Zero and Shark fix

[Gary Benson]

Hi all,

This commit removes a hack that became unnecessary with the updated
frame anchor code in my last commit.   Xerxes, this might be the
source of the failures you were seeing.

Cheers,
Gary

-- 
http://gbenson.net/
-------------- next part --------------
diff -r 9bc0c7b7f327 ChangeLog
--- a/ChangeLog	Mon May 10 11:44:31 2010 +0100
+++ b/ChangeLog	Tue May 11 09:26:27 2010 +0100
@@ -1,3 +1,8 @@
+2010-05-11  Gary Benson  <gbenson at redhat.com>
+
+	* patches/icedtea-shark.patch
+	(Deoptimization::create_vframeArray): Remove a now-unnecessary hack.
+
 2010-05-10  Gary Benson  <gbenson at redhat.com>
 
 	PR icedtea/323
diff -r 9bc0c7b7f327 patches/icedtea-shark.patch
--- a/patches/icedtea-shark.patch	Mon May 10 11:44:31 2010 +0100
+++ b/patches/icedtea-shark.patch	Tue May 11 09:26:27 2010 +0100
@@ -101,27 +101,6 @@
  
    UnrollBlock* info = new UnrollBlock(array->frame_size() * BytesPerWord,
                                        caller_adjustment * BytesPerWord,
-@@ -890,7 +897,20 @@
-   // stuff a C2I adapter we can properly fill in the callee-save
-   // register locations.
-   frame caller = fr.sender(reg_map);
-+#ifdef ZERO
-+  int frame_size;
-+  {
-+    // In zero, frame::sp() is the *end* of the frame, so
-+    // caller.sp() - fr.sp() is the size of the *caller*.
-+    RegisterMap dummy_map(thread, false);
-+    frame frame_1 = thread->last_frame();
-+    frame frame_2 = frame_1.sender(&dummy_map);
-+    assert(frame_2.sp() == fr.sp(), "should be");
-+    frame_size = frame_2.sp() - frame_1.sp();
-+  }
-+#else
-   int frame_size = caller.sp() - fr.sp();
-+#endif // ZERO
- 
-   frame sender = caller;
- 
 @@ -1057,7 +1077,7 @@
  JRT_END