Re: [PATCH v4 RFC 1/6] x86/PCI: add config space write abstract intercept logic

[Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>]

On June 23, 2015 3:21:17 AM EDT, Jan Beulich <JBeulich@suse.com> wrote:
>>>> On 22.06.15 at 21:31, <konrad.wilk@oracle.com> wrote:
>>> @@ -1804,8 +1804,12 @@ static bool_t pci_cfg_ok(struct domain *
>>>              start |= CF8_ADDR_HI(currd->arch.pci_cf8);
>>>      }
>>>  
>>> -    return !xsm_pci_config_permission(XSM_HOOK, currd, machine_bdf,
>>> -                                      start, start + size - 1,
>write);
>>> +    if ( xsm_pci_config_permission(XSM_HOOK, currd, machine_bdf,
>>> +                                   start, start + size - 1,
>!!write) != 0 )
>>> +         return 0;
>>> +
>>> +    return !write ||
>>> +           pci_conf_write_intercept(0, machine_bdf, start, size,
>write) >= 0;
>> 
>> Won't the 'write' parameter cause an compiler error as it expects an 
>> pointer?
>
>No, certainly not - !write means the same as write != NULL, but is
>(imo) easier to read.

I meant the

 pci_conf_write_intercept(...,write).

The prototype for the last parameter is for *uint32?

>
>Jan