On Sunday 21 January 2007 18:27, Shaohua Li wrote: > On Fri, 2007-01-19 at 12:46 -0700, Bjorn Helgaas wrote: > > On Thursday 18 January 2007 18:12, Shaohua Li wrote: > > > > +device_initcall(acpi_reserve_resources); > > > Will this be called before PCI assigned resources to PCI devices? We use > > > fs_initcall in motherboard to avoid PCI devices use motherboard's > > > resources before. > > > > I think we're OK. Here's the current order: > > > > pnp_system_init fs_initcall drivers/pnp/system.c > > pcibios_assign_resources fs_initcall arch/i386/pci/i386.c > > acpi_reserve_resources device_initcall drivers/acpi/osl.c > > > > So pnp_system_init() will reserve all the motherboard resources before > > the PCI resources are assigned. What do you think? > > In this way, we could just remove acpi_reserve_resources as pci already > assigns resources to pci devices. The concern is motherboard might not > list all acpi resources in its _CRS, but I don't know if there are such > BIOS. Microsoft seems to reserve motherboard resources in the same way. > Maybe we could just remove acpi_reserve_resources and give it a spin in > -mm to see if anybody complains. I agree that in theory we should be able to remove acpi_reserve_resources() and just depend on pnp_system_init() reserving all the motherboard resources. But your concern is very valid: it looks like both HP DL360 (x86) and rx2600 (ia64) firmware list things in the FADT that aren't included in _CRS for motherboard devices. I think these are BIOS bugs, but for now, it seems safest to keep acpi_reserve_resources() to work around the bugs. Bjorn - 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
