With AMD Cataylst driver 13.8~beta1 and AMD A8-5600k (AMD Radeon HD 7560D) screen redraws are as somewhat randomised traingles containing part of the area to be redrawn in the area to be redrwan. Reproducible: Always Steps to Reproduce: 1. Open Dolphin 2. Set list view 3. Drag a file with a long name Using git bisect I have tracked the problem to commit 1f0bb6d36075b0b14c927b91d5c3df6720826afd
Created attachment 83976 [details] Screenshot
Thanks for the bisect. Can you also attach the outputs of "qdbus org.kde.kwin /KWin supportInformation" and "glxinfo"? I assume returning false in libkwineffects/kwinglutils.cpp:1964 bool GLVertexBuffer::supportsIndexedQuads() { return false; // just for testing return GLVertexBufferPrivate::supportsIndexedQuads; } would fix it as well then?
Created attachment 84115 [details] Debug Info
Making a simular change fixes it, I removed the code that sets GLVertexBufferPrivate::supportsIndexedQuads to true.
I assume it's a problem in the fglrx beta and bet that GLX_MESA_copy_sub_buffer is a lie. Can you install the stable (former, one that doesn't cause this) fglrx version and provide a counter glxinfo?
Hello! I have same problem with HD8670D (A10-6800K APU). Now using xorg 1.13 (same result with 1.14). I've set up dual monitor and the effect is more intense here then on the screenshot attached. I tried lots of combination of fglrx/xorg and settings. Until now the fglrx 13.4 and xorg 1.13 does not produce the effect (didn't try the 13.9 / 1.13 combo yet), however that still can't use because the whole environment become unstable (painful lag spikes on rendering). I figured some lines in .xsession-errors when the effect happens: QWidget::setMinimumSize: (/KCommonDecorationButton) Negative sizes (-6,-6) are not possible kwin(4688) KWin::x11ErrorHandler: kwin: X Error ( "error: BadPixmap [4], request: X_ChangeGC[56]" ) ... Also this when the lag spike happens with the fglrx 13.4: kwin(4724) KWin::x11ErrorHandler: kwin: X Error ( "error: BadDrawable [9], request: X_GetGeometry[14]" ) kwin(4724) KWin::x11ErrorHandler: kwin: X Error ( "error: BadDrawable [9], request: RenderCreatePicture[RENDER+4]" ) ... This happens all the time with every combination: kwin(4724) KWin::x11ErrorHandler: kwin: X Error ( "error: BadDamage [DAMAGE+0], request: XDamageDestroy[DAMAGE+2], resource: 0x7400075" ) kwin(4724) KWin::x11ErrorHandler: kwin: X Error ( "error: BadDamage [DAMAGE+0], request: XDamageDestroy[DAMAGE+2], resource: 0x740007e" ) kwin(4724) KWin::x11ErrorHandler: kwin: X Error ( "error: BadDamage [DAMAGE+0], request: XDamageDestroy[DAMAGE+2], resource: 0x1800068" ) ... I also tried to play with the "UseFastTLS" and the "DamageEvents" setting in xorg.conf's Device section, but no luck. I collected some stuff: xorg logs, xorg settings, glxinfos, kwin support, xsession-errors, screenshots, etc., say if you want more. I'm attaching as a compressed tar. PS: GLX_MESA_copy_sub_buffer seems wrong way, since all ati-drivers (incl. stables) has it.
Created attachment 84430 [details] xorg logs, xorg settings, glxinfos, kwin support, xsession-errors, screenshots, etc.
Can you verify comments #2 & #4 (means to recompile kwin)
Hi, just wanted to say that comment 2 fixed this *annoying* issue I've been having for *months*.
Fredrik thinks, the actual issue is in libkwineffects/kwinglutils.cpp:1442 void IndexBuffer::accomodate(int count) For a very wild shot, could you inject "glFlush();" behind glBufferData(GL_ELEMENT_ARRAY_BUFFER, size, NULL, GL_STATIC_DRAW); and behind glCopyBufferSubData(GL_COPY_READ_BUFFER, GL_ELEMENT_ARRAY_BUFFER, 0, 0, m_size); ?
I am seeing a similar problem with KDE 4.11 (on kubuntu 13.10) using fglrx 13.251 (catalyst 13.12). I will attach a screenshot. What can I do to determine if it is a duplicate bug or not?
Created attachment 85426 [details] fglrx 13.251 screenshot Corruption when minimizing or maximizing a window on KDE 4.11 when using fglrx 13.251
(In reply to comment #11) > What can I do to determine if it is a duplicate bug or not? Can you compile a patched version of kwin?
I made the changes suggested in comment #10 and it didn't resolve the issue. I also am not seeing anything added to .xsession-errors which makes me believe I am seeing a different problem.
See comment #2 - it's nowhere clear whether the suggestion in comment #10 could resolve this (well, unless the change from comment #2 resolves it for you ;-)
(In reply to comment #15) > See comment #2 - it's nowhere clear whether the suggestion in comment #10 > could resolve this (well, unless the change from comment #2 resolves it for > you ;-) The changes from Comment #2 stop the glitches. Anything else you want me to try (so I can use indexed quads if they are helpful)?
This bug report hasn't seen any comment for almost 3 years. Given that I assume that it's no longer an issue and was a problem with the driver. In case it still happens, please reopen.