At 08:02 PM 12/3/2004, Tim Kelly wrote:
>At 1:11 PM -0800 12/3/04, Matt Thomas wrote:
> >Since we are in the idle routine, it doesn't really matter when interrupts
> >are enabled.  So enabling them before lowering IPL makes a lot of sense.
>
>In all honesty, I think this really says a lot about the state of this
>port. As has been pointed out to me, I don't know that much about Unix
>underpennings, yet a bug that takes all of moving the enabling of an
>interrupt to before spllower is called to fix is ignored until a newbie
>like me figures out that this is keeping a CPU unable to receive external
>interrupts while processing deferred interrupts. Apparently it really
>_does_ matter when interrupts are enabled, and this was a really stupid
>bug. You'd think the person responsible for maintaining the locore_subr.S
>code would at least acknowledge that they should have caught this a long
>time ago.

Believe it or not, I did not write that code.  While I have endeavored to
clean it up, I have not tried rewriting it from scratch.  I did that
for the OEA pmap code and that enough pain for me.  Enough though I should
have learned by now, I trust that the original author knew what he was doing.
In some cases, such as this, that trust is misplaced.

In reality, I don't spend much time on the macppc port; I have other
powerpc ports to waste my time on.

>Is this the best effort core can put forth for the macppc port?

Effort on a volunteer project is always hit and miss.  First you have
find someone with the interest and then hope they also have the ability.
It rarely happens.

-- 
Matt Thomas                     email: matt@3am-software.com
3am Software Foundry              www: http://3am-software.com/bio/matt/
Cupertino, CA              disclaimer: I avow all knowledge of this message.