http://bugs.winehq.org/show_bug.cgi?id=27746
Summary: fbo vs backbuffer regression in wine 1.3.24 Product: Wine Version: 1.3.24 Platform: x86 OS/Version: Linux Status: UNCONFIRMED Severity: normal Priority: P2 Component: directx-d3d AssignedTo: wine-bugs@winehq.org ReportedBy: galtgendo@o2.pl CC: stefan@codeweavers.com
This is a follow up of bug 25717.
As noted there, commit cd96c59d9168fbda25536d328edb24c0bc0e5139 (wined3d: Track framebuffer changes.) caused a regression, in that certain elements aren't drawn if fbo is used, while they're still displayed correctly with backbuffer.
The game in question spits following warnings repeatedly, regardless of commit, that caused the regression: err:d3d:loadVertexData >>>>>>>>>>>>>>>>> GL_INVALID_ENUM (0x500) from glBindBufferARB @ state.c / 4319 err:d3d:loadVertexData >>>>>>>>>>>>>>>>> GL_INVALID_ENUM (0x500) from glBindBufferARB @ state.c / 4319 err:d3d_draw:drawStridedSlow >>>>>>>>>>>>>>>>> GL_INVALID_ENUM (0x500) from glEnd and previous calls @ drawprim.c / 309
If it matters: linux-2.6.39, r200, libdrm-2.4.25, mesa-7.10.3, xf86-video-ati-6.14.2
http://bugs.winehq.org/show_bug.cgi?id=27746
Dmitry Timoshkov dmitry@baikal.ru changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |regression Summary|fbo vs backbuffer |Сertain elements aren't |regression in wine 1.3.24 |drawn if fbo is used
--- Comment #1 from Dmitry Timoshkov dmitry@baikal.ru 2011-07-09 14:03:38 CDT --- ... in which application?
http://bugs.winehq.org/show_bug.cgi?id=27746
Stefan Dösinger stefandoesinger@gmx.at changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |stefandoesinger@gmx.at
--- Comment #2 from Stefan Dösinger stefandoesinger@gmx.at 2011-07-10 15:12:58 CDT --- The app was some japanese app whose name I can neither spell nor pronounce, you can find it at http://holyseal.net/cgi-bin/mlistview.cgi?prdcode=8222 . The app in question has "ver.0.03" next to it.
I could not reproduce this bug on OSX with a Geforce 9600, so I suspect Mesa has a bug that causes a order dependence between some states. I also could not reproduce the FBO-unrelated GL errors reported in the first post.
Personally I'll wait and see if any other regressions are reported from the FBO patches and reinvestigate this closer if bugs occur on other drivers. r200 chips are rather old and nobody really cares about them any more.
http://bugs.winehq.org/show_bug.cgi?id=27746
--- Comment #3 from Rafał Mużyło galtgendo@o2.pl 2011-08-02 08:45:58 CDT --- Well, a quite amusing change took place. Mesa 7.11 was released and the glyphs appeared, but so did "Mesa 7.11 implementation error: Unexpected texture format in radeon_update_wrapper()" warnings (from src/mesa/drivers/dri/radeon/radeon_fbo.c). Now, the funny part is that no significant changes seem to have been made in r200 private code - it looks like this was an effect in the shared radeon code.
Not that I did a bisect (cause I didn't), but given the warnings, after looking at the git history, I'd suspect "mesa: Switch generate_mipmaps_compressed() to using TexImage2D to upload." commit.
Now, the question is "is mesa doing something odd or is wine".
GL_INVALID_ENUM warnings are still there, but it seems we've already established, that those are about something regarding GL_ATI_fragment_shader extension, that (according to comments in wine sources) seems unused outside r200.
http://bugs.winehq.org/show_bug.cgi?id=27746
--- Comment #4 from Stefan Dösinger stefandoesinger@gmx.at 2011-08-02 08:59:50 CDT --- I'm not sure about the mesa specific parts - I once fixed a minor issue in the r200 code(ATI_fragment_shader wasn't enabled with kernel mode setting), but beyond that I don't know the code.
And yes, ATI_fragment_shader only makes sense for radeon 8500 to radeon 9200 cards(r200, r250, r280). Older cards don't support it, and newer cards support ARB_fragment_program, which we prefer over ATI_fragment_shader. Fglrx used to support ATI_fragment_shader on newer cards while it still supported those cards, but the implementation was broken.
http://bugs.winehq.org/show_bug.cgi?id=27746
Henri Verbeet hverbeet@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Regression SHA1| |cd96c59d9168fbda25536d328ed | |b24c0bc0e5139
http://bugs.winehq.org/show_bug.cgi?id=27746
Stefan Dösinger stefandoesinger@gmx.at changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|UNCONFIRMED |RESOLVED Resolution| |INVALID
--- Comment #5 from Stefan Dösinger stefandoesinger@gmx.at 2011-08-13 16:37:26 CDT --- I'll resolve this bug as invalid since it looks like a bug in the driver of an ancient GPU. Feel free to open a separate bug for the GL errors.
http://bugs.winehq.org/show_bug.cgi?id=27746
Austin English austinenglish@gmail.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|RESOLVED |CLOSED
--- Comment #6 from Austin English austinenglish@gmail.com 2011-08-15 16:35:07 CDT --- Closing.
-
wine-bugs＠winehq.org