On Tue, Sep 20, 2011 at 7:43 PM, Huang Ying <ying.huang@xxxxxxxxx> wrote: > On 09/21/2011 12:26 AM, Bjorn Helgaas wrote: >> On Mon, Sep 19, 2011 at 8:34 PM, Huang Ying <ying.huang@xxxxxxxxx> wrote: >>> On 09/20/2011 10:09 AM, Bjorn Helgaas wrote: >>>> On Mon, Sep 19, 2011 at 7:38 PM, Huang Ying <ying.huang@xxxxxxxxx> wrote: >>>>> Some firmware will access memory in ACPI NVS region via APEI. That >>>>> is, instructions in APEI ERST/EINJ table will read/write ACPI NVS >>>>> region. The original resource conflict checking in APEI code will >>>>> check memory/ioport accessed by APEI via general resource management >>>>> mech. But ACPI NVS region is marked as busy already, so that the >>>>> false resource conflict will prevent APEI ERST/EINJ to work. >>>>> >>>>> To fix this, this patch excludes ACPI NVS regions when APEI components >>>>> request resources. So that they will not conflict with ACPI NVS >>>>> regions. >>>> >>>> I think this is much, much too complicated. >>>> >>>> Yinghai's three-line e820.c patch to leave ACPI NVS regions in the >>>> iomem_resource tree, but as not busy, is far better. >>> >>> ACPI NVS should only be used by firmware or firmware interpreter instead >>> of the ordinary drivers. So I think that is reasonable to make it busy >>> in iomem resource tree. >> >> "My driver is not like ordinary drivers" is a common excuse for adding >> special cases. I don't buy it. >> >> These patches (3 and 4) add a lot of complexity but I don't believe >> they add any real protection. >> >> Regions are marked busy by their owners, i.e., by drivers that claim >> devices and know how to operate them. The e820 code is not an owner >> of ACPI NVS regions, so it should not mark them busy. >> >> I don't really think we have a problem here that needs to be solved. >> Ordinary drivers have no way of learning an address in ACPI NVS, so >> they aren't even going to try to use it. > > So what resource conflict checking is for? If something wrong with > driver configuration, resource description in ACPI table etc, the driver > may request iomem inside ACPI NVS regions. > > ACPI NVS regions already have a user, that is the ACPI AML interpreter, > so it is always busy. If the AML interpreter is the user, *it* should mark the regions busy, not e820. 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
