Re: Re: 4.0/4.1 requests - IO-APIC EOI v4 [RFC]

[Keir Fraser <keir.xen@gmail.com>]

On 12/09/2011 11:23, "Jan Beulich" <JBeulich@suse.com> wrote:

>>>> On 12.09.11 at 12:15, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
>> On 12/09/11 07:50, Jan Beulich wrote:
>>>>>> On 09.09.11 at 18:47, Andrew Cooper <andrew.cooper3@citrix.com> wrote:
>>>> +                     * Xen code doesn't actually prevent multiple IO-APIC
>>>> +                     * entries being assigned the same vector, so EOI all
>>>> +                     * pins which have the correct vector.
>>>> +                     *
>>>> +                     * Remove the following code when the above assertion
>>>> +                     * is fulfilled. */
>>>> +
>>> Why don't you just call __io_apic_eoi() recursively here?
>>> 
>>> Jan
>> 
>> If I call the function recursively, it will loop forever.  Anyway, the
> 
> Why would it loop forever? You get in here only with pin == -1, and
> for the recursive call you'd pass the pin number you determined.

Exactly. Please re-send the patch with the recursive call. It makes the
function quite a bit shorter.

 -- Keir

>> need to clear multiple pins is only temorary until George finishes his
>> per-device AMD interrupt remap patch which will enforce vector
>> uniqueness in each IO-APIC.  My expectation is that this issue will be
>> fixed in the next few weeks.
> 
> Sure.
> 
> Jan
>