On 09/27/2018 10:53 AM, Andrew Haley wrote: > On 09/27/2018 09:30 AM, Aleksey Shipilev wrote: >> I was mostly concerned with having the same control flow as before, >> did I miss some change that is actually non-trivial? > > That's not my point: we should not put break statements in places we > don't expect to reach. It's misleading for the reader. Code changes > to "shut up" the compiler have do be done with great care and > knowledge of what the code does.
Okay, fair point. >> On the second read, this change in c1_LIRAssembler_aarch64.cpp looks >> suspicious, as it elevates ShouldNotReachHere to default case, >> rather than letting default thing fall-through? >> >> break; >> + default: >> ShouldNotReachHere(); >> } > > I think that one is actually OK. > > http://cr.openjdk.java.net/~aph/8211207/ Looks good! -Aleksey