| Summary: | 3.1 context on Mesa 9.0.1 is buggy | ||
|---|---|---|---|
| Product: | [JogAmp] Jogl | Reporter: | Mark Raynsford <org.jogamp> |
| Component: | core | Assignee: | Sven Gothel <sgothel> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | sgothel |
| Priority: | --- | ||
| Version: | 2 | ||
| Hardware: | pc_all | ||
| OS: | linux | ||
| Type: | --- | SCM Refs: |
jogl 3ce0aa6e36d9474ac3105cab491a60327860757d
|
| Workaround: | TRUE | ||
| Attachments: |
Two unit tests
The second test as native code with GLFW The first test as native code with GLFW |
||
Created attachment 403 [details]
The second test as native code with GLFW
Created attachment 404 [details]
The first test as native code with GLFW
The native tests give the following output on buggy contexts: renderer : Mesa DRI Intel(R) Sandybridge Mobile version : 3.1 Mesa 9.0.1 vendor : Intel Open Source Technology Center glVertexAttribPointer: 502 renderer : Mesa DRI Intel(R) Sandybridge Mobile version : 3.1 Mesa 9.0.1 vendor : Intel Open Source Technology Center test-poly: glPolygonMode: 1280 I've just discovered that glPolygonMode(GL_FRONT, ...); and glPolygonMode(GL_BACK, ...); are invalid under 3.1 - only GL_FRONT_AND_BACK is allowed. I'm still investigating the second case, in case it's not a bug either. Fix Bug 658 (Mesa 9.0 3.1 Intel compat quirk, 3.1 core only) ; No PROFILE_ALIASING compat -> core ; Fix setGLFunctionAvailability(..) failure path @ profile query - Add GLRendererQuirks.GLNonCompliant, marking a GL context/profile non compliant. Currently: 'Mesa DRI Intel(R) Sandybridge Desktop' && 3.1 compat profile - Fix Bug 658 (Mesa 9.0 3.1 Intel compat quirk, 3.1 core only) Detect case using new GLRendererQuirks.GLNonCompliant in setGLFunctionAvailability() and return 'false'. ... http://jogamp.org/git/?p=jogl.git;a=commit;h=3ce0aa6e36d9474ac3105cab491a60327860757d |
Created attachment 402 [details] Two unit tests The 3.1 context exposed by the recent 9.0.1 update is defective. At least two function calls give nonsensical errors (the second of which actually prevents the use of VBOs at all). I've only confirmed this on Intel. See the attached test cases (both of which should *pass* on the buggy setup). Note that forcing a 3.0 context with -Djogl.debug.GLContext.NoProfileAliasing results in a working 3.0 context that won't give the errors detected in the unit tests.