I'm by no means either a lawyer or a license expert, and even less so on (I forgot the name) the license Sun used for OpenSolaris, but from what I've read on the topic (partly in the context of the ZFS GPL2 license incompatibility, thus making it Linux kernel licensing incompatible), one thing they did that the GPL2 didn't do, is include language specifically designed to address patents. While how far the fork could diverge and remain within that safety net I'm not going to even venture a guess at, it does seem they should be safe enough at least as long as they stay relatively conservative in their "creativity". Of course, if that's indeed a limiting factor of the patent coverage, it's something Oracle could well be counting on to limit viability, longer term.
I'd certainly do well my legal consulting before I ventured to use the code for other purposes, however, even if explicitly covered by the same license.
(That's where Google got into trouble, as they explicitly chose NOT to go with Sun's chosen GPL+classpath-exceptions, since it would limit people building proprietary code on it and they didn't believe that was commercially viable, while attempting and failing to come to an agreement with Sun on a commercial Java license, as well. So Google's not even covered by the implied patent coverage of the GPL2. And given that they /did/ try to negotiate a Java license and did a reimplementation when that failed, Oracle would at least at at first glance have a genuine reason for the treble damages for deliberate infringement claim that they're making. FWIW, I don't believe Oracle's as much a bad guy here as they're being made out to be, particularly as the case could be seen as direct support for the GPL, given Google's explicit choice NOT to abide by the terms of the GPL under which Java was open sourced. That's also likely why OIN, the SFLC, etc, aren't speaking out. Google clearly took liberties with the code and license here, and arguably Oracle has every right to nail them for it. OTOH, it may well be that Google clean-roomed it well enough that the copyrights don't apply and avoided the specifics of the patents. I don't think anyone outside of Google itself really knows how well they did that, and they may have simply decided it was an untested legal area and they'd take their chances. That would appear to be what the court case is likely to turn on, from my specifically not professionally qualified viewpoint. It certainly doesn't seem anything like another SCO to me, nor, really, does there seem any indication Oracle is headed that way, because their OpenSolaris repositioning could in fact be seen as choosing to more directly support Linux at least on the open source front, as well, while keeping Solaris more proprietary. Time will tell, I guess.)