Re: [patch] x86_64: fix boot hang caused by CALGARY_IOMMU_ENABLED_BY_DEFAULT

[Ingo Molnar <mingo@elte.hu>]

* Muli Ben-Yehuda <muli@il.ibm.com> wrote:

> > but i still /strongly/ disagree with your attitude that mainline is 
> > 'experimental' and hence there's nothing to see here, move over.
> 
> We can agree to disagree about how "experimental" mainline should be. 
> [...]

there's not much to disagree about. Mainline early bootup _must not 
break_, and if it breaks, it must be easy for the tester to figure it 
out. Simple as that. If it ever breaks and the user cannot give us other 
feedback but: "my laptop hung", we screwed up the process!

once the system has booted up into a minimal state, up to the stage 
where say netconsole works, we've got an exponentially increasing number 
of measures to find /all the other bugs/. But early bootup is like 
sacred. It's not experimental at all. Really. Having a system that 
doesnt even boot and gives no feedback at all is an absolute showstopper 
and a lost tester to us.

if we need draconian measures such as having two versions of early 
bootup code present in the kernel and a runtime boot switch to trigger 
between the old-trusted and the new-unknown one [perhaps even 
automatically, via the help of Grub] then so it be - but we cannot 
tolerate hung bootups.

	Ingo