On 28 July 2011 16:45, Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > My experience in this area is _extremely_ limited. Take a look at > commits b14e033e17d0ea0ba (PNPACPI: Add support for remote wakeup) and > 3e6e15a862a0bc201 (Input: enable remote wakeup for PNP i8042 keyboard > ports). Thanks for the pointers. Indeed, in b14e033e17d0ea0ba you are enabling wakeup on pnp (ACPI) devices, not the actual input device. In the sysfs tree, the pnp device is completely disconnected from the device hierarchy that includes the actual input device (the one controlled in drivers/input/input.c). By this I mean that with your patch, in order to enable/disable keyboard wakeups on my non-OLPC laptop I need to: echo -n disabled > "/sys/bus/pnp/drivers/i8042 kbd/00:06/power/wakeup" However, the input device that is responsible for my keyboard is /sys/devices/platform/i8042/serio0/input/input3 - and there is no linkage between that and the pnp node. If it would be more appropriate, I could also create a disconnected and distant /sys node for OLPC's keyboard/mouse wakeup capability, matching the way that ACPI works. But I thought it would be better to put the capability directly in the i8042/input device hierarchy. (and then we have the separate problem with i8042 input devices being unconditionally reset in 3 layers, losing the event that woke the system - this would be harder to solve if the wakeup-capable node was distant/unrelated) > At this point I have reached my level of incompetence... so I'll shut > up now. :-) Cool! I finally found something that you don't know everything about! Thanks for the help! I'm also new to this area and welcome some kind of verification/criticism of the design I have come up with. Daniel _______________________________________________ linux-pm mailing list linux-pm@xxxxxxxxxxxxxxxxxxxxxxxxxx https://lists.linux-foundation.org/mailman/listinfo/linux-pm
