Messages in this thread | | | From | Thomas Renninger <> | Subject | Re: [Devel] [PATCH 1/2] ACPICA: Introduce acpi_os_phys_table_override function | Date | Wed, 31 Aug 2011 11:28:49 +0200 |
| |
On Wednesday, August 31, 2011 04:43:42 AM Lin Ming wrote: > On Wed, 2011-08-24 at 17:48 +0800, Thomas Renninger wrote: ... > You add a new interface. Yes, is this a bigger problem?
> Can we just extend the existing interface: acpi_os_table_override? Not sure how to do that without OS/ACPICA API changes. The virtual address handling is nasty. You have to differ early mappings (early_ioremap) and later mappings (io/memremap). Re-mapping later is not possible because the physical address is lost with the current overriding interface. The physical address usage is transparent and from what I can see the only way to provide proper table overriding.
If it's ok to add more paramters to acpi_os_table_override and either pass the virtual (as before) or the physical address, this would work: acpi_os_table_override(struct acpi_table_header *existing_table, struct acpi_table_header **new_table, acpi_physical_address *address, u32 *table_length);
This would be an interface change which looked even worse to me, than adding a new function.
Thomas
| |