Messages in this thread | | | Date | Fri, 18 Jul 2008 09:43:04 +0100 | From | "Jan Beulich" <> | Subject | RE: ACPI WARNING: at drivers/acpi/tables/tbfadt.c:348acpi_tb_create_local_fadt+0x147/0x2f4() |
| |
>>> "Moore, Robert" <robert.moore@intel.com> 17.07.08 19:20 >>> >So far, in the number of the cases like this that I've seen, it's the v2 >fields that have problems. Perhaps the heuristic should be something >like "if there is an inconsistency between the v1 and v2 fields, fall >back to v1".
While extending the patch to do so, I realize that other v2 fields are used as-is, no matter whether their bit_width (or other fields) are wrong. Is that perhaps why hardware/hwregs.c uses hard-coded constants rather than the specified widths? If so (and if the v1 fields are considered reliable), shouldn't the v2 ones be sanity-checked against the v1 ones and then the specified widths be used as intended by the spec?
Jan
| |