On Sep 12, 2011, at 1:19 PM, Michael Büsch wrote:
> On Mon, 12 Sep 2011 12:58:52 +0200
> [email protected] wrote:
>
>> On Sep 12, 2011, at 12:35 PM, Michael Büsch wrote:
>>
>>> On Mon, 12 Sep 2011 12:09:01 +0200
>>> [email protected] wrote:
>>>
>>>> but trying on a = 0x8000 and b = 0x7fff helped me understanding. a - b =
>>>> 0x8000 + 2'(0x7fff) = 0x0001 > 0. While 0x8000 < 0x7fff.
>>>
>>> I don't get it.
>>> Can you write simple pseudocode for your instruction?
>> Will try for jdn.
>>
>>> For example, for jls, we have this pseudocode:
>>>
>>> if (xxx < yyy)
>>> pc := jjj
>>> (where xxx and yyy are two's complement)
>>>
>>
>> ** jump if difference is negative
>>
>> 0d6 xxx yyy jjj
>>
>> if ( xxx - yyy < 0 )
>> pc := jjj
>>
>> C-pseudocode for jdn
>>
>> short c = xxx - yyy;
>> if ( c < 0 )
>> goto jjj;
>
> Ok. So the existing signed-compare jumps look at the carry of the subtraction
> operation, but the new jdX instructions look at bit 0x8000 of the subtraction
> result. (if set -> negative, otherwise positive). jdX ignores the carry.
Exactly. For instance for jdn we have
0d6 xxx yyy jjj
tmp := xxx - yyy;
if( tmp & 0x8000 )
pc := jjj
or alternatively
tmp := xxx - yyy;
if( tmp < 0 )
pc := jjj
Tested both in the firmware and they are the same.
Do you agree with those names?
Thanks,
-Francesco
_______________________________________________
b43-dev mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/b43-dev
