12.102.3.0)]
  Quirks         [NoDoubleBufferedBitmap]
  GL_VENDOR      ATI Technologies Inc.
  GL_RENDERER    AMD Radeon HD 6300M Series
  GL_VERSION     4.2.12198 Compatibility Profile Context 12.102.3.0

+++

Hope that works for all this class of GPUs ..

If not, please re-open and report versions to refine Quirk definition.

Pls check w/ next aggregated builds at (in a few hours):
  <http://jogamp.org/deployment/archive/master/?C=M;O=D>

Thank you.