[Mono-bugs] [Bug 70241][Wis] New - The GC commits memory on allocation

[bugzilla-daemon@bugzilla.ximian.com]

Please do not reply to this email- if you want to comment on the bug, go to the
URL shown below and enter your comments there.

Changed by bmaurer@users.sf.net.

http://bugzilla.ximian.com/show_bug.cgi?id=70241

--- shadow/70241	2004-12-05 21:57:48.000000000 -0500
+++ shadow/70241.tmp.17223	2004-12-05 21:57:49.000000000 -0500
@@ -0,0 +1,54 @@
+Bug#: 70241
+Product: Mono: Runtime
+Version: 1.1
+OS: 
+OS Details: 
+Status: NEW   
+Resolution: 
+Severity: 
+Priority: Wishlist
+Component: misc
+AssignedTo: mono-bugs@ximian.com                            
+ReportedBy: bmaurer@users.sf.net               
+QAContact: mono-bugs@ximian.com
+TargetMilestone: ---
+URL: 
+Cc: 
+Summary: The GC commits memory on allocation
+
+Consider the following test case:
+
+class X {
+	static void Main ()
+	{
+		int size = int.MaxValue;
+		do {
+			try {
+				int [] x = new int [size];
+				
+				x [1] = 1;
+				x [size / 2] = 1;
+				x [size / 3] = 1;
+				x [size - 1] = 1;
+			} catch {
+			}
+		} while ((size >>= 1) != 0);
+	}
+}
+
+In this test, I am creating a very large, but very sparse, array.
+
+What should happen in this case is that only 4 pages get allocated (each
+page that I touched). The other pages should not be comitted to physical
+memory. If you write the test in C and use calloc, this is what happens. On
+MSFT the test case returns instantly, and there is no sign of excess memory
+comitting (at least, according to the task manager)
+
+However, Mono is committing the pages when they are allocated. This uses up
+nearly all the memory on the system.
+
+I am not sure if turnning on mmap support for libgc will fix this. It might
+still clear the pages with memset, which would render this useless.
+
+Even so, we probably want a behavior like libc malloc -- mmap large
+allocations only.