Fix for setting/clearing breaks take 2

[Jonathan Larmour]

John Newlin wrote:
> Jonathan Larmour wrote:
> 
>> John Newlin wrote:
>>
>>> My previous patch had a small bug in it.  It really works this time.  ;)
>>
>>
>>
>> Can you give a bit more background on this? This patch doesn't seem 
>> right to me.... surely you should be defining things so that 
>> "*(t_inst*)pc = (t_inst)HAL_BREAKINST;" will work?
> 
> 
> Unfortunately that does not work on the Xtensa architecture.  Xtensa has 
> instructions that are 2 bytes or 3 bytes.  The break instruction is 
> 2-bytes, however it can be on any alignment (due to the intermixed 
> 3-byte instructions throwing off the alignment).  Yes, the architecture 
> is weird that way, instructions can be on any alignment, but the 
> load/store instructions require aligned data.  It does make for some 
> compact code though.  :)

Yes that's certainly exotic. Thanks for the explanation (and the rest I 
snipped). I've checked in your patch. Thanks!

Jifl
-- 
eCosCentric    http://www.eCosCentric.com/    The eCos and RedBoot experts
--["No sense being pessimistic, it wouldn't work anyway"]-- Opinions==mine