s390/xor: optimized xor routing using the XC instruction

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>

3 files changed, 155 insertions, 2 deletions

diff --git a/arch/s390/include/asm/xor.h b/arch/s390/include/asm/xor.h
index c82eb12a5b18..c988df744a70 100644
--- a/arch/s390/include/asm/xor.h
+++ b/arch/s390/include/asm/xor.h
@@ -1 +1,20 @@
-#include <asm-generic/xor.h>
+/*
+ * Optimited xor routines
+ *
+ * Copyright IBM Corp. 2016
+ * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>
+ */
+#ifndef _ASM_S390_XOR_H
+#define _ASM_S390_XOR_H
+
+extern struct xor_block_template xor_block_xc;
+
+#undef XOR_TRY_TEMPLATES
+#define XOR_TRY_TEMPLATES				\
+do {							\
+	xor_speed(&xor_block_xc);			\
+} while (0)
+
+#define XOR_SELECT_TEMPLATE(FASTEST)	(&xor_block_xc)
+
+#endif /* _ASM_S390_XOR_H */
diff --git a/arch/s390/lib/Makefile b/arch/s390/lib/Makefile
index 0e8fefe5b0ce..1d1af31e8354 100644
--- a/arch/s390/lib/Makefile
+++ b/arch/s390/lib/Makefile
@@ -3,7 +3,7 @@
 #
 
 lib-y += delay.o string.o uaccess.o find.o
-obj-y += mem.o
+obj-y += mem.o xor.o
 lib-$(CONFIG_SMP) += spinlock.o
 lib-$(CONFIG_KPROBES) += probes.o
 lib-$(CONFIG_UPROBES) += probes.o
diff --git a/arch/s390/lib/xor.c b/arch/s390/lib/xor.c
new file mode 100644
index 000000000000..7d94e3ec34a9
--- /dev/null
+++ b/arch/s390/lib/xor.c
@@ -0,0 +1,134 @@
+/*
+ * Optimized xor_block operation for RAID4/5
+ *
+ * Copyright IBM Corp. 2016
+ * Author(s): Martin Schwidefsky <schwidefsky@de.ibm.com>
+ */
+
+#include <linux/types.h>
+#include <linux/module.h>
+#include <linux/raid/xor.h>
+
+static void xor_xc_2(unsigned long bytes, unsigned long *p1, unsigned long *p2)
+{
+	asm volatile(
+		"	larl	1,2f\n"
+		"	aghi	%0,-1\n"
+		"	jm	3f\n"
+		"	srlg	0,%0,8\n"
+		"	ltgr	0,0\n"
+		"	jz	1f\n"
+		"0:	xc	0(256,%1),0(%2)\n"
+		"	la	%1,256(%1)\n"
+		"	la	%2,256(%2)\n"
+		"	brctg	0,0b\n"
+		"1:	ex	%0,0(1)\n"
+		"	j	3f\n"
+		"2:	xc	0(1,%1),0(%2)\n"
+		"3:\n"
+		: : "d" (bytes), "a" (p1), "a" (p2)
+		: "0", "1", "cc", "memory");
+}
+
+static void xor_xc_3(unsigned long bytes, unsigned long *p1, unsigned long *p2,
+		     unsigned long *p3)
+{
+	asm volatile(
+		"	larl	1,2f\n"
+		"	aghi	%0,-1\n"
+		"	jm	3f\n"
+		"	srlg	0,%0,8\n"
+		"	ltgr	0,0\n"
+		"	jz	1f\n"
+		"0:	xc	0(256,%1),0(%2)\n"
+		"	xc	0(256,%1),0(%3)\n"
+		"	la	%1,256(%1)\n"
+		"	la	%2,256(%2)\n"
+		"	la	%3,256(%3)\n"
+		"	brctg	0,0b\n"
+		"1:	ex	%0,0(1)\n"
+		"	ex	%0,6(1)\n"
+		"	j	3f\n"
+		"2:	xc	0(1,%1),0(%2)\n"
+		"	xc	0(1,%1),0(%3)\n"
+		"3:\n"
+		: "+d" (bytes), "+a" (p1), "+a" (p2), "+a" (p3)
+		: : "0", "1", "cc", "memory");
+}
+
+static void xor_xc_4(unsigned long bytes, unsigned long *p1, unsigned long *p2,
+		     unsigned long *p3, unsigned long *p4)
+{
+	asm volatile(
+		"	larl	1,2f\n"
+		"	aghi	%0,-1\n"
+		"	jm	3f\n"
+		"	srlg	0,%0,8\n"
+		"	ltgr	0,0\n"
+		"	jz	1f\n"
+		"0:	xc	0(256,%1),0(%2)\n"
+		"	xc	0(256,%1),0(%3)\n"
+		"	xc	0(256,%1),0(%4)\n"
+		"	la	%1,256(%1)\n"
+		"	la	%2,256(%2)\n"
+		"	la	%3,256(%3)\n"
+		"	la	%4,256(%4)\n"
+		"	brctg	0,0b\n"
+		"1:	ex	%0,0(1)\n"
+		"	ex	%0,6(1)\n"
+		"	ex	%0,12(1)\n"
+		"	j	3f\n"
+		"2:	xc	0(1,%1),0(%2)\n"
+		"	xc	0(1,%1),0(%3)\n"
+		"	xc	0(1,%1),0(%4)\n"
+		"3:\n"
+		: "+d" (bytes), "+a" (p1), "+a" (p2), "+a" (p3), "+a" (p4)
+		: : "0", "1", "cc", "memory");
+}
+
+static void xor_xc_5(unsigned long bytes, unsigned long *p1, unsigned long *p2,
+		     unsigned long *p3, unsigned long *p4, unsigned long *p5)
+{
+	/* Get around a gcc oddity */
+	register unsigned long *reg7 asm ("7") = p5;
+
+	asm volatile(
+		"	larl	1,2f\n"
+		"	aghi	%0,-1\n"
+		"	jm	3f\n"
+		"	srlg	0,%0,8\n"
+		"	ltgr	0,0\n"
+		"	jz	1f\n"
+		"0:	xc	0(256,%1),0(%2)\n"
+		"	xc	0(256,%1),0(%3)\n"
+		"	xc	0(256,%1),0(%4)\n"
+		"	xc	0(256,%1),0(%5)\n"
+		"	la	%1,256(%1)\n"
+		"	la	%2,256(%2)\n"
+		"	la	%3,256(%3)\n"
+		"	la	%4,256(%4)\n"
+		"	la	%5,256(%5)\n"
+		"	brctg	0,0b\n"
+		"1:	ex	%0,0(1)\n"
+		"	ex	%0,6(1)\n"
+		"	ex	%0,12(1)\n"
+		"	ex	%0,18(1)\n"
+		"	j	3f\n"
+		"2:	xc	0(1,%1),0(%2)\n"
+		"	xc	0(1,%1),0(%3)\n"
+		"	xc	0(1,%1),0(%4)\n"
+		"	xc	0(1,%1),0(%5)\n"
+		"3:\n"
+		: "+d" (bytes), "+a" (p1), "+a" (p2), "+a" (p3), "+a" (p4),
+		  "+a" (reg7)
+		: : "0", "1", "cc", "memory");
+}
+
+struct xor_block_template xor_block_xc = {
+	.name = "xc",
+	.do_2 = xor_xc_2,
+	.do_3 = xor_xc_3,
+	.do_4 = xor_xc_4,
+	.do_5 = xor_xc_5,
+};
+EXPORT_SYMBOL(xor_block_xc);