memmove/bcopy from libkern, almost certainly the unrolling here is

beneficial because the compiler tends to do small known-size blocks
inline.  Continue using the MI memcpy.c for now.
ok miod

2 files changed, 162 insertions, 2 deletions

diff --git a/lib/libc/arch/hppa/string/Makefile.inc b/lib/libc/arch/hppa/string/Makefile.inc
index 68abbfc3226..7dc5311d52a 100644
--- a/lib/libc/arch/hppa/string/Makefile.inc
+++ b/lib/libc/arch/hppa/string/Makefile.inc
@@ -1,6 +1,6 @@
-#	$OpenBSD: Makefile.inc,v 1.8 2014/11/30 19:43:56 deraadt Exp $
+#	$OpenBSD: Makefile.inc,v 1.9 2014/12/05 01:30:44 deraadt Exp $
 
-SRCS+=	bcopy.c memcpy.c memmove.c \
+SRCS+=	memmove.S memcpy.c \
 	index.c rindex.c strchr.c strrchr.c \
 	bcmp.c bzero.c ffs.c memchr.c memcmp.c \
 	memset.c strcat.c strcmp.c strcpy.c strcspn.c strlen.c \
diff --git a/lib/libc/arch/hppa/string/memmove.S b/lib/libc/arch/hppa/string/memmove.S
new file mode 100644
index 00000000000..8ed210a168a
--- /dev/null
+++ b/lib/libc/arch/hppa/string/memmove.S
@@ -0,0 +1,160 @@
+/* This is a generated file. DO NOT EDIT. */
+/*
+ * Generated from:
+ *
+ *	OpenBSD: bcopy.m4
+ */
+/*
+ * Copyright (c) 1999 Michael Shalayeff
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR `AS IS' AND ANY EXPRESS OR
+ * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
+ * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
+ * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
+ * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
+ * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
+ * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+
+
+#undef _LOCORE
+#define _LOCORE
+#include <machine/asm.h>
+#include <machine/frame.h>
+
+
+LEAF_ENTRY(bcopy)
+	copy	arg0, ret0
+	copy	arg1, arg0
+	copy	ret0, arg1
+ALTENTRY(memmove)
+	comb,>,n arg0, arg1, $bcopy.reverse
+// ALTENTRY(memcpy)
+	copy	arg0, ret0
+	
+	comib,>=,n 15, arg2, $bcopy_f.byte
+
+	extru	arg1, 31, 2, t3
+	extru	arg0, 31, 2, t4
+	add	arg2, t4, arg2
+	comb,<> t3, t4, $bcopy_f.unaligned
+	dep	r0, 31, 2, arg1
+	
+	addi	-16, arg2, arg2
+$bcopy_f.loop16a
+	ldws,ma	4(sr0, arg1), t1
+	ldws,ma	4(sr0, arg1), t2
+	ldws,ma	4(sr0, arg1), t3
+	ldws,ma	4(sr0, arg1), t4
+	stbys,b,m t1, 4(sr0, arg0)
+	stws,ma	t2, 4(sr0, arg0)
+	stws,ma	t3, 4(sr0, arg0)
+	addib,>= -16, arg2, $bcopy_f.loop16a
+	stws,ma	t4, 4(sr0, arg0)
+
+	addib,<,n 12, arg2, $bcopy_f.cleanup
+$bcopy_f.word
+	ldws,ma	4(sr0, arg1), t1
+	addib,>= -4, arg2, $bcopy_f.word
+	stws,ma	t1, 4(sr0, arg0)
+
+$bcopy_f.cleanup
+	addib,=,n 4, arg2, $bcopy_f.done
+	ldws	0(sr0, arg1), t1
+	add	arg0, arg2, arg0
+	b	$bcopy_f.done
+	stbys,e	t1, 0(sr0, arg0)
+
+$bcopy_f.unaligned
+	sub,>=	t4, t3, t2
+	ldwm	4(sr0, arg1), ret1
+	zdep	t2, 28, 29, t1
+	mtsar	t1
+	
+	addi	-16, arg2, arg2
+$bcopy_f.loop16u
+	ldws,ma	4(sr0, arg1), t1
+	ldws,ma	4(sr0, arg1), t2
+	ldws,ma	4(sr0, arg1), t3
+	ldws,ma	4(sr0, arg1), t4
+	vshd	ret1, t1, r31
+	stbys,b,m r31, 4(sr0, arg0)
+	vshd	t1, t2, r31
+	stws,ma	r31, 4(sr0, arg0)
+	vshd	t2, t3, r31
+	stws,ma	r31, 4(sr0, arg0)
+	vshd	t3, t4, r31
+	stws,ma	r31, 4(sr0, arg0)
+	addib,>= -16, arg2, $bcopy_f.loop16u
+	copy	t4, ret1
+
+	addib,<,n 12, arg2, $bcopy_f.cleanup_un
+$bcopy_f.word_un
+	ldws,ma	4(sr0, arg1), t1
+	vshd	ret1, t1, t2
+	addib,<	-4, arg2, $bcopy_f.cleanup1_un
+	stws,ma	t2, 4(sr0, arg0)
+	ldws,ma	4(sr0, arg1), ret1
+	vshd	t1, ret1, t2
+	addib,>= -4, arg2, $bcopy_f.word_un
+	stws,ma	t2, 4(sr0, arg0)
+
+$bcopy_f.cleanup_un
+	addib,<=,n 4, arg2, $bcopy_f.done
+	mfctl	sar, t4
+	add	arg0, arg2, arg0
+	extru	t4, 28, 2, t4
+	sub,<=	arg2, t4, r0
+	ldws,ma	4(sr0, arg1), t1
+	vshd	ret1, t1, t2
+	b	$bcopy_f.done
+	stbys,e	t2, 0(sr0, arg0)
+
+$bcopy_f.cleanup1_un
+	b	$bcopy_f.cleanup_un
+	copy	t1, ret1
+
+$bcopy_f.byte
+	comb,>=,n r0, arg2, $bcopy_f.done
+$bcopy_f.byte_loop
+	ldbs,ma	1(sr0, arg1), t1
+	addib,<> -1, arg2, $bcopy_f.byte_loop
+	stbs,ma	t1, 1(sr0, arg0)
+$bcopy_f.done
+
+	bv	0(rp)
+	nop
+$bcopy.reverse
+	copy	arg0, ret0
+	add	arg1, arg2, arg1
+	add	arg0, arg2, arg0
+
+
+$bcopy_r.byte
+	comb,>=,n r0, arg2, $bcopy_r.done
+$bcopy_r.byte_loop
+	ldbs,mb	-1(sr0, arg1), t1
+	addib,<> -1, arg2, $bcopy_r.byte_loop
+	stbs,mb	t1, -1(sr0, arg0)
+$bcopy_r.done
+
+	bv	0(rp)
+	nop
+EXIT(bcopy)
+
+	.end