Reviewers: Toon Verwaest,
Message:
Please take a look. See the bug for the reasoning. This doesn't seem to
affect
our usual benchmarks.
Is adding a test worth it? We could only assert that the function from the
bug
stays optimized; I'm not quite convinced of the usefulness of such tests.
Description:
HPhis with HParameter inputs must have Tagged representation
BUG=v8:3670
LOG=n
Please review this at https://codereview.chromium.org/702053002/
Base URL: https://v8.googlecode.com/svn/branches/bleeding_edge
Affected files (+6, -0 lines):
M src/hydrogen-instructions.h
Index: src/hydrogen-instructions.h
diff --git a/src/hydrogen-instructions.h b/src/hydrogen-instructions.h
index
810a2fdd520d26e0d636e97ee470681210f293ac..a610fdb94732088c81c3ce01a3f1b61564a3fea1
100644
--- a/src/hydrogen-instructions.h
+++ b/src/hydrogen-instructions.h
@@ -5312,6 +5312,12 @@ class HParameter FINAL : public
HTemplateInstruction<0> {
return Representation::None();
}
+ virtual Representation KnownOptimalRepresentation() OVERRIDE {
+ // If a parameter is an input to a phi, that phi should not
+ // choose any more optimistic representation than Tagged.
+ return Representation::Tagged();
+ }
+
DECLARE_CONCRETE_INSTRUCTION(Parameter)
private:
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
