On 25/02/16 15:07, Andrew Cooper wrote: > On 25/02/16 14:27, Jan Beulich wrote: >>>>> On 25.02.16 at 14:16, <andrew.coop...@citrix.com> wrote: >>> No OS is plausibly going to hide non-IP information in FIP. >>> >>> If some theoretical OS does do something like that, there is always the >>> override available. >> But who says it needs to be an OS to do so? User mode code could >> as much (and would imo even be more likely to do so, compared to >> an OS). > > Why does this matter? We *cannot* make a perfect heuristic in this case. > > The new heuristic is wrong substantially less often than the old > heuristic, so is an improvement.
Well, no. The current one is wrong 0.01% of the time[1], and is now wrong 0.0100000000000000001% of the time. I would suggest that this isn't a significant different. Anyway, since it was trivial to reorder the series, I'm posting a v4 with this controversial patch at the end. David [1] Based on the number of XenServer test cases it breaks. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xen.org http://lists.xen.org/xen-devel
