Bug 300527
| Summary: | [WPE] API test TestDisplayWayland failing on dual GPU systems | ||
|---|---|---|---|
| Product: | WebKit | Reporter: | Lauro Moura <lmoura> |
| Component: | WPE WebKit | Assignee: | Nobody <webkit-unassigned> |
| Status: | NEW | ||
| Severity: | Normal | CC: | bugs-noreply |
| Priority: | P2 | ||
| Version: | WebKit Nightly Build | ||
| Hardware: | Unspecified | ||
| OS: | Unspecified | ||
Lauro Moura
./Tools/Scripts/run-wpe-tests --release WPEPlatform/TestDisplay WPEPlatform/TestDisplayWayland
On my laptop (dual intel/nvidia), the TestDisplayWayland are failing due to the following (fatal) warning:
** (TestDisplayWayland:569583): WARNING **: 13:10:40.312: Infered DRM device (/dev/dri/card1) using libdrm but multiple were found, you can override this with WPE_DRM_DEVICE
It's emitted by `wpeDRMDeviceCreateForDevice`, which `WPEDisplayWayland` calls with nullptr. Meanwhile, `WPE_DRM_DEVICE` is just checked lazily in `wpe_display_get_drm_device`, before asking the child class for the drm device.
In fact, IIUC, currently, if we use `WPE_DRM_DEVICE, we can have the scenario where `WPEDisplayWayland` might be using a given `WPEDRMDevice` (its `priv->drmDevice`) that's different from the one returned by `wpe_display_get_drm_device`, which would be `WPEDisplay`'s `priv->overridenDRMDevice`.
| Attachments | ||
|---|---|---|
| Add attachment proposed patch, testcase, etc. |