@@ -1419,8 +1419,11 @@ int analogix_dp_bind(struct device *dev, struct drm_device *drm_dev,
goto err_disable_pm_runtime;
}
- phy_power_off(dp->phy);
pm_runtime_put(dev);
+ /*
+ * intentionally keep PHY powered on, Chromebook2 Peach-PIT board
+ * freezes otherwise
+ */
return 0;
@@ -1452,6 +1455,7 @@ void analogix_dp_unbind(struct device *dev, struct device *master,
drm_dp_aux_unregister(&dp->aux);
pm_runtime_disable(dev);
+ phy_power_off(dp->phy);
clk_disable_unprepare(dp->clock);
}
EXPORT_SYMBOL_GPL(analogix_dp_unbind);
Patch f0a8b49c03d2 ("drm/bridge: analogix dp: Fix runtime PM state on driver bind") fixed unbalanced call to phy_power_on() in analogix_dp_bind() function by calling phy_power_off() at the end of bind operation. However it turned out that having PHY powered is required for proper DRM display pipeline initialization on Peach-Pit Chromebook2 board, as this board freezes otherwise when PHY power off is called in bind. Fix this by keeping PHY powered for the whole bind/unbind driver life cycle. The freeze is probably related to the fact that the display pipeline (Exynos FIMD CTRC -> Exynos/Analogix DP bridge -> PS8625 dp2lvds bridge -> B116XW03 panel) is already configured and enabled by the bootloader and requires reset of all components for proper shutdown. Having PHY powered is also needed for proper EDID handling, which has been fixed by commit 510353a63796 ("drm/bridge: analogix dp: Fix runtime PM state in get_modes() callback"). Fixes: f0a8b49c03d2 ("drm/bridge: analogix dp: Fix runtime PM state on driver bind") Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com> --- Changelog: v2: - added comment in the code about Chromebook2 Peach-PIT issue --- drivers/gpu/drm/bridge/analogix/analogix_dp_core.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)