From 032bcffef610e94406c192e39d67b5231c2d7ab1 Mon Sep 17 00:00:00 2001
From: George Koehler <xkernigh@netscape.net>
Date: Mon, 23 Jan 2017 17:16:39 -0500
Subject: [PATCH] In PowerPC libem, use the new features of our assembler.

The new features are the hi16/lo16 and ha16/lo16 syntax for
relocations, and the extended mnemonics like "blr".

Use ha16/lo16 to load some double floats with 2 instructions (lis/lfd)
instead of 3 (lis/ori/lfd).

Use the extended names for branches, comparisons, and bit rotations,
so I can more easily read the code.  The new names often encode the
same machine instructions as the old names, except in a few places
where I changed the instructions.

Stop using andi. when we don't need to set cr0.  In inn.s, I change
andi. to extrwi to extract the same bits.  In los.s and sts.s, I
change "andi. r3, r3, ~3" to "clrrwi r3, r3, 2".  This avoids setting
cr0 and also stops clearing the high 16 bits of r3.

In csa.s, los.s, sts.s, I change some comparisons and right shifts
from signed to unsigned (cmplw, cmplwi, srwi), because the sizes are
unsigned.  In inn.s, the right shift can be signed (sraw) or unsigned
(srw), but I use srw because we don't need the carry bit.

In fef8.s, I save an instruction by using rlwinm instead of addis/andc
to rlwinm to clear a field.  The code no longer kills r7.  In both
fef8.s and fif8.s, I remove the list of killed registers.

Also remove some whitespace from ends of lines.
---
 mach/powerpc/libem/aar4.s        | 26 +++++++-----------
 mach/powerpc/libem/and.s         |  8 +++---
 mach/powerpc/libem/cfi8.s        |  9 +------
 mach/powerpc/libem/cfu8.s        | 31 +++++++++-------------
 mach/powerpc/libem/cif8.s        | 21 ++++++---------
 mach/powerpc/libem/cms.s         | 14 +++++-----
 mach/powerpc/libem/com.s         |  8 +++---
 mach/powerpc/libem/csa.s         | 29 ++++++--------------
 mach/powerpc/libem/csb.s         | 25 +++++++-----------
 mach/powerpc/libem/cuf8.s        | 23 ++++++----------
 mach/powerpc/libem/fd_00000000.s |  7 +----
 mach/powerpc/libem/fd_80000000.s |  7 +----
 mach/powerpc/libem/fd_FFFFFFFF.s |  9 ++-----
 mach/powerpc/libem/fef8.s        | 28 +++++++++-----------
 mach/powerpc/libem/fif8.s        | 28 +++++++++-----------
 mach/powerpc/libem/inn.s         | 12 ++++-----
 mach/powerpc/libem/ior.s         |  8 +++---
 mach/powerpc/libem/lar4.s        | 19 ++++++--------
 mach/powerpc/libem/los.s         | 45 ++++++++++++++------------------
 mach/powerpc/libem/rck.s         | 12 ++++-----
 mach/powerpc/libem/ret.s         | 17 ++++--------
 mach/powerpc/libem/sar4.s        | 19 ++++++--------
 mach/powerpc/libem/set.s         | 16 +++++-------
 mach/powerpc/libem/sts.s         | 45 ++++++++++++++------------------
 mach/powerpc/libem/xor.s         |  8 +++---
 mach/powerpc/libem/zer.s         | 10 +++----
 26 files changed, 184 insertions(+), 300 deletions(-)

diff --git a/mach/powerpc/libem/aar4.s b/mach/powerpc/libem/aar4.s
index 2c65af643..5e4155091 100644
--- a/mach/powerpc/libem/aar4.s
+++ b/mach/powerpc/libem/aar4.s
@@ -1,12 +1,5 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-
 .sect .text
-	
+
 ! Index into a bounds-checked array.
 !
 ! On entry:
@@ -20,19 +13,20 @@
 
 .define .aar4
 .aar4:
-	li32 r0, .trap_earray
+	lis r0, hi16[.trap_earray]
+	ori r0, r0, lo16[.trap_earray]
 	mtspr ctr, r0            ! load CTR with trap address
 
 	lwz r0, 0(r3)
 	subf. r4, r0, r4         ! adjust range
-	bcctr IFTRUE, LT, 0      ! check lower bound
-	
+	bltctr                   ! check lower bound
+
 	lwz r0, 4(r3)
-	cmpl cr0, 0, r4, r3
-	bcctr IFFALSE, LT, 0     ! check upper bound
-	
+	cmplw r4, r3
+	bgectr                   ! check upper bound
+
 	lwz r0, 8(r3)
 	mullw r4, r4, r0         ! scale index
 	add r3, r4, r5           ! calculate element address
-	
-	bclr ALWAYS, 0, 0
+
+	blr
diff --git a/mach/powerpc/libem/and.s b/mach/powerpc/libem/and.s
index 727d79ec0..cb4e1e54a 100644
--- a/mach/powerpc/libem/and.s
+++ b/mach/powerpc/libem/and.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Set intersection.
@@ -12,7 +10,7 @@
 
 	mr	r4, sp			! r4 = ptr to set a
 	add	r5, sp, r3		! r5 = ptr to set b
-	rlwinm	r6, r3, 30, 2, 31
+	srwi	r6, r3, 2
 	mtspr	ctr, r6			! ctr = r3 / 4
 1:
 	lwz	r7, 0(r4)
@@ -21,6 +19,6 @@
 	stw	r8, 0(r5)
 	addi	r4, r4, 4
 	addi	r5, r5, 4
-	bc	DNZ, 0, 1b		! loop ctr times
+	bdnz	1b			! loop ctr times
 	add	sp, sp, r3
-	bclr	ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/cfi8.s b/mach/powerpc/libem/cfi8.s
index 7142c694b..9a87e99ae 100644
--- a/mach/powerpc/libem/cfi8.s
+++ b/mach/powerpc/libem/cfi8.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-	
 .sect .text
 
 ! Converts a 64-bit double into a 32-bit integer.
@@ -17,4 +10,4 @@
 	fctiwz f0, f0
 	stfd f0, 0(sp)
 	addi sp, sp, 4
-	bclr ALWAYS, 0, 0        ! ...and return 
+	blr
diff --git a/mach/powerpc/libem/cfu8.s b/mach/powerpc/libem/cfu8.s
index 758df8572..915f84dd2 100644
--- a/mach/powerpc/libem/cfu8.s
+++ b/mach/powerpc/libem/cfu8.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-
 .sect .text
 
 ! Converts a 64-bit double into a 32-bit unsigned integer.
@@ -13,17 +6,17 @@
 
 .define .cfu8
 .cfu8:
-	li32 r3, .fd_00000000
-	lfd f0, 0(r3)            ! f0 = 0.0
-	
+	lis r3, ha16[.fd_00000000]
+	lfd f0, lo16[.fd_00000000](r3) ! f0 = 0.0
+
 	lfd f1, 0(sp)            ! value to be converted
 
-	li32 r3, .fd_FFFFFFFF
-	lfd f3, 0(r3)            ! f3 = 0xFFFFFFFF	
+	lis r3, ha16[.fd_FFFFFFFF]
+	lfd f3, lo16[.fd_FFFFFFFF](r3) ! f3 = 0xFFFFFFFF
+
+	lis r3, ha16[.fd_80000000]
+	lfd f4, lo16[.fd_80000000](r3) ! f4 = 0x80000000
 
-	li32 r3, .fd_80000000
-	lfd f4, 0(r3)            ! f4 = 0x80000000 	
-	
 	fsel f2, f1, f1, f0
 	fsub f5, f3, f1
 	fsel f2, f5, f2, f3
@@ -34,11 +27,11 @@
 	
 	stfd f2, 0(sp)
 	addi sp, sp, 4
-	
-	bclr IFTRUE, LT, 0
-	
+
+	bltlr
+
 	lwz r3, 0(sp)
 	xoris r3, r3, 0x8000
 	stw r3, 0(sp)
 
-	bclr ALWAYS, 0, 0 
+	blr
diff --git a/mach/powerpc/libem/cif8.s b/mach/powerpc/libem/cif8.s
index d2c82e54e..13abbcade 100644
--- a/mach/powerpc/libem/cif8.s
+++ b/mach/powerpc/libem/cif8.s
@@ -1,9 +1,4 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
+.sect .text; .sect .rom; .sect .data; .sect .bss
 
 .sect .text
 
@@ -14,22 +9,22 @@
 .define .cif8
 .cif8:
 	addi sp, sp, -4          ! make space for the double
-	
+
 	lwz r3, 4(sp)
 	xoris r3, r3, 0x8000
 	stw r3, 4(sp)            ! flip sign of integer value
-	
+
 	addis r3, r0, 0x4330
 	stw r3, 0(sp)            ! set high word to construct a double
-	
+
 	lfd f0, 0(sp)            ! load value
 
-	li32 r3, pivot
-	lfd f1, 0(r3)            ! load pivot value
+	lis r3, ha16[pivot]
+	lfd f1, lo16[pivot](r3)  ! load pivot value
 	fsub f0, f0, f1          ! adjust
-	                         
+
 	stfd f0, 0(sp)           ! save value again...
-	bclr ALWAYS, 0, 0        ! ...and return 
+	blr                      ! ...and return 
 
 .sect .rom
 pivot:
diff --git a/mach/powerpc/libem/cms.s b/mach/powerpc/libem/cms.s
index 53cb65691..30aaccd20 100644
--- a/mach/powerpc/libem/cms.s
+++ b/mach/powerpc/libem/cms.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Compare sets a, b.
@@ -12,21 +10,21 @@
 	mr	r4, sp			! r4 = ptr to set a
 	add	r5, sp, r3		! r5 = ptr to set b
 	mr	r6, r3			! r6 = size
-	rlwinm	r3, r3, 30, 2, 31
+	srwi	r3, r3, 2
 	mtspr	ctr, r3			! ctr = size / 4
 1:
 	lwz	r7, 0(r4)
 	lwz	r8, 0(r5)
-	cmp	cr0, 0, r7, r8		! compare words in sets
+	cmpw	cr0, r7, r8		! compare words in sets
 	addi	r4, r4, 4
 	addi	r5, r5, 4
-	bc	IFFALSE, EQ, 2f		! branch if not equal
-	bc	DNZ, 0, 1b		! loop ctr times
+	bne	cr0, 2f			! branch if not equal
+	bdnz	1b			! loop ctr times
 	addi	r3, r0, 0		! equal: return 0
 	b	3f
 2:
 	addi	r3, r0, 1		! not equal: return 1
 3:
-	rlwinm	r6, r6, 1, 0, 30	! r6 = size * 2
+	slwi	r6, r6, 1		! r6 = size * 2
 	add	sp, sp, r6		! remove sets from stack
-	bclr	ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/com.s b/mach/powerpc/libem/com.s
index 084eeeb62..3168cfe17 100644
--- a/mach/powerpc/libem/com.s
+++ b/mach/powerpc/libem/com.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Set complement.
@@ -11,12 +9,12 @@
 	addi sp, sp, 4
 
 	mr	r4, sp			! r4 = pointer to set a
-	rlwinm	r5, r3, 30, 2, 31
+	srwi	r5, r3, 2
 	mtspr	ctr, r5			! ctr = r3 / 4
 1:
 	lwz	r6, 0(r4)
 	nor	r6, r6, r6		! complement of word
 	stw	r6, 0(r4)
 	addi	r4, r4, 4
-	bc	DNZ, 0, 1b		! loop ctr times
-	bclr	ALWAYS, 0, 0
+	bdnz	1b			! loop ctr times
+	blr
diff --git a/mach/powerpc/libem/csa.s b/mach/powerpc/libem/csa.s
index 88e6e176a..3898241c4 100644
--- a/mach/powerpc/libem/csa.s
+++ b/mach/powerpc/libem/csa.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-	
 .sect .text
 
 ! this is not a subroutine, but just a
@@ -25,23 +18,17 @@
 	
 	lwz r5, 4(r3)            ! fetch lower bound
 	subf. r4, r5, r4         ! adjust value
-	bcctr IFTRUE, LT, 0      ! jump to default if out of range
-	
+	bltctr                   ! jump to default if out of range
+
 	lwz r5, 8(r3)            ! fetch range
-	cmp cr0, 0, r4, r5
-	bcctr IFTRUE, GT, 0      ! jump to default if out of range
-	
+	cmplw r4, r5
+	bgtctr                   ! jump to default if out of range
+
 	addi r3, r3, 12          ! skip header
-	rlwinm r4, r4, 2, 0, 31-2 ! scale value (<<2)
-	b 1f
-1:
+	slwi r4, r4, 2           ! scale value (<<2)
 	lwzx r5, r3, r4          ! load target
-	b 1f
-1:
 	mtspr ctr, r5
-	 
+
 	or. r5, r5, r5           ! test it
-	b 1f
-1:
-	bcctr IFFALSE, EQ, 0     ! jump to target if non-zero
+	bnectr                   ! jump to target if non-zero
 	b .trap_ecase            ! otherwise trap
diff --git a/mach/powerpc/libem/csb.s b/mach/powerpc/libem/csb.s
index a8df85d7f..571bfc210 100644
--- a/mach/powerpc/libem/csb.s
+++ b/mach/powerpc/libem/csb.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-	
 .sect .text
 
 ! this is not a subroutine, but just a
@@ -22,21 +15,21 @@
 
 	lwz r5, 0(r3)            ! load default
 	mtspr ctr, r5
-	
+
 	lwz r6, 4(r3)            ! fetch count
-	
+
 1:
 	or. r6, r6, r6           ! test count
-	bcctr IFTRUE, EQ, 0      ! exit if zero
+	beqctr                   ! exit if zero
 	addi r6, r6, -1          ! otherwise decrement
-	
+
 	lwzu r7, 8(r3)           ! fetch target index, increment pointer
-	cmp cr0, 0, r4, r7       ! compare with value
-	bc IFFALSE, EQ, 1b       ! if not equal, go again
-	
+	cmpw r4, r7              ! compare with value
+	bne 1b                   ! if not equal, go again
+
 	lwz r7, 4(r3)            ! fetch target address
 	mtspr ctr, r7
-	
+
 	or. r7, r7, r7           ! test it
-	bcctr IFFALSE, EQ, 0     ! jump to target if non-zero
+	bnectr                   ! jump to target if non-zero
 	b .trap_ecase            ! otherwise trap
diff --git a/mach/powerpc/libem/cuf8.s b/mach/powerpc/libem/cuf8.s
index 5d5a12988..ce9932aa1 100644
--- a/mach/powerpc/libem/cuf8.s
+++ b/mach/powerpc/libem/cuf8.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-	
 .sect .text
 
 ! Converts a 32-bit unsigned integer into a 64-bit double.
@@ -14,18 +7,18 @@
 .define .cuf8
 .cuf8:
 	addi sp, sp, -4          ! make space for the double
-	
-	addis r3, r0, 0x4330
+
+	lis r3, 0x4330
 	stw r3, 0(sp)            ! set high word to construct a double
-	
+
 	lfd f0, 0(sp)            ! load value
-	
-	li32 r3, pivot
-	lfd f1, 0(r3)            ! load pivot value
+
+	lis r3, ha16[pivot]
+	lfd f1, lo16[pivot](r3)  ! load pivot value
 	fsub f0, f0, f1          ! adjust
-	                         
+
 	stfd f0, 0(sp)           ! save value again...
-	bclr ALWAYS, 0, 0        ! ...and return 
+	blr                      ! ...and return
 
 .sect .rom
 pivot:
diff --git a/mach/powerpc/libem/fd_00000000.s b/mach/powerpc/libem/fd_00000000.s
index cefa91b8e..8ffe44a24 100644
--- a/mach/powerpc/libem/fd_00000000.s
+++ b/mach/powerpc/libem/fd_00000000.s
@@ -1,10 +1,5 @@
-#
-! $Source$
-! $State$
-! $Revision$
+.sect .text; .sect .rom; .sect .data; .sect .bss
 
-#include "powerpc.h"
-	
 .sect .rom
 
 ! Contains a handy double-precision zero. (Also works as a single-precision
diff --git a/mach/powerpc/libem/fd_80000000.s b/mach/powerpc/libem/fd_80000000.s
index 50eacd586..5c153bba8 100644
--- a/mach/powerpc/libem/fd_80000000.s
+++ b/mach/powerpc/libem/fd_80000000.s
@@ -1,10 +1,5 @@
-#
-! $Source$
-! $State$
-! $Revision$
+.sect .text; .sect .rom; .sect .data; .sect .bss
 
-#include "powerpc.h"
-	
 .sect .rom
 
 ! Contains a handy double-precision 0x80000000.
diff --git a/mach/powerpc/libem/fd_FFFFFFFF.s b/mach/powerpc/libem/fd_FFFFFFFF.s
index 9218f2726..88cf04bd9 100644
--- a/mach/powerpc/libem/fd_FFFFFFFF.s
+++ b/mach/powerpc/libem/fd_FFFFFFFF.s
@@ -1,15 +1,10 @@
-#
-! $Source$
-! $State$
-! $Revision$
+.sect .text; .sect .rom; .sect .data; .sect .bss
 
-#include "powerpc.h"
-	
 .sect .rom
 
 ! Contains a handy double-precision 0xFFFFFFFF.
 
 .define .fd_FFFFFFFF
-.fd_FFFFFFFF:	
+.fd_FFFFFFFF:
 	!float 4.294967295e+9 sz 8
 	.data1 0101,0357,0377,0377,0377,0340,00,00
diff --git a/mach/powerpc/libem/fef8.s b/mach/powerpc/libem/fef8.s
index fc72b04f2..f71ab8e38 100644
--- a/mach/powerpc/libem/fef8.s
+++ b/mach/powerpc/libem/fef8.s
@@ -1,4 +1,4 @@
-#include "powerpc.h"
+.sect .text; .sect .rom; .sect .data; .sect .bss
 
 .sect .text
 
@@ -10,46 +10,44 @@
 !  r3 = fraction, high word (bits 0..31)
 !  r4 = fraction, low word (bits 32..63)
 !  r5 = exponent
-! Kills: cr0 f0 f1 r6 r7
 
 .define .fef8
 .fef8:
 	! IEEE double-precision format:
 	!   sign  exponent  fraction
 	!   0     1..11     12..63
-	rlwinm r6, r3, 12, 21, 31	! r6 = IEEE exponent
-	addis r7, r0, 0x7ff0		! r7 = exponent mask
+	extrwi r6, r3, 11, 1		! r6 = IEEE exponent
 	addi r5, r6, -1022		! r5 = true exponent
-	cmpi cr0, 0, r6, 2047
-	bclr IFTRUE, EQ, 0		! return if infinity or NaN
-	cmpi cr0, 0, r6, 0
-	bc IFFALSE, EQ, 1f		! jump if normalized number
+	cmpwi r6, 2047
+	beqlr				! return if infinity or NaN
+	cmpwi r6, 0
+	bne 1f				! jump if normalized number
 
 	! Got denormalized number or zero, probably zero.
-	rlwinm r6, r3, 0, 12, 31
+	extrwi r6, r3, 22, 12
 	addi r5, r0, 0			! r5 = true exponent = 0
 	or. r6, r6, r4			! r6 = high|low fraction
-	bclr IFTRUE, EQ, 0		! return if zero
+	beqlr				! return if zero
 
 	! Got denormalized number, not zero.
 	stwu r4, -4(sp)
 	stwu r3, -4(sp)
-	li32 r6, _2_64
 	lfd f0, 0(sp)
-	lfd f1, 0(r6)
+	lis r6, ha16[_2_64]
+	lfd f1, lo16[_2_64](r6)
 	fmul f0, f0, f1			! multiply it by 2**64
 	stfd f0, 0(sp)
 	lwz r3, 0(sp)
 	lwz r4, 4(sp)
-	rlwinm r6, r3, 12, 21, 31	! r6 = IEEE exponent
+	extrwi r6, r3, 11, 1		! r6 = IEEE exponent
 	addi sp, sp, 8
 	addi r5, r6, -1022 - 64		! r5 = true exponent
 1:
 	! Put fraction in [0.5, 1) or (-1, -0.5] by setting its
 	! exponent to true 0, IEEE 1022.
-	andc r3, r3, r7			! clear old exponent
+	rlwinm r3, r3, 0, 12, 0		! clear old exponent
 	oris r3, r3, 1022 << 4		! set new exponent
-	bclr ALWAYS, 0, 0
+	blr
 
 .sect .rom
 _2_64:
diff --git a/mach/powerpc/libem/fif8.s b/mach/powerpc/libem/fif8.s
index a26c77830..eda9b04f2 100644
--- a/mach/powerpc/libem/fif8.s
+++ b/mach/powerpc/libem/fif8.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Multiplies two double-precision floats, then splits the product into
@@ -9,7 +7,6 @@
 ! Yields:
 !  f1 = fraction
 !  f2 = integer
-! Kills: cr0 f1 f2 r3 r4 r5 r6
 
 .define .fif8
 .fif8:
@@ -25,17 +22,16 @@
 	! 0 to 51, then the IEEE fraction has that many integer bits.
 	! (IEEE has an implicit 1 before its fraction.  If the IEEE
 	! fraction has 0 integer bits, we still have an integer.)
-	rlwinm r5, r3, 12, 21, 31	! r5 = IEEE exponent
+	extrwi r5, r3, 11, 1		! r5 = IEEE exponent
 	addic. r5, r5, -1023		! r5 = nr of integer bits
-	bc IFTRUE, LT, no_int
-	cmpi cr0, 0, r5, 21
-	bc IFTRUE, LT, small_int
-	cmpi cr0, 0, r5, 52
-	bc IFTRUE, LT, big_int
+	blt no_int
+	cmpwi r5, 21
+	blt small_int
+	cmpwi r5, 52
+	blt big_int
 
-	! f1 is an integer without fraction.  Jump to calculate
-	! fraction f1 = f2 - f1.  It will be zero (or perhaps NaN).
-	fmr f2, f1
+	! f1 is an integer without fraction (or infinity or NaN).
+	fmr f2, f1			! integer = f1
 	b subtract
 
 no_int:
@@ -46,17 +42,17 @@ no_int:
 small_int:
 	! f1 has r5 = 0 to 20 integer bits in the IEEE fraction.
 	! High word has 20 - r5 fraction bits.
-	addi r6, r0, 20
+	li r6, 20
 	subf r6, r5, r6
 	srw r3, r3, r6
-	addi r4, r0, 0			! clear low word
+	li r4, 0			! clear low word
 	slw r3, r3, r6			! clear fraction in high word
 	b move_int
 
 big_int:
 	! f1 has r5 = 21 to 51 to integer bits.
 	! Low word has 52 - r5 fraction bits.
-	addi r6, r0, 52
+	li r6, 52
 	subf r6, r5, r6
 	srw r4, r4, r6
 	slw r4, r4, r6			! clear fraction in low word
@@ -68,4 +64,4 @@ subtract:
 	fsub f1, f1, f2			! fraction = value - integer
 done:
 	addi sp, sp, 8			! restore stack pointer
-	bclr ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/inn.s b/mach/powerpc/libem/inn.s
index 9770ac094..8925e776e 100644
--- a/mach/powerpc/libem/inn.s
+++ b/mach/powerpc/libem/inn.s
@@ -1,4 +1,4 @@
-#include "powerpc.h"
+#
 
 .sect .text
 
@@ -14,13 +14,13 @@
     addi r5, sp, 8      /* r5 = base address of bit set */
 
     rlwinm r6, r4, 29, 3, 29 /* r6 = byte index of word in set */
-    andi. r7, r4, 31    /* r7 = bit within word */
+    extrwi r7, r4, 5, 27     /* r7 = bit number within word */
 
-    lwzx r8, r5, r6     /* r8 = individual byte from set */
-    sraw r8, r8, r7
-    rlwinm r8, r8, 0, 31, 31
+    lwzx r8, r5, r6     /* r8 = individual word from set */
+    srw r8, r8, r7
+    extrwi r8, r8, 1, 31
 
     addi sp, sp, 8      /* retract over the two words */
     add sp, sp, r3      /* retract over bitfield */
     stwu r8, -4(sp)     /* push result */
-    bclr ALWAYS, 0, 0   /* return */
+    blr                 /* return */
diff --git a/mach/powerpc/libem/ior.s b/mach/powerpc/libem/ior.s
index 363799e1d..e6cd1844e 100644
--- a/mach/powerpc/libem/ior.s
+++ b/mach/powerpc/libem/ior.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Set union.
@@ -12,7 +10,7 @@
 
 	mr	r4, sp			! r4 = ptr to set a
 	add	r5, sp, r3		! r5 = ptr to set b
-	rlwinm	r6, r3, 30, 2, 31
+	srwi	r6, r3, 2
 	mtspr	ctr, r6			! ctr = r3 / 4
 1:
 	lwz	r7, 0(r4)
@@ -21,6 +19,6 @@
 	stw	r8, 0(r5)
 	addi	r4, r4, 4
 	addi	r5, r5, 4
-	bc	DNZ, 0, 1b		! loop ctr times
+	bdnz	1b			! loop ctr times
 	add	sp, sp, r3
-	bclr	ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/lar4.s b/mach/powerpc/libem/lar4.s
index 6375979c4..2f5c3683c 100644
--- a/mach/powerpc/libem/lar4.s
+++ b/mach/powerpc/libem/lar4.s
@@ -1,6 +1,3 @@
-#
-#include "powerpc.h"
-
 .sect .text
 
 ! Load from bounds-checked array.
@@ -18,19 +15,19 @@
 	! r3 = ptr to element
 	! r0 = size of element
 
-	cmpi cr0, 0, r0, 1
-	bc IFFALSE, EQ, 1f
+	cmpwi r0, 1
+	bne 1f
 	! Load 1 byte.
 	lbz r4, 0(r3)
 	stwu r4, -4(sp)
-	bclr ALWAYS, 0, 0
+	blr
 1:
-	cmpi cr0, 0, r0, 2
-	bc IFFALSE, EQ, 2f
+	cmpwi r0, 2
+	bne 2f
 	! Load 2 bytes.
 	lhz r4, 0(r3)
 	stwu r4, -4(sp)
-	bclr ALWAYS, 0, 0
+	blr
 2:
 	! Load r0 bytes, where r0 must be a positive multiple of 4.
 	subf sp, r0, sp		! move stack pointer down
@@ -39,5 +36,5 @@
 	addic. r5, r5, -4	! r5 -= 4
 	lwzx r4, r5, r3
 	stwx r4, r5, sp
-	bc IFTRUE, GT, 3b	! loop if r5 > 0
-	bclr ALWAYS, 0, 0
+	bgt 3b			! loop if r5 > 0
+	blr
diff --git a/mach/powerpc/libem/los.s b/mach/powerpc/libem/los.s
index f867fe770..2d412bce8 100644
--- a/mach/powerpc/libem/los.s
+++ b/mach/powerpc/libem/los.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-	
 .sect .text
 
 ! Loads a variable-sized structure onto the stack.
@@ -15,32 +8,32 @@
 .define .los
 .los:
 	! These sizes are handled specially.
-	
-	cmpi cr0, 0, r3, 1
-	bc IFFALSE, GT, size1
-	
-	cmpi cr0, 0, r3, 2
-	bc IFFALSE, GT, size2
-	
-	cmpi cr0, 0, r3, 4
-	bc IFFALSE, GT, size4
-	
+
+	cmplwi r3, 1
+	ble size1
+
+	cmplwi r3, 2
+	ble size2
+
+	cmplwi r3, 4
+	ble size4
+
 	! Variable-sized structure.
-	
+
 	addi r3, r3, 3
-	andi. r3, r3, ~3         ! align size
-	
+	clrrwi r3, r3, 2         ! align size
+
 	add r4, r4, r3           ! adjust address to top of block
 
-	srawi r3, r3, 2          ! convert size to the number of words
+	srwi r3, r3, 2           ! convert size to the number of words
 	mtspr ctr, r3
-	
+
 1:
 	lwzu r5, -4(r4)
 	stwu r5, -4(sp)
-	bc DNZ, 0, 1b            ! decrement CTR, jump if non-zero
-	bclr ALWAYS, 0, 0
-	
+	bdnz 1b                  ! decrement CTR, jump if non-zero
+	blr
+
 size1:
 	lbz r3, 0(r4)
 	b 1f
@@ -51,4 +44,4 @@ size4:
 	lwz r3, 0(r4)
 1:
 	stwu r3, -4(sp)
-	bclr ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/rck.s b/mach/powerpc/libem/rck.s
index 0d5717f16..9008be610 100644
--- a/mach/powerpc/libem/rck.s
+++ b/mach/powerpc/libem/rck.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Bounds check. Traps if the value is out of range.
@@ -12,11 +10,11 @@
     addi sp, sp, 4 ! leave value on stack
 
     lwz r5, 0 (r3)
-    cmp cr0, 0, r4, r5
-    bc IFTRUE, LT, .trap_erange
+    cmpw r4, r5
+    blt .trap_erange
 
     lwz r5, 4 (r3)
-    cmp cr0, 0, r4, r5
-    bc IFTRUE, GT, .trap_erange
+    cmpw r4, r5
+    bgt .trap_erange
 
-	bclr ALWAYS, 0, 0
+    blr
diff --git a/mach/powerpc/libem/ret.s b/mach/powerpc/libem/ret.s
index cca79ae86..c498af240 100644
--- a/mach/powerpc/libem/ret.s
+++ b/mach/powerpc/libem/ret.s
@@ -1,19 +1,12 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-
 .sect .text
-	
+
 ! Standard boilerplate for returning from functions.
 
 .define .ret
 .ret:
 	lwz r0, 4(fp)
 	mtspr lr, r0
-	lwz r0, 0(fp)            ! our stack frame becomes invalid as soon as...
-	addi sp, fp, 8           ! ...we change sp
-	or fp, r0, r0
-	bclr ALWAYS, 0, 0
+	lwz r0, 0(fp)   ! our stack frame becomes invalid as soon as...
+	addi sp, fp, 8  ! ...we change sp
+	mr fp, r0
+	blr
diff --git a/mach/powerpc/libem/sar4.s b/mach/powerpc/libem/sar4.s
index 0c1368af1..7c9778958 100644
--- a/mach/powerpc/libem/sar4.s
+++ b/mach/powerpc/libem/sar4.s
@@ -1,6 +1,3 @@
-#
-#include "powerpc.h"
-
 .sect .text
 
 ! Store to bounds-checked array.
@@ -18,21 +15,21 @@
 	! r3 = ptr to element
 	! r0 = size of element
 
-	cmpi cr0, 0, r0, 1
-	bc IFFALSE, EQ, 1f
+	cmpwi r0, 1
+	bne 1f
 	! Store 1 byte.
 	lwz r4, 0(sp)
 	addi sp, sp, 4
 	stb r4, 0(r3)
-	bclr ALWAYS, 0, 0
+	blr
 1:
-	cmpi cr0, 0, r0, 2
-	bc IFFALSE, EQ, 2f
+	cmpwi r0, 2
+	bne 2f
 	! Store 2 bytes.
 	lwz r4, 0(sp)
 	addi sp, sp, 4
 	sth r4, 0(r3)
-	bclr ALWAYS, 0, 0
+	blr
 2:
 	! Store r0 bytes, where r0 must be a positive multiple of 4.
 	or r5, r0, r0		! index r5 = length r0
@@ -40,6 +37,6 @@
 	addic. r5, r5, -4	! r5 -= 4
 	lwzx r4, r5, sp
 	stwx r4, r5, r3
-	bc IFTRUE, GT, 3b	! loop if r5 > 0
+	bgt 3b			! loop if r5 > 0
 	add sp, r0, sp		! move stack pointer up
-	bclr ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/set.s b/mach/powerpc/libem/set.s
index b42881cd7..3c4a9e579 100644
--- a/mach/powerpc/libem/set.s
+++ b/mach/powerpc/libem/set.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Create singleton set.
@@ -11,22 +9,22 @@
 	lwz     r4, 4 (sp)
 	addi    sp, sp, 8
 
-	rlwinm	r7, r3, 30, 2, 31
+	srwi	r7, r3, 2
 	neg	r5, r3
 	add	sp, sp, r5		! allocate set
 	mr	r6, sp			! r6 = ptr to set
 	mtspr	ctr, r7			! ctr = r3 / 4
 1:
-	rlwinm.	r7, r4, 0, 0, 26	! r7 = r4 & ~31
-	bc	IFTRUE, EQ, 2f		! branch if r4 in 0..31
-	addi	r5, r0, 0		! no bit, word is zero
+	clrrwi.	r7, r4, 5		! r7 = r4 & ~31
+	beq	2f			! branch if r4 in 0..31
+	li	r5, 0			! no bit, word is zero
 	b	3f
 2:
-	addi	r5, r0, 1
+	li	r5, 1
 	slw	r5, r5, r4		! yes bit, set bit in word
 3:
 	stw	r5, 0(r6)		! store word in set
 	addi	r4, r4, -32
 	addi	r6, r6, 4
-	bc	DNZ, 0, 1b		! loop ctr times
-	bclr	ALWAYS, 0, 0
+	bdnz	1b			! loop ctr times
+	blr
diff --git a/mach/powerpc/libem/sts.s b/mach/powerpc/libem/sts.s
index 2f8022ad9..411b0fb66 100644
--- a/mach/powerpc/libem/sts.s
+++ b/mach/powerpc/libem/sts.s
@@ -1,10 +1,3 @@
-#
-! $Source$
-! $State$
-! $Revision$
-
-#include "powerpc.h"
-	
 .sect .text
 
 ! Stores a variable-sized structure from the stack.
@@ -15,35 +8,35 @@
 .define .sts
 .sts:
 	! These sizes are handled specially.
-	
+
 	lwz r5, 0(sp)
 
-	cmpi cr0, 0, r3, 1
-	bc IFFALSE, GT, size1
-	
-	cmpi cr0, 0, r3, 2
-	bc IFFALSE, GT, size2
-	
-	cmpi cr0, 0, r3, 4
-	bc IFFALSE, GT, size4
-	
+	cmplwi r3, 1
+	ble size1
+
+	cmplwi r3, 2
+	ble size2
+
+	cmplwi r3, 4
+	ble size4
+
 	! Variable-sized structure.
-	
+
 	addi r3, r3, 3
-	andi. r3, r3, ~3         ! align size
-	
-	srawi r3, r3, 2          ! convert size to the number of words
+	clrrwi r3, r3, 2         ! align size
+
+	srwi r3, r3, 2           ! convert size to the number of words
 	mtspr ctr, r3
-	
+
 1:
 	lwz r5, 0(sp)
 	addi sp, sp, 4
 	stw r5, 0(r4)
 	addi r4, r4, 4
 
-	bc DNZ, 0, 1b            ! decrement CTR, jump if non-zero
-	bclr ALWAYS, 0, 0
-	
+	bdnz 1b                  ! decrement CTR, jump if non-zero
+	blr
+
 size1:
 	stb r5, 0(r4)
 	b 1f
@@ -54,4 +47,4 @@ size4:
 	stw r5, 0(r4)
 1:
 	addi sp, sp, 4
-	bclr ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/xor.s b/mach/powerpc/libem/xor.s
index 9d4bc76b9..acb02a032 100644
--- a/mach/powerpc/libem/xor.s
+++ b/mach/powerpc/libem/xor.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Set symmetric difference.
@@ -10,7 +8,7 @@
 .xor:
 	mr	r4, sp			! r4 = ptr to set a
 	add	r5, sp, r3		! r5 = ptr to set b
-	rlwinm	r6, r3, 30, 2, 31
+	srwi	r6, r3, 2
 	mtspr	ctr, r6			! ctr = r3 / 4
 1:
 	lwz	r7, 0(r4)
@@ -19,6 +17,6 @@
 	stw	r8, 0(r5)
 	addi	r4, r4, 4
 	addi	r5, r5, 4
-	bc	DNZ, 0, 1b		! loop ctr times
+	bdnz	1b			! loop ctr times
 	add	sp, sp, r3
-	bclr	ALWAYS, 0, 0
+	blr
diff --git a/mach/powerpc/libem/zer.s b/mach/powerpc/libem/zer.s
index 697a5715f..a47a150cc 100644
--- a/mach/powerpc/libem/zer.s
+++ b/mach/powerpc/libem/zer.s
@@ -1,5 +1,3 @@
-#include "powerpc.h"
-
 .sect .text
 
 ! Create empty set.
@@ -10,8 +8,8 @@
 	lwz     r3, 0(sp)
 	addi    sp, sp, 4
 
-	rlwinm	r7, r3, 30, 2, 31
-	addi	r4, r0, 0		! r4 = zero
+	srwi	r7, r3, 2
+	li	r4, 0			! r4 = zero
 	neg	r5, r3
 	add	sp, sp, r5		! allocate set
 	mr	r6, sp			! r6 = ptr to set
@@ -19,5 +17,5 @@
 1:
 	stw	r4, 0(r6)		! store zero in set
 	addi	r6, r6, 4
-	bc	DNZ, 0, 1b		! loop ctr times
-	bclr	ALWAYS, 0, 0
+	bdnz	1b			! loop ctr times
+	blr