On 04.11.2021 15:21, Roger Pau Monné wrote:
> On Wed, Nov 03, 2021 at 03:40:55PM +0100, Jan Beulich wrote:
>> While commit XXXXXXXXXXXX ("x86/IOMMU: mark IOMMU / intremap not in use
>> when ACPI tables are missing") deals with apic_x2apic_probe() as called
>> from x2apic_bsp_setup(), the check_x2apic_preenabled() path is similarly
>> affected: The call needs to occur after acpi_boot_init() (which is what
>> calls acpi_iommu_init()), such that iommu_intremap getting disabled
>> there can be properly taken into account by apic_x2apic_probe().
>>
>> Signed-off-by: Jan Beulich <jbeul...@suse.com>
>
> LGTM. I cannot find any dependency between acpi_boot_init and having
> initialized the apic.
Sadly there is, and I've now learned that when believing I would test
the change yesterday I actually didn't (or else I would have spotted
the problem that I've spotted now), and instead I did boot an older
binary: acpi_process_madt() calls clustered_apic_check() and hence
relies on genapic to have got filled before.
I'll have to see if I can come up with some variant avoiding the issue,
but I suspect that's not going to be 4.16 material anymore then. My
first try is likely going to be to simply pull out acpi_iommu_init()
from acpi_boot_init().
Jan
