[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 Mark Janes changed: What|Removed |Added CC||lem...@gmail.com -- You are receiving this mail because: You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #19 from Tapani Pälli --- (In reply to Mark Janes from comment #18) > The fix for this bug triggered a piglit failure: > > spec.!opengl 3_2.get-active-attrib-returns-all-inputs > > /tmp/build_root/m32/lib/piglit/bin/gl-get-active-attrib-returns-all-inputs > -auto -fbo > piglit_vertex was not counted as active. > > Should I write up a separate bug for this? Yes please, assign to me. I think I know what's up here, it's probably the builtin attribs that are not referenced by the vertex stage. -- You are receiving this mail because: You are the QA Contact for the bug. You are on the CC list for the bug. You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #18 from Mark Janes --- The fix for this bug triggered a piglit failure: spec.!opengl 3_2.get-active-attrib-returns-all-inputs /tmp/build_root/m32/lib/piglit/bin/gl-get-active-attrib-returns-all-inputs -auto -fbo piglit_vertex was not counted as active. Should I write up a separate bug for this? -- You are receiving this mail because: You are the QA Contact for the bug. You are on the CC list for the bug. You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 José Fonseca changed: What|Removed |Added Status|NEW |RESOLVED Resolution|--- |FIXED --- Comment #17 from José Fonseca --- (In reply to Tapani Pälli from comment #16) > Yep it is correct, just like with _mesa_longest_attribute_name_length. Thanks. I pushed the fix now. > Not sure why I did not manage to hit this. It's random -- I think that is_active_attrib ends up interpreting random pointer addresses as enums and decide on them. Maybe the randomness is higher with gallium drivers. It was only when I try to print var->name that things started to crash consistently for me. > > I think we should invest the time to make RESOURCE_VAR and friends more > > robust, by adding assertions that the types are indeed correct. (Replace > > DECL_RESOURCE_FUNC helper macro by manually written code if necessary.) > > > > This sort of bugs is hard to find otherwise. > > I agree, current way is too fragile. IMO a type template would be best > solution. Maybe first just adding the assert for the correct type is good > start. Yep. -- You are receiving this mail because: You are the QA Contact for the bug. You are on the CC list for the bug. You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207
--- Comment #16 from Tapani Pälli ---
(In reply to José Fonseca from comment #13)
> I think I figured out the problem.
>
> As commented on DECL_RESOURCE_FUNC macro, the RESOURCE_VAR inline function
> is not type safe, and stuff that's not a ir_variable is wrongly being casted
> into it.
>
> This patch seems to do the trick, but I'm not 100% sure.
Yep it is correct, just like with _mesa_longest_attribute_name_length. Not sure
why I did not manage to hit this.
> diff --git a/src/mesa/main/shader_query.cpp b/src/mesa/main/shader_query.cpp
> index a84ec84..d2ca49b 100644
> --- a/src/mesa/main/shader_query.cpp
> +++ b/src/mesa/main/shader_query.cpp
> @@ -302,8 +302,10 @@ _mesa_count_active_attribs(struct gl_shader_program
> *shProg)
> struct gl_program_resource *res = shProg->ProgramResourceList;
> unsigned count = 0;
> for (unsigned j = 0; j < shProg->NumProgramResourceList; j++, res++) {
> - if (is_active_attrib(RESOURCE_VAR(res)))
> -count++;
> + if (res->Type == GL_PROGRAM_INPUT &&
> + res->StageReferences & (1 << MESA_SHADER_VERTEX) &&
> + is_active_attrib(RESOURCE_VAR(res)))
> + count++;
> }
> return count;
> }
>
>
>
> I think we should invest the time to make RESOURCE_VAR and friends more
> robust, by adding assertions that the types are indeed correct. (Replace
> DECL_RESOURCE_FUNC helper macro by manually written code if necessary.)
>
> This sort of bugs is hard to find otherwise.
I agree, current way is too fragile. IMO a type template would be best
solution. Maybe first just adding the assert for the correct type is good
start.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
You are the assignee for the bug.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #15 from Dieter Nützel --- (In reply to Aaron Watry from comment #14) > (In reply to José Fonseca from comment #13) > > I think I figured out the problem. > > > > As commented on DECL_RESOURCE_FUNC macro, the RESOURCE_VAR inline function > > is not type safe, and stuff that's not a ir_variable is wrongly being casted > > into it. > > > > This patch seems to do the trick, but I'm not 100% sure. > > > > That patch fixes the crash in Konqueror for me. > > Whether or not it's correct, that's up to people who know the code better > than myself. For me, too. git-6fe0d4f + José's patch Thanks José! Your WebGL implementation: Renderer: WebKit WebGL Vendor: WebKit WebGL version: WebGL 1.0 (3.0 Mesa 10.6.0-devel (git-6fe0d4f)) GLSL version: WebGL GLSL ES 1.0 (1.30) Supported WebGL extensions: OES_texture_float <-- Were is this defined? ;-) OES_standard_derivatives WEBKIT_EXT_texture_filter_anisotropic OES_vertex_array_object OES_element_index_uint WEBKIT_WEBGL_lose_context WEBKIT_WEBGL_compressed_texture_s3tc WEBKIT_WEBGL_depth_texture -- You are receiving this mail because: You are the QA Contact for the bug. You are on the CC list for the bug. You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #14 from Aaron Watry --- (In reply to José Fonseca from comment #13) > I think I figured out the problem. > > As commented on DECL_RESOURCE_FUNC macro, the RESOURCE_VAR inline function > is not type safe, and stuff that's not a ir_variable is wrongly being casted > into it. > > This patch seems to do the trick, but I'm not 100% sure. > That patch fixes the crash in Konqueror for me. Whether or not it's correct, that's up to people who know the code better than myself. -- You are receiving this mail because: You are the QA Contact for the bug. You are on the CC list for the bug. You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207
--- Comment #13 from José Fonseca ---
I think I figured out the problem.
As commented on DECL_RESOURCE_FUNC macro, the RESOURCE_VAR inline function is
not type safe, and stuff that's not a ir_variable is wrongly being casted into
it.
This patch seems to do the trick, but I'm not 100% sure.
diff --git a/src/mesa/main/shader_query.cpp b/src/mesa/main/shader_query.cpp
index a84ec84..d2ca49b 100644
--- a/src/mesa/main/shader_query.cpp
+++ b/src/mesa/main/shader_query.cpp
@@ -302,8 +302,10 @@ _mesa_count_active_attribs(struct gl_shader_program
*shProg)
struct gl_program_resource *res = shProg->ProgramResourceList;
unsigned count = 0;
for (unsigned j = 0; j < shProg->NumProgramResourceList; j++, res++) {
- if (is_active_attrib(RESOURCE_VAR(res)))
-count++;
+ if (res->Type == GL_PROGRAM_INPUT &&
+ res->StageReferences & (1 << MESA_SHADER_VERTEX) &&
+ is_active_attrib(RESOURCE_VAR(res)))
+ count++;
}
return count;
}
I think we should invest the time to make RESOURCE_VAR and friends more robust,
by adding assertions that the types are indeed correct. (Replace
DECL_RESOURCE_FUNC helper macro by manually written code if necessary.)
This sort of bugs is hard to find otherwise.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
You are the assignee for the bug.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #12 from José Fonseca --- I ran the test inside apitrace multiple times. When it passes we have glGetProgramiv(program = 36, pname = GL_ACTIVE_ATTRIBUTES, params = &2) when it fails we have glGetProgramiv(program = 36, pname = GL_ACTIVE_ATTRIBUTES, params = &3) I.e, the bug is somewhere inside _mesa_count_active_attribs. I also tried to run the test inside valgrind sucessively, but then it stops failing -- You are receiving this mail because: You are the QA Contact for the bug. You are on the CC list for the bug. You are the assignee for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207
José Fonseca changed:
What|Removed |Added
CC||jfons...@vmware.com
Component|Drivers/Gallium/r600|Mesa core
Assignee|dri-devel@lists.freedesktop |mesa-dev@lists.freedesktop.
|.org|org
QA Contact|dri-devel@lists.freedesktop |mesa-dev@lists.freedesktop.
|.org|org
--- Comment #11 from José Fonseca ---
(In reply to Dieter Nützel from comment #0)
[...]
> a563689a408b7a28c710fb0e382272a0d823f38a is the first bad commit
> commit a563689a408b7a28c710fb0e382272a0d823f38a
> Author: Tapani Pälli
> Date: Thu Apr 23 11:13:17 2015 +0300
>
> mesa: refactor active attrib queries for glGetProgramiv
>
> Main motivation here is to get rid of iterating IR and
> encapsulate queries within program resources.
> No functional changes.
>
> Piglit tests calling the modified functionality:
>
>- gl-get-active-attrib-returns-all-inputs
>- glsl-1.50-get-active-attrib-array
>- getactiveattrib
>
> [...]
Piglit `getactiveattrib 110` test started to randomly fail on llvmpipe:
$ for i in `seq 1 100`; do .../piglit/bin/getactiveattrib 110 -auto -fbo; done
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
Mesa: User error: GL_INVALID_VALUE in glGetActiveAttrib(index)
Failing shader:
attribute vec4 vertex;
attribute vec2 alternate;
uniform bool use_alternate;
void main() {
gl_Position = vertex;
if (use_alternate) gl_Position = alternate.xyxy;
}
Attribute `alternate' listed multiple times in active list.
PIGLIT: {"result": "fail" }
PIGLIT: {"result": "pass" }
PIGLIT: {"result": "pass" }
> When I revert 'a563689' all is smooth, again.
And likewise, if I revert a563689 it passes consistently.
--
You are receiving this mail because:
You are the QA Contact for the bug.
You are on the CC list for the bug.
You are the assignee for the bug.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #10 from Aaron Watry --- Created attachment 115405 --> https://bugs.freedesktop.org/attachment.cgi?id=115405&action=edit API Trace of konqueror crash -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #9 from Aaron Watry --- Note that firefox 37 works fine for me on my CEDAR (Radeon 5400) on r600g on current mesa master. Konqueror 4.14.6 crashes while loading the OpenGL demo. When running konqueror with LIBGL_DEBUG=verbose, I get the following: awatry@ws-awatry:~$ LIBGL_DEBUG=verbose konqueror libGL: OpenDriver: trying /usr/local/lib/dri/tls/r600_dri.so libGL: OpenDriver: trying /usr/local/lib/dri/r600_dri.so libGL: OpenDriver: trying /usr/local/lib/dri/tls/r600_dri.so libGL: OpenDriver: trying /usr/local/lib/dri/r600_dri.so NOT SANDBOXED Mesa: User error: GL_INVALID_VALUE in glGetActiveAttrib(index) KCrash: Application 'konqueror' crashing... -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #8 from Tapani Pälli --- I cannot reproduce this bug. I used following page to verify that Konqueror runs against my Mesa: http://renderingpipeline.com/webgl-extension-viewer/ "WebGL version: WebGL 1.0 (3.0 Mesa 10.6.0-devel (git-9143940))" webgl-water and other webgl tests run fine, Konqueror version information: konqueror-14.12.3-1.fc21.x86_64 kwebkitpart-1.3.4-5.fc21.x86_64 this is on Ivybridge laptop, i965 driver. -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #7 from Tapani Pälli --- (In reply to Dieter Nützel from comment #6) > (In reply to Dieter Nützel from comment #5) > > (In reply to Dieter Nützel from comment #4) > > > (In reply to Tapani Pälli from comment #3) > > > > My konqueror does not seems to support webgl (?) is there something > > > > special > > > > I need to do to enable it? > > > > > > Hello Tapani! > > > > [snip] > > > > > Maybe you have to use WebKit not KHTML. > > > http://html5test.com/index.html > > > > It is in since > > > > 11./13.10.2011 > > > > through WebKit. > > kwebkitpart > kwebkitpart-1.3.3-3.1.5.x86_64 Yes, installing this package made it work, demos seem to work for me though .. I will try some more and also investigate possible side-effects caused by my patch. -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #6 from Dieter Nützel --- (In reply to Dieter Nützel from comment #5) > (In reply to Dieter Nützel from comment #4) > > (In reply to Tapani Pälli from comment #3) > > > My konqueror does not seems to support webgl (?) is there something > > > special > > > I need to do to enable it? > > > > Hello Tapani! > > [snip] > > > Maybe you have to use WebKit not KHTML. > > http://html5test.com/index.html > > It is in since > > 11./13.10.2011 > > through WebKit. kwebkitpart kwebkitpart-1.3.3-3.1.5.x86_64 -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #5 from Dieter Nützel --- (In reply to Dieter Nützel from comment #4) > (In reply to Tapani Pälli from comment #3) > > My konqueror does not seems to support webgl (?) is there something special > > I need to do to enable it? > > Hello Tapani! [snip] > Maybe you have to use WebKit not KHTML. > http://html5test.com/index.html It is in since 11./13.10.2011 through WebKit. -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #4 from Dieter Nützel --- (In reply to Tapani Pälli from comment #3) > My konqueror does not seems to support webgl (?) is there something special > I need to do to enable it? Hello Tapani! My Konqueror 4.14.6 and former versions (on openSUSE x86 and x86-64) support it for some months (years?) out of the box. What's weird that it do NOT run webgl if I go back to repro package Mesa-10.5.4 (from 27 Apr 2015). I can swear it worked with former versions... Have to retest on my poor RV730/x86 system tomorrow. Maybe you have to use WebKit not KHTML. http://html5test.com/index.html -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #3 from Tapani Pälli --- My konqueror does not seems to support webgl (?) is there something special I need to do to enable it? -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #2 from Tapani Pälli --- seems to work with chrome and firefox, will attempt to reproduce with konqueror -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 Dieter Nützel changed: What|Removed |Added Hardware|Other |x86-64 (AMD64) OS|All |Linux (All) Severity|normal |critical -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 --- Comment #1 from Dieter Nützel --- Created attachment 115389 --> https://bugs.freedesktop.org/attachment.cgi?id=115389&action=edit konqueror-20150428-051127.kcrash.txt -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
[Mesa-dev] [Bug 90207] [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff)
https://bugs.freedesktop.org/show_bug.cgi?id=90207 Bug ID: 90207 Summary: [r600g, bisected] regression: NI/Turks crash on WebGL Water (most WebGL stuff) Product: Mesa Version: git Hardware: Other OS: All Status: NEW Severity: normal Priority: medium Component: Drivers/Gallium/r600 Assignee: dri-de...@lists.freedesktop.org Reporter: die...@nuetzel-hh.de QA Contact: dri-de...@lists.freedesktop.org CC: mesa-dev@lists.freedesktop.org Current Mesa git (9143940) crash on WebGL Water demo and most WebGL stuff. http://madebyevan.com/webgl-water/ http://www.ibiblio.org/e-notes/webgl/webgl.htm I've bisected it to: /opt/mesa> git bisect good a563689a408b7a28c710fb0e382272a0d823f38a is the first bad commit commit a563689a408b7a28c710fb0e382272a0d823f38a Author: Tapani Pälli Date: Thu Apr 23 11:13:17 2015 +0300 mesa: refactor active attrib queries for glGetProgramiv Main motivation here is to get rid of iterating IR and encapsulate queries within program resources. No functional changes. Piglit tests calling the modified functionality: - gl-get-active-attrib-returns-all-inputs - glsl-1.50-get-active-attrib-array - getactiveattrib Signed-off-by: Tapani Pälli Reviewed-by: Martin Peres :04 04 8983df05bac6cc3a455b2e5100a2038939e24eaa cd5fa256be39b9b02e9201a9ce6d313c670f2942 M src When I revert 'a563689' all is smooth, again. -- You are receiving this mail because: You are on the CC list for the bug. ___ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
