Re: The parisc SMP boot problem

On Sun, Feb 18, 2007 at 11:44:13AM -0600, James Bottomley wrote:
> After a bit more of an analysis of our changesets (I concluded it must
> have been something we introduced).  This is the source of the SMP boot
> hang on 2.6.20-rcX:
> 
> commit 01a9fa750904beed08814364fcbd7039eca8ec71
> Author: Randolph <[email protected]>
> Date:   Mon Dec 11 16:07:51 2006 -0800
> 
>     Optimize TLB flush on SMP systems
>     
>     Signed-off-by: Randolph Chung <[email protected]>
> 
> As predicted, the reason it passed all the tests at the time it was
> introduced is that because of the piggin bug, we were only scheduling on
> a single CPU, and so the SMP piece of this wasn't truly tested.
> 
> Reverting this patch allows ioz to boot correctly.
> 

Ok. I'll revert it in the for-linus tree as well.

Any idea what was wrong with it?

Cheers,
	Kyle