Re: [Xen-devel] [PATCH v2 2/2] x86/mmcfg/drhd: Move acpi_mmcfg_init() call before calling acpi_parse_dmar()

["Jan Beulich" <JBeulich@xxxxxxxx>]

>>> On 20.08.18 at 11:38, <zhenzhong.duan@xxxxxxxxxx> wrote:
> On 2018/8/20 15:45, Jan Beulich wrote:
>>>>> On 20.08.18 at 05:38, <zhenzhong.duan@xxxxxxxxxx> wrote:
>>> I'm thinking about moving below piece of code earlier too, and I checked
>>> pci_mmcfg_check_hostbridge() carefully, it's secure, what do you think
>>> about that?
>>>
>>>       mmio_ro_ranges = rangeset_new(NULL, "r/o mmio ranges",
>>>                                     RANGESETF_prettyprint_hex);
>>>
>> 
>> That's a reasonable thing to do, and is (as pointed out) a necessary
>> prereq. But to be very clear - you'll also have to prove it's sufficient,
>> and for that it doesn't suffice to consider pci_mmcfg_check_hostbridge()
>> alone.
> Not sure how to prove, I checked over acpi_mmcfg_init() carefully, 
> acpi_disabled and DMI info are used and they are initialized earlier 
> than acpi_dmar_init() call, I only found mmio_ro_ranges need to be moved.

But that's only half of it: Checking just acpi_mmcfg_init() is insufficient.
You also need to check everything between the old and new call sites.
And the result of this checking wants to be summarized (read: in a
brief but nevertheless sufficient form) in the patch description.

Jan



_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-devel