The reported generation sizes does not add up correctly when DefNew GC promote to a full GC

Examples:
65.491: [GC (Allocation Failure) 65.491: [DefNew: 56773K->4616K(58560K), 0.0145990 secs]65.506: [Tenured: 130447K->84239K(130492K), 0.1636040 secs] 186077K->84239K(189052K), [Metaspace: 4477K->4477K(5424K)], 0.1783900 secs] [Times: user=0.18 sys=0.00, real=0.18 secs]

... should be 187220K->88855K(189052K)

100.226: [GC (Allocation Failure) 100.226: [DefNew: 76229K->76229K(78656K), 0.0000260 secs]100.226: [Tenured: 174572K->110727K(174784K), 0.2232460 secs] 250802K->110727K(253440K), [Metaspace: 4477K->4477K(5424K)], 0.2233610 secs] [Times: user=0.22 sys=0.00, real=0.23 secs]

... should be 250802K->186956K(253440K)

Impact=Low, just a logging issue
Likelihood=Medium, will happen when DefNew expects a promotion failure
Workaround=Low, can get the information from PrintHeapAtGC

ILW=LML -> P5