From maxim.kartashev at jetbrains.com Thu Jun 2 14:18:38 2022 From: maxim.kartashev at jetbrains.com (Maxim Kartashev) Date: Thu, 2 Jun 2022 17:18:38 +0300 Subject: Another XWayland-related issue Message-ID: I just filed a bug (8287731 [XWayland] Portions of window get painted with solid white color) that has something to do with XRender implementation in XWayland. The bug manifests itself as white rectangles appearing instead of portions of the window that are painted from pre-created translucent "compatible images" (GraphicsConfiguration.createCompatibleImage()). It doesn't reproduce always, but shows up often enough. Disabling Glamor *or* switching to the OpenGL pipeline makes the bug go away. At this point it is still not clear whether this is a Java thing or something's wrong with the XRender implementation inside Xorg or even farther down the stack. If you have seen this one before and/or have a suggestion on how to further narrow down the bug's cause, please respond. P.S. I have only looked at this inside a VMWare virtual machine instance, so not 100% sure that this can be seen on "real" hardware. Ref: https://bugs.openjdk.java.net/browse/JDK-8287731 From ofourdan at redhat.com Thu Jun 2 14:41:44 2022 From: ofourdan at redhat.com (Olivier Fourdan) Date: Thu, 2 Jun 2022 16:41:44 +0200 Subject: Another XWayland-related issue In-Reply-To: References: Message-ID: Hi Maxim, On Thu, Jun 2, 2022 at 4:19 PM Maxim Kartashev wrote: > [?] > > If you have seen this one before and/or have a suggestion on how to further > narrow down the bug's cause, please respond. > > P.S. I have only looked at this inside a VMWare virtual machine instance, > so not 100% sure that this can be seen on "real" hardware. > > Ref: > https://bugs.openjdk.java.net/browse/JDK-8287731 Since disabling glamor gets rid of the issue, then the root cause is between GLAMOR (in the Xserver) and the underlying GL implementation. For example I reckon the somehow similar rendering issue [2] reported by Alexander is more of a GL implementation bug in the virtualization stack than GLAMOR (GLAMOR being used by Xwayland and Xorg with modesetting for example). So to help narrow the issue, the next thing would be to reproduce on real hardware. FWIW, I just tried that on my intel laptop here and I could not reproduce the issue, no matter how many times I repeated the test. I also tried on an NVIDIA laptop with the NVIDIA proprietary driver (to test another GL implementation, not just Mesa) and it works as well. So I suspect this is the same as with [2], it's possibly more of a GL implementation issue in the virtualization stack. Cheers Olivier [2] https://gitlab.freedesktop.org/xorg/xserver/-/issues/1333 From maxim.kartashev at jetbrains.com Thu Jun 2 16:51:53 2022 From: maxim.kartashev at jetbrains.com (Maxim Kartashev) Date: Thu, 2 Jun 2022 19:51:53 +0300 Subject: Another XWayland-related issue In-Reply-To: References: Message-ID: Thanks for your analysis, Olivier! FYI: I filed an XOrg issue here: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1349 On Thu, Jun 2, 2022 at 5:42 PM Olivier Fourdan wrote: > Hi Maxim, > > On Thu, Jun 2, 2022 at 4:19 PM Maxim Kartashev > wrote: > > [?] > > > > If you have seen this one before and/or have a suggestion on how to > further > > narrow down the bug's cause, please respond. > > > > P.S. I have only looked at this inside a VMWare virtual machine instance, > > so not 100% sure that this can be seen on "real" hardware. > > > > Ref: > > https://bugs.openjdk.java.net/browse/JDK-8287731 > > Since disabling glamor gets rid of the issue, then the root cause is > between GLAMOR (in the Xserver) and the underlying GL implementation. > > For example I reckon the somehow similar rendering issue [2] reported > by Alexander is more of a GL implementation bug in the virtualization > stack than GLAMOR (GLAMOR being used by Xwayland and Xorg with > modesetting for example). > > So to help narrow the issue, the next thing would be to reproduce on > real hardware. > > FWIW, I just tried that on my intel laptop here and I could not > reproduce the issue, no matter how many times I repeated the test. > > I also tried on an NVIDIA laptop with the NVIDIA proprietary driver > (to test another GL implementation, not just Mesa) and it works as > well. > > So I suspect this is the same as with [2], it's possibly more of a GL > implementation issue in the virtualization stack. > > Cheers > Olivier > > [2] https://gitlab.freedesktop.org/xorg/xserver/-/issues/1333 > > From duke at openjdk.org Thu Jun 30 09:57:16 2022 From: duke at openjdk.org (duke) Date: Thu, 30 Jun 2022 09:57:16 GMT Subject: git: openjdk/wakefield: pure_wl_toolkit: Improved sun.awt.wl.WLGraphicsEnvironment to support createCraphics() Message-ID: Changeset: 3d2576cc Author: Alexey Ushakov Date: 2022-06-30 09:54:52 +0000 URL: https://git.openjdk.org/wakefield/commit/3d2576cc490299c10fc097cf7e74d9743233618c Improved sun.awt.wl.WLGraphicsEnvironment to support createCraphics() ! src/java.desktop/unix/classes/sun/awt/wl/WLGraphicsEnvironment.java From duke at openjdk.org Thu Jun 30 12:20:57 2022 From: duke at openjdk.org (duke) Date: Thu, 30 Jun 2022 12:20:57 GMT Subject: git: openjdk/wakefield: pure_wl_toolkit: Fixed crash on WLSD_Dispose Message-ID: Changeset: 0bc3e256 Author: Alexey Ushakov Date: 2022-06-30 14:18:47 +0000 URL: https://git.openjdk.org/wakefield/commit/0bc3e2567637114e3053ee0f3dd6190b8cd8f96f Fixed crash on WLSD_Dispose ! src/java.desktop/unix/native/common/java2d/wl/WLSurfaceData.c From duke at openjdk.org Thu Jun 30 15:30:52 2022 From: duke at openjdk.org (duke) Date: Thu, 30 Jun 2022 15:30:52 GMT Subject: git: openjdk/wakefield: pure_wl_toolkit: Replaced crash in Robot with logging Message-ID: <6328d5f6-2a34-43d4-b6cc-50f7724f10ed@openjdk.org> Changeset: 453aaead Author: Alexey Ushakov Date: 2022-06-30 17:30:21 +0000 URL: https://git.openjdk.org/wakefield/commit/453aaeadd32f4b17bf80c19674ac7ed4509eff66 Replaced crash in Robot with logging ! src/java.desktop/unix/classes/sun/awt/wl/WLComponentPeer.java ! src/java.desktop/unix/native/libawt_wlawt/WLRobotPeer.c