I sometimes have more ideas than I can handle in a day and my mind resets every night so every morning I am a blank slate like that guy from that movie. So I'm going to start writing them down and adding a hashtag. #StaticCheckerIdeas I'm going to CC kernel janitors until the Smatch list gets archived on lore.kernel.org There is a bug in ti_sci_probe() where it does: info = devm_kzalloc(dev, sizeof(*info), GFP_KERNEL); ... list_add_tail(&info->node, &ti_sci_list); ... return of_platform_populate(dev->of_node, NULL, NULL, dev); Idea 1: If of_platform_populate() fails, then info freed but it's it's still on the the &ti_sci_list list. Look for list_add of devm_ data and ensure that it's removed from the list before an error code. There are other lifetime issues surrounding devm_ data. We need a list_del in the release function as well. Idea 2: The other error paths all have clean up so last error path should have clean up too. regards, dan carpenter
