From 6e040d83ea743f56389ce5c69e58c9ae8da3017f Mon Sep 17 00:00:00 2001
From: Ingo Weinhold <ingo_weinhold@gmx.de>
Date: Sun, 11 Dec 2005 13:26:50 +0000
Subject: [PATCH] Added __swap_{float,double} for PPC. They are probably
 horribly suboptimal (if working at all) -- review is appreciated.

git-svn-id: file:///srv/svn/repos/haiku/haiku/trunk@15486 a95241bf-73f2-0310-859d-f6bbb57e9c96
---
 src/system/libroot/os/arch/ppc/byteorder.S | 61 +++++++++++++++++++++-
 1 file changed, 60 insertions(+), 1 deletion(-)

diff --git a/src/system/libroot/os/arch/ppc/byteorder.S b/src/system/libroot/os/arch/ppc/byteorder.S
index 7ad16f1d98..8f1b1536b8 100644
--- a/src/system/libroot/os/arch/ppc/byteorder.S
+++ b/src/system/libroot/os/arch/ppc/byteorder.S
@@ -1,5 +1,5 @@
 /*
-** Copyright 2003, Axel Dörfler, axeld@pinc-software.de. All rights reserved.
+** Copyright 2003, Axel D�fler, axeld@pinc-software.de. All rights reserved.
 ** Distributed under the terms of the OpenBeOS License.
 */
 
@@ -41,3 +41,62 @@ FUNCTION(__swap_int64):
 		mr		%r4, %r5				// copy lower 32 bits
 		blr
 
+
+/* TODO: The following functions can surely be optimized. A simple optimization
+ * would be to define macros with the contents of the __swap_int{32,64}
+ * functions and use those instead of calling the functions.
+ */
+
+/* float __swap_float(float value)
+ *					  f1
+ */
+FUNCTION(__swap_float):
+		// push a stack frame
+		stwu	%r1, -32(%r1)
+		mflr	%r0
+		stw		%r0, 36(%r1)
+
+		// %f1 -> %r3
+		stfs	%f1, 20(%r1)
+		lwz		%r3, 20(%r1)
+
+		// let __swap_int32 convert %r3
+		bl		__swap_int32
+
+		// %r3 -> %f1
+		stw		%r3, 20(%r1)
+		lfs		%f1, 20(%r1)
+		
+		// pop the stack frame
+		lwz		%r0, 36(%r1)
+		mtlr	%r0
+		addi	%r1, %r1, 32
+		blr
+
+/* double __swap_double(double value)
+ *						f1
+ */
+FUNCTION(__swap_double):
+		// push a stack frame
+		stwu	%r1, -32(%r1)
+		mflr	%r0
+		stw		%r0, 36(%r1)
+
+		// %f1 -> (%r3:%r4)
+		stfd	%f1, 20(%r1)
+		lwz		%r3, 20(%r1)
+		lwz		%r4, 24(%r1)
+
+		// let __swap_int64 convert %r3:%r4
+		bl		__swap_int64
+
+		// (%r3:%r4) -> %f1
+		stw		%r3, 20(%r1)
+		stw		%r4, 24(%r1)
+		lfd		%f1, 20(%r1)
+		
+		// pop the stack frame
+		lwz		%r0, 36(%r1)
+		mtlr	%r0
+		addi	%r1, %r1, 32
+		blr