Re: [PATCH 9/9] drivers/acpi: Use explicitly specified types

[Jan Beulich <jbeulich@xxxxxxxx>]

On 22.06.2022 13:09, Michal Orzel wrote:
> On 22.06.2022 12:36, Jan Beulich wrote:
>> On 20.06.2022 09:02, Michal Orzel wrote:
>>> --- a/xen/drivers/acpi/tables/tbutils.c
>>> +++ b/xen/drivers/acpi/tables/tbutils.c
>>> @@ -481,7 +481,7 @@ acpi_tb_parse_root_table(acpi_physical_address 
>>> rsdp_address, u8 flags)
>>>                     if (ACPI_FAILURE(status)) {
>>>                             ACPI_WARNING((AE_INFO,
>>>                                           "Truncating %u table entries!",
>>> -                                         (unsigned)
>>> +                                         (unsigned int)
>>>                                           (acpi_gbl_root_table_list.size -
>>>                                            acpi_gbl_root_table_list.
>>>                                            count)));
>>
>> Same here then, except PRIu32 wouldn't be correct to use in this case.
>>
> Why is it so given that both size and count are of type u32?

Because the promoted type (i.e. the type of the result of the subtraction)
isn't uint32_t (and will never be). It'll be "unsigned int" when
sizeof(int) == 4 (and in this case it'll happen to alias uint32_t) and
just "int" when sizeof(int) > 4 (not even aliasing int32_t).

Jan