@@ -903,7 +903,8 @@ static void g2d_runqueue_worker(struct work_struct *work)
g2d->runqueue_node = NULL;
if (runqueue_node) {
- pm_runtime_put(g2d->dev);
+ pm_runtime_mark_last_busy(g2d->dev);
+ pm_runtime_put_autosuspend(g2d->dev);
complete(&runqueue_node->complete);
if (runqueue_node->async)
@@ -1022,7 +1023,8 @@ static void g2d_wait_finish(struct g2d_data *g2d, struct drm_file *file)
* the IRQ which triggers the PM runtime put().
* So do this manually here.
*/
- pm_runtime_put(dev);
+ pm_runtime_mark_last_busy(dev);
+ pm_runtime_put_autosuspend(dev);
complete(&runqueue_node->complete);
if (runqueue_node->async)
@@ -1519,6 +1521,8 @@ static int g2d_probe(struct platform_device *pdev)
goto err_destroy_workqueue;
}
+ pm_runtime_use_autosuspend(dev);
+ pm_runtime_set_autosuspend_delay(dev, 2000);
pm_runtime_enable(dev);
clear_bit(G2D_BIT_SUSPEND_RUNQUEUE, &g2d->flags);
clear_bit(G2D_BIT_ENGINE_BUSY, &g2d->flags);
@@ -1590,6 +1594,7 @@ static int g2d_remove(struct platform_device *pdev)
/* There should be no locking needed here. */
g2d_remove_runqueue_nodes(g2d, NULL);
+ pm_runtime_dont_use_autosuspend(&pdev->dev);
pm_runtime_disable(&pdev->dev);
g2d_fini_cmdlist(g2d);
The runqueue worker currently issues a get() when a new node is processed, and a put() once a node is completed. The corresponding suspend and resume calls currently only do clock gating, but with the upcoming introduction of IOMMU runpm also the corresponding IOMMU domain gets enabled (for get()) and disabled (for put()). This introduces performance regressions with we mitigate here. Switch PM runtime to autosuspend, such that clock gating and IOMMU control only happens when the engine is idle for a 'long' time. Signed-off-by: Tobias Jakobi <tjakobi@math.uni-bielefeld.de> --- drivers/gpu/drm/exynos/exynos_drm_g2d.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-)