Ok, so now I've spent the time reading all of the posts (thanks, I have nothing better to do with a toddler in the house :/) and I still have to stand by the statement that Linus, whilst often blunt to the point of traumatic, is no liar. He's quite honest.
The apology only comes AFTER being chewed out -- the initial stand from Mauro is erring on the side of stating that PA is broken (which it may be, but that's not the point here; the point is that Linus has made it very clear in the past that kernel changes cannot break userspace because when they do, users (desktop and server) as well as userland developers lose confidence in the base system. It's bad for everyone.
I don't think you fully understand the terms "cowboy" and "rock star" and just how detrimental these people are. Sure, I only described "cowboy" for you, but since you're so intent on full completion, let me remind you that rock stars don't work well in teams because they believe their stuff is simply better than everyone else's. Often, they may even be right -- which causes a feedback loop that just makes the situation worse. I've had to maintain a rock star's code before -- sure, it was good, functional, fast. It was also incredibly difficult to read and when I fixed a flaw which became quite obvious, the tantrum that ensued made sure that I would never helpfully commit to that code base again. I had to interface with that code so when I found routines that were of interest to me, I would often rip them out, make them readable, fix any edge cases and keep them in my repo -- instead of just committing the (imo) higher quality code back to the original author.
Regarding your last comment: I still think that the linked mail from Linus (whilst quite blunt) is not completely off-kilter. A project of any reasonable size requires a technical lead who understands the project and its requirements and the unbreakable rules that come with those requirements. Linus is that guy and I think he does a reasonable job of it considering the size and requirements of the project. It's his leadership which gives me confidence in the kernel -- because it's no longer a hackthing: it's actually a solid work now (and has been for quite a few years; bear in mind where Linux came from) and Linus is putting that ahead of the feelings of the contributors -- which, at the end of the day, is good for the user and, eventually the contributors.
Mauro is properly apologetic *after* being chewed out. And I believe that (a) he won't make a mistake like that again and (b) he will be a useful member of the contributing team, having learned that the rules that have been established for the project are, in fact, immutable. He's no arch-villain, in the same way that Linus is no hero: Mauro made a mistake, got flamed and learned. Linus is just doing his job.