From 4fcc031a7085a47b9a027a20a919574f8aab0768 Mon Sep 17 00:00:00 2001
From: Mike Frysinger <vapier@gentoo.org>
Date: Tue, 30 May 2006 09:13:53 +0000
Subject: import some optimized functions from blackfin cvs

---
 libc/string/bfin/memcmp.S | 101 ++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 101 insertions(+)
 create mode 100644 libc/string/bfin/memcmp.S

(limited to 'libc/string/bfin/memcmp.S')

diff --git a/libc/string/bfin/memcmp.S b/libc/string/bfin/memcmp.S
new file mode 100644
index 000000000..f2679d5ae
--- /dev/null
+++ b/libc/string/bfin/memcmp.S
@@ -0,0 +1,101 @@
+/* memcmp.S
+ * Copyright (C) 2003, 2005, 2006 Analog Devices Inc., All Rights Reserved.
+ *
+ * This file is subject to the terms and conditions of the GNU Library General
+ * Public License. See the file "COPYING.LIB" in the main directory of this
+ * archive for more details.
+ *
+ * Non-LGPL License also available as part of VisualDSP++
+ * http://www.analog.com/processors/resources/crosscore/visualDspDevSoftware.html
+ */
+
+/* int memcmp(const void *s1, const void *s2, size_t n);
+ * R0 = First Address (s1)
+ * R1 = Second Address (s2)
+ * R2 = count (n)
+ *
+ * Favours word aligned data.
+ */
+
+.text
+
+.align 2
+
+.global _memcmp
+.type _memcmp, STT_FUNC
+_memcmp:
+	I1 = P3;
+	P0 = R0;              // P0 = s1 address
+	P3 = R1;              // P3 = s2 Address
+	P2 = R2 ;             // P2 = count
+	CC = R2 <= 7(IU);
+	IF CC JUMP  too_small;
+	I0 = R1;		    // s2
+	R1 = R1 | R0;         // OR addresses together
+	R1 <<= 30;            // check bottom two bits
+	CC =  AZ;             // AZ set if zero.
+	IF !CC JUMP  bytes ;  // Jump if addrs not aligned.
+
+	P1 = P2 >> 2;          // count = n/4
+	R3 =  3;
+	R2 = R2 & R3;         // remainder
+	P2 = R2;               // set remainder
+
+	LSETUP (quad_loop_s , quad_loop_e) LC0=P1;
+quad_loop_s:
+#if !defined(__WORKAROUND_AVOID_DAG1)
+	MNOP || R0 = [P0++] || R1 = [I0++];
+#else
+	R0 = [P0++];
+	R1 = [I0++];
+#endif
+	CC = R0 == R1;
+	IF !CC JUMP quad_different;
+quad_loop_e:
+	NOP;
+
+	P3 = I0;                 // s2
+too_small:
+	CC = P2 == 0;            //Check zero count
+	IF CC JUMP finished;     // very unlikely
+
+bytes:
+	LSETUP (byte_loop_s , byte_loop_e) LC0=P2;
+byte_loop_s:
+	R1 = B[P3++](Z);	// *s2
+	R0 = B[P0++](Z);	// *s1
+	CC = R0 == R1;
+	IF !CC JUMP different;
+byte_loop_e:
+	NOP;
+
+different:
+	R0 = R0 - R1;
+	P3 = I1;
+	RTS;
+
+quad_different:
+	// We've read two quads which don't match.
+	// Can't just compare them, because we're
+	// a little-endian machine, so the MSBs of
+	// the regs occur at later addresses in the
+	// string.
+	// Arrange to re-read those two quads again,
+	// byte-by-byte.
+	P0 += -4;	// back up to the start of the
+	P3 = I0;	// quads, and increase the
+	P2 += 4;	// remainder count
+	P3 += -4;
+	JUMP bytes;
+
+finished:
+	R0 = 0;
+	P3 = I1;
+	RTS;
+.size _memcmp,.-_memcmp
+
+libc_hidden_def (memcmp)
+
+#ifdef __UCLIBC_SUSV3_LEGACY__
+strong_alias (memcmp,bcmp)
+#endif
-- 
cgit v1.2.3