zequanwu added a comment.
Thanks for reviewing.
================
Comment at: clang/test/CoverageMapping/switch.cpp:62
default: // CHECK-NEXT: File 0, [[@LINE]]:3 -> [[@LINE+1]]:10 = #4
break; // CHECK-NEXT: Branch,File 0, [[@LINE-1]]:3 ->
[[@LINE-1]]:10 = #4, (#0 - #4)
+ }
----------------
vsk wrote:
> zequanwu wrote:
> > vsk wrote:
> > > The blank space after `default: break;` and before the closing '}' should
> > > have the same count as the switch condition, I thought (this goes for all
> > > of the unreachable code within a switch body)? The idea is to prevent
> > > 'not-executed' regions from appearing after the `break` stmt.
> > Yes, they are unreachable. So, shouldn't they always have zero count
> > instead of same count as switch condition?
> I see, you're right. We use a gap region with zero count to cover the area
> after break. That's actually documented here:
> https://clang.llvm.org/docs/SourceBasedCodeCoverage.html#switch-statements.
> One interesting point is that there's only one gap region that covers the
> whole body; in this example, that would be `Gap,File 0, [[@LINE]]:13 ->
> [[@LINE+8]]:10 = 0`. So, perhaps the new gaps emitted in switches are
> redundant, but getting rid of them is just a performance optimization.
Yes, they are redundant.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D97101/new/
https://reviews.llvm.org/D97101
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
