On Friday, January 04, 2013 12:16:48 AM Rafael J. Wysocki wrote: > Hi All, > > There are some problems with our handling of ACPI power resources that have > accumulated over time or have never been addressed. > > First of all, the interface represented by > acpi_power_resource_[un]register_device() doesn't really work as it should, > so it needs to be redesigned (details in the changelog of patch [1/12]). > > Second, power resources really have to be treated in a special way by the > ACPI namespace scanning code, because they may need to be ready to use as > soon as they are discovered. Moreover, the are represented by struct > acpi_device objects, so that they appear in sysfs along with "regular" ACPI > device nodes, they really don't represent devices, so they need special > configuration treatment in some cases. This special casing is obnoxiously > present in scan.c, but if we accept the fact that power resources really are > a special case, things can be simplified quite a bit (patch [3/12] with some > preparations in [2/12]). > > If the above changes are made, then it turns out that the entire > acpi_power_driver structure is not necessary any more and we don't need the > device power states of power resources for anything (patches [4-5/12]). > > Moreover, ACPI power resources are supposed to be ordered. Namely, they have > an attribute defining the order in which they are supposed to turned on and > off. Currently, though, we don't take that attribute into accound, although > we obviously should do that. After the previous changes that may be > implemented at a relatively low cost (patch [6-7/12]). > > Patches [8-9/12] move code around and clean it up in preparation for patch > [10/12], which consolidates the way in which power resources lists are > extracted from the output of _PRW and _PR[0-3]. > > Patch [11/12] improves the handling of power resource initialization errors > and patch [12/12] causes the system level attribute of wakeup power resources > to be taken into account during device setup. > > All patches are on top of the git branch at: > > git://git.kernel.org/pub/scm/linux/kernel/git/rafael/linux-pm.git linux-next > > Tested on HP nx6325 (and other boxes where ACPI power resources aren't used). There has been a week without a single comment, so I gather the thing is not controversial at all. I'll put it into linux-next for v3.9, then. Thanks, Rafael -- I speak only for myself. Rafael J. Wysocki, Intel Open Source Technology Center. -- To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
