JVM: erase conditional suspension markers from inline-only functions

Otherwise, the generated bytecode is unnecessarily suboptimal in some
(arguably weird) cases.

In the JVM backend, this was an accidental regression in #3260, as I had
not noticed that effectively inline-only functions were handled by a
separate branch in FunctionCodegen. In JVM_IR, I'm pretty sure the
redundant markers have always been there as `isSuspensionPoint` in
ExpressionCodegen never checked for effectively-inline-only-ness.
This commit is contained in:
pyos
2020-04-07 18:08:33 +02:00
committed by Ilmir Usmanov
parent f6d9867fa4
commit 0037d5a7c8
6 changed files with 34 additions and 14 deletions
@@ -1391,6 +1391,11 @@ public class BytecodeTextTestGenerated extends AbstractBytecodeTextTest {
runTest("compiler/testData/codegen/bytecodeText/coroutines/doNotReassignContinuation.kt");
}
@TestMetadata("effectivelyInlineOnly.kt")
public void testEffectivelyInlineOnly() throws Exception {
runTest("compiler/testData/codegen/bytecodeText/coroutines/effectivelyInlineOnly.kt");
}
@TestMetadata("nonLocalReturn.kt")
public void testNonLocalReturn() throws Exception {
runTest("compiler/testData/codegen/bytecodeText/coroutines/nonLocalReturn.kt");