On Fri, Jun 26, 2009 at 3:56 PM, Robert Vincent > I care and I love the infrastructure that has been created. However, it > seems as though there are devices that do not conform to the paradigm or > maybe they are not truly in "low power" mode. My guess is the latter > otherwise there would be a flurry of complaints. Unfortunately, it's a little more complicated. In the beginning, none of the drivers did power management, and nobody cared because they were all PCI cards and nobody would notice a 1 watt difference in consumption on a 300 watt power supply. These devices would maintain persistent state across v4l closes because the chips were never powered down. Other devices, such as some USB devices, do have the power management hooks implemented. I don't know what the percentages are here (I would have to look at the driver code to figure that out). These devices power down the chips when asked to by the bridge, and it is typically triggered when no userland apps still have the v4l device open. In cases such as this, the power management works but it would break cases where people used scripts to control the device. There's probably a third class of devices worth mentioning: those that really should be doing power management but aren't. This includes all the USB devices which burn your fingers and drain your laptop battery from the time you plug it in until the time you unplug it, regardless of whether you are using it. It's not about "caring" or how much we do or do not love the infrastructure. It's about deciding what are the most important goals given limited developer resources. In this case, it's a question of which is more important: incurring the cost of overhauling all the drivers that do power management to have state persist after a power down versus telling those users who use scripts to manually disable power management. Since we have no idea how many users use scripts to control their tuners, and right now we don't know how many devices are effected, we cannot really make any decisions one way or the other. My hope is to see power management properly implemented in more drivers since I'm concerned about the environment. However, if I have to do a huge overhaul of the state management in every driver just to accommodate an unknown quantity of users, then I would have to think twice about that. Devin -- Devin J. Heitmueller - Kernel Labs http://www.kernellabs.com -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
