https://bugs.winehq.org/show_bug.cgi?id=38326
Bug ID: 38326 Summary: 3DMark03: Could not find vertex shader attribute 'vs_in1' Product: Wine Version: 1.7.39 Hardware: x86 OS: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: directx-d3d Assignee: wine-bugs@winehq.org Reporter: isakov-sl@bk.ru Distribution: ---
Running 3DMark03 I got several messages like --- fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #92: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request.
--- This probably appears because of using vs_link instead of vs_in. Crossover codes differs by this usage and shown 4 times better result in Rag Troll test.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #1 from Sergey Isakov isakov-sl@bk.ru --- Tested unmodified wine-1.7.42-137-g6d323d8 The bug is still present ---- fixme:d3d:resource_check_usage Unhandled usage flags 0x40. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #78: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #2 from Sergey Isakov isakov-sl@bk.ru --- The bug is reproduced by tests wine c:/tests/d3d9_test.exe.so visual ---- fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #3: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. visual.c:14042: Test failed: Expected color 0x000000ff, 0x00007f00, 0x0000ff00 or 0x00007f00 for instruction "def1", got 0x00000000. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #3: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #3: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #3 from Sergey Isakov isakov-sl@bk.ru --- Wine-1.7.45 3DMark05 Hundred of such messages --- fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #7: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #9: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #7: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #9: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #7: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #9: fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in1' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log WARNING: Could not find vertex shader attribute 'vs_in2' to match BindAttributeLocation request. fixme:d3d_shader:print_glsl_info_log Info log received from GLSL shader #7:
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #4 from Sergey Isakov isakov-sl@bk.ru --- Wow, victory!
The bug is resolved by the patch from bug #39057
Can be closed duplicated.
https://bugs.winehq.org/show_bug.cgi?id=38326
super_man@post.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |super_man@post.com
--- Comment #5 from super_man@post.com --- (In reply to Sergey Isakov from comment #4)
Wow, victory!
The bug is resolved by the patch from bug #39057
Can be closed duplicated.
You can close as dupes bugs you have created yourself. I don't have permissions to mark as dupe.
Great finding anyway.
https://bugs.winehq.org/show_bug.cgi?id=38326
Sergey Isakov isakov-sl@bk.ru changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution|--- |DUPLICATE
--- Comment #6 from Sergey Isakov isakov-sl@bk.ru --- These messages disappeared after applying patch suggested in Bug 39057
*** This bug has been marked as a duplicate of bug 39057 ***
https://bugs.winehq.org/show_bug.cgi?id=38326
swswine@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |swswine@gmail.com
--- Comment #7 from swswine@gmail.com --- My patch in bug 39057 looks unrelated to this issue, seems like nothing in that patch that could help with shader compile warnings listed here. Perhaps something else has also changed between your 3DMark tests. I've run 3DMark03 with stock wine 1.7.53 without any custom patches (with -nosysteminfo option, without which it does not work due to some COM related issues). It worked for me without any problems or warnings like yours.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #8 from Sergey Isakov isakov-sl@bk.ru --- I also wonder. I have two Wine version, one before applying this patch and second after it. I tested several apps with both engines and confirm that only this patch is a solution. As well affected app is Heroes VI of Might and Magic. May be my bugs appeared only on my hardware/software but the patch is really helps.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #9 from swswine@gmail.com --- At the same time I have 3Dmark03 (and 3DMark06 also) working with stock wine version (tested 1.7.53), without any patches. WineHQ database also suggests the same: https://appdb.winehq.org/objectManager.php?sClass=application&iId=3120
I am pretty sure so far that indexed vertex blending is not used there. I can hardly imagine how the changes made by the patch can trigger the fix for your problem. Most likely there is some other difference. Did you make both builds of wine (with and without the patch) yourselves the same way and installed 3DMark the same way to clean wineprefix? If you are comparing some stock wine build with your build with the patch than the difference might rather come from build environment. Or (the most common reason) can you have different wineprefix setup between the tests?
If you give some details on your test configuration (see below), it may give some chance to guess what can be the reason:
1. wine versions with and without patch (version number and build source: your own, or some prebuilt versions) 2. wineprefix: are you using the same wineprefix for both tests and switch wine versions, or different? Did you start from clean wine prefix in both cases? Do you have any direct3d tuning keys in registry (HKEY_CURRENT_USER/Software/Wine/Direct3d)? 3. Your OS and version 4. Your GPU hardware & driver info (glxinfo | grep OpenGL, or optirun glxinfo | grep OpenGL if using discrete video card with bumblebee)
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #10 from Sergey Isakov isakov-sl@bk.ru --- May be you are right and this patch influenced on me indirect way. Anyway my setup. This is not linux, there is no bumlbee and glxinfo. I am using iMac12,2 with AMD Radeon 6000. OSX 10.9.5 with stock drivers. Wineprefix is hidden from me. I installed Wine into Wineskin and sure it changed every time when I ask to do this.
The bug was reported with wine-1.7.39 and remains here up to 1.7.54. Yes, I can compile unmodified WineHQ from latest git. As well, I can compile unmodified WineStaging from git sources from https://github.com/wine-compholio/wine-patched I also has my local sources with unapproved hacks and patches. For example your patch 39057. Doing this I found the log is clean from the messages of this bug. Checking latest changes I found this patch. Now test: took WineHQ-1.7.54. The bug presents. Apply patch-39057. The bug dissappears.
If you proposed I made a mistake then I will repeat this process again using latest git version.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #11 from Sergey Isakov isakov-sl@bk.ru --- OK, I retested to be sure for myself. ~~~~ iMac:wine-hq sergey$ patch -p1 </Users/sergey/Desktop/wine_tmp/patches/patch-39057.txt patching file dlls/wined3d/glsl_shader.c Hunk #5 succeeded at 8011 (offset -2 lines). Hunk #6 succeeded at 8316 (offset -2 lines). patching file dlls/wined3d/utils.c Hunk #1 succeeded at 4873 (offset -71 lines). patching file dlls/wined3d/vertexdeclaration.c patching file dlls/wined3d/wined3d_private.h patch unexpectedly ends in middle of line iMac:wine-hq sergey $ git describe head wine-1.7.54-332-g370254c iMac:wine-hq sergey $
Game L"C:\Games\Might and Magic Heroes VI - Shades of Darkness\Might & Magic Heroes VI.exe"
Before this patch there are tens messages about "... WARNING: Could not find vertex shader ...."
After the patch these messages disappeared.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #12 from swswine@gmail.com --- There is much of strange things about it. 1. The patch is not completely applied: look at your message: "patching file dlls/wined3d/wined3d_private.h patch unexpectedly ends in middle of line"
Perhaps you've got patch file truncated somehow.
I wonder how this compiles at all as for wined3d_private.h patch adds a field to a structure which is used in the other file.
2. The other files are patched with hunk displacement which means that you are actually patching a bit different version. The patch relates to released version of 1.7.54 (https://dl.winehq.org/wine/source/1.7/, file wine-1.7.54.tar.bz2).
Unfortunately I have completely no experience with Mac/OSX/Wineskin (I was initially confused by your 'Hardware' field telling it is Linux) and thus can't guess any further what may be happening in your environment (e. g. I do not know how build is performed through wineskin), so I suggest we give up for now. Hopefully this strange behavior will be clarified someday by you or someone running these programs on OSX.
https://bugs.winehq.org/show_bug.cgi?id=38326
Matteo Bruni matteo.mystral@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- OS|Linux |Mac OS X
--- Comment #13 from Matteo Bruni matteo.mystral@gmail.com --- (In reply to swswine from comment #9)
I am pretty sure so far that indexed vertex blending is not used there. I can hardly imagine how the changes made by the patch can trigger the fix for your problem.
It's probably the first hunk of your patch, I imagine 3DMark at some point sets a vertex declaration with a BLENDINDEX vertex attribute, not necessarily enabling indexed vertex blending. Apparently only OS X drivers complain about the glBindAttribLocation() call for a nonexistent attribute.
Sergey, as you should know by now, warnings or generally messages don't necessarily imply that there is an issue. If 3DMark draws correctly with plain Wine, this isn't an actual bug.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #14 from Sergey Isakov isakov-sl@bk.ru --- (In reply to swswine from comment #12)
There is much of strange things about it.
- The patch is not completely applied: look at your message:
"patching file dlls/wined3d/wined3d_private.h patch unexpectedly ends in middle of line"
This message appears because there is no end line. If I apply just one empty line the message disappears. Results will be the same so why all is compiled and works.
Perhaps you've got patch file truncated somehow.
I wonder how this compiles at all as for wined3d_private.h patch adds a field to a structure which is used in the other file.
- The other files are patched with hunk displacement which means that you
are actually patching a bit different version. The patch relates to released version of 1.7.54 (https://dl.winehq.org/wine/source/1.7/, file wine-1.7.54.tar.bz2).
No, I take no this version. I got git latest version wine-1.7.54-332-g370254c
Unfortunately I have completely no experience with Mac/OSX/Wineskin (I was initially confused by your 'Hardware' field telling it is Linux) and thus can't guess any further what may be happening in your environment (e. g. I do not know how build is performed through wineskin), so I suggest we give up for now. Hopefully this strange behavior will be clarified someday by you or someone running these programs on OSX.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #15 from Sergey Isakov isakov-sl@bk.ru --- (In reply to Matteo Bruni from comment #13)
(In reply to swswine from comment #9)
I am pretty sure so far that indexed vertex blending is not used there. I can hardly imagine how the changes made by the patch can trigger the fix for your problem.
It's probably the first hunk of your patch, I imagine 3DMark at some point sets a vertex declaration with a BLENDINDEX vertex attribute, not necessarily enabling indexed vertex blending. Apparently only OS X drivers complain about the glBindAttribLocation() call for a nonexistent attribute.
Sergey, as you should know by now, warnings or generally messages don't necessarily imply that there is an issue. If 3DMark draws correctly with plain Wine, this isn't an actual bug.
I just don't know if the image is not full.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #16 from Matteo Bruni matteo.mystral@gmail.com --- (In reply to Sergey Isakov from comment #15)
(In reply to Matteo Bruni from comment #13)
(In reply to swswine from comment #9)
I am pretty sure so far that indexed vertex blending is not used there. I can hardly imagine how the changes made by the patch can trigger the fix for your problem.
It's probably the first hunk of your patch, I imagine 3DMark at some point sets a vertex declaration with a BLENDINDEX vertex attribute, not necessarily enabling indexed vertex blending. Apparently only OS X drivers complain about the glBindAttribLocation() call for a nonexistent attribute.
Sergey, as you should know by now, warnings or generally messages don't necessarily imply that there is an issue. If 3DMark draws correctly with plain Wine, this isn't an actual bug.
I just don't know if the image is not full.
Is there any visible difference in the rendered scenes between unpatched and patched wine?
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #17 from Sergey Isakov isakov-sl@bk.ru --- Hope to find them.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #18 from Sergey Isakov isakov-sl@bk.ru --- (In reply to Matteo Bruni from comment #13)
Sergey, as you should know by now, warnings or generally messages don't necessarily imply that there is an issue. If 3DMark draws correctly with plain Wine, this isn't an actual bug.
All warnings deserve to be eliminated.
https://bugs.winehq.org/show_bug.cgi?id=38326
--- Comment #19 from Matteo Bruni matteo.mystral@gmail.com --- (In reply to Sergey Isakov from comment #18)
(In reply to Matteo Bruni from comment #13)
Sergey, as you should know by now, warnings or generally messages don't necessarily imply that there is an issue. If 3DMark draws correctly with plain Wine, this isn't an actual bug.
All warnings deserve to be eliminated.
No, that's in general unpractical and uninteresting.
https://bugs.winehq.org/show_bug.cgi?id=38326
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #20 from Austin English austinenglish@gmail.com --- Closing.