| Summary: | Vivante: Memory corruption in wayland client process after hiding window | ||||||
|---|---|---|---|---|---|---|---|
| Product: | [Build System, Metadata & Runtime] BSPs | Reporter: | Wieland Hagen <wieland.hagen> | ||||
| Component: | bsps-meta-fsl-arm | Assignee: | Tom Hochstein <tom.hochstein> | ||||
| Status: | RESOLVED WONTFIX | QA Contact: | |||||
| Severity: | normal | ||||||
| Priority: | Low | CC: | otavio, prabhusundar | ||||
| Version: | 2.1.2 | ||||||
| Target Milestone: | Future | ||||||
| Hardware: | Other | ||||||
| OS: | arm | ||||||
| Whiteboard: | |||||||
| OS type for building Yocto: | --- | Type of Regression: | New (Never tested) | ||||
| Verified: | Documentation change: | No (bug/feature does not impact docs) | |||||
| Attachments: |
|
||||||
It is already fixed and it will be available in upcoming release. 5.0.11P8.6 As explained by Prabhu, this will be fixed when the next BSP release is out. That's good to hear. When will 5.0.11P8.6 be released? Is there already a change-log for the fixed issues? And in case it's not possible for us to use a new yocto release, is it possible to use the new driver as a drop-in replacement for our system (kernel version 4.1.15-1.1.0+ga4d2a08)? Waiting For Upstream status is being removed, changing to Accepted. bsps-meta-fsl-arm issues are now tracked in GitHub. Please see: https://github.com/Freescale/meta-freescale |
Created attachment 3436 [details] Testcase app See the attached testcase. When hiding a wayland window by attaching a NIL-buffer to the wayland surface, memory corruptions occur, which are visible later on by malloc assertions failing. This is only reproducible when - a shader program is used while rendering - after attaching the NIL-buffer we call 1. eglDestroySurface(display, surface) 2. eglCreateWindowSurface(display, cfg, window, 0) 3. eglMakeCurrent(display, surface, surface, context) (this is what's happening in QtWayland when calling QWindow::hide() with the QtWayland backend) The attached code shows the bug. Run like ./wayland_window_hide 100 bug on a target device. This code runs fine on intel or nvidia host machines.