loongarch: move the XOR code to lib/raid/

Move the optimized XOR into lib/raid and include it it in xor.ko instead
of always building it into the main kernel image.

Link: https://lkml.kernel.org/r/20260327061704.3707577-15-hch@lst.de
Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Eric Biggers <ebiggers@kernel.org>
Tested-by: Eric Biggers <ebiggers@kernel.org>
Cc: Albert Ou <aou@eecs.berkeley.edu>
Cc: Alexander Gordeev <agordeev@linux.ibm.com>
Cc: Alexandre Ghiti <alex@ghiti.fr>
Cc: Andreas Larsson <andreas@gaisler.com>
Cc: Anton Ivanov <anton.ivanov@cambridgegreys.com>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: "Borislav Petkov (AMD)" <bp@alien8.de>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Chris Mason <clm@fb.com>
Cc: Christian Borntraeger <borntraeger@linux.ibm.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: David Sterba <dsterba@suse.com>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Huacai Chen <chenhuacai@kernel.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Jason A. Donenfeld <jason@zx2c4.com>
Cc: Johannes Berg <johannes@sipsolutions.net>
Cc: Li Nan <linan122@huawei.com>
Cc: Madhavan Srinivasan <maddy@linux.ibm.com>
Cc: Magnus Lindholm <linmag7@gmail.com>
Cc: Matt Turner <mattst88@gmail.com>
Cc: Michael Ellerman <mpe@ellerman.id.au>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: Palmer Dabbelt <palmer@dabbelt.com>
Cc: Richard Henderson <richard.henderson@linaro.org>
Cc: Richard Weinberger <richard@nod.at>
Cc: Russell King <linux@armlinux.org.uk>
Cc: Song Liu <song@kernel.org>
Cc: Sven Schnelle <svens@linux.ibm.com>
Cc: Ted Ts'o <tytso@mit.edu>
Cc: Vasily Gorbik <gor@linux.ibm.com>
Cc: WANG Xuerui <kernel@xen0n.name>
Cc: Will Deacon <will@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

5 files changed, 320 insertions, 0 deletions

diff --git a/lib/raid/xor/Makefile b/lib/raid/xor/Makefile
index 4ab0e7411ff7..e8868f5fc396 100644
--- a/lib/raid/xor/Makefile
+++ b/lib/raid/xor/Makefile
@@ -14,6 +14,8 @@ ifeq ($(CONFIG_ARM),y)
 xor-$(CONFIG_KERNEL_MODE_NEON)	+= arm/xor-neon.o arm/xor-neon-glue.o
 endif
 xor-$(CONFIG_ARM64)		+= arm64/xor-neon.o arm64/xor-neon-glue.o
+xor-$(CONFIG_CPU_HAS_LSX)	+= loongarch/xor_simd.o
+xor-$(CONFIG_CPU_HAS_LSX)	+= loongarch/xor_simd_glue.o
 
 
 CFLAGS_arm/xor-neon.o		+= $(CC_FLAGS_FPU)
diff --git a/lib/raid/xor/loongarch/xor_simd.c b/lib/raid/xor/loongarch/xor_simd.c
new file mode 100644
index 000000000000..84cd24b728c4
--- /dev/null
+++ b/lib/raid/xor/loongarch/xor_simd.c
@@ -0,0 +1,93 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * LoongArch SIMD XOR operations
+ *
+ * Copyright (C) 2023 WANG Xuerui <git@xen0n.name>
+ */
+
+#include "xor_simd.h"
+
+/*
+ * Process one cache line (64 bytes) per loop. This is assuming all future
+ * popular LoongArch cores are similar performance-characteristics-wise to the
+ * current models.
+ */
+#define LINE_WIDTH 64
+
+#ifdef CONFIG_CPU_HAS_LSX
+
+#define LD(reg, base, offset)	\
+	"vld $vr" #reg ", %[" #base "], " #offset "\n\t"
+#define ST(reg, base, offset)	\
+	"vst $vr" #reg ", %[" #base "], " #offset "\n\t"
+#define XOR(dj, k)	"vxor.v $vr" #dj ", $vr" #dj ", $vr" #k "\n\t"
+
+#define LD_INOUT_LINE(base)	\
+	LD(0, base, 0)		\
+	LD(1, base, 16)		\
+	LD(2, base, 32)		\
+	LD(3, base, 48)
+
+#define LD_AND_XOR_LINE(base)	\
+	LD(4, base, 0)		\
+	LD(5, base, 16)		\
+	LD(6, base, 32)		\
+	LD(7, base, 48)		\
+	XOR(0, 4)		\
+	XOR(1, 5)		\
+	XOR(2, 6)		\
+	XOR(3, 7)
+
+#define ST_LINE(base)		\
+	ST(0, base, 0)		\
+	ST(1, base, 16)		\
+	ST(2, base, 32)		\
+	ST(3, base, 48)
+
+#define XOR_FUNC_NAME(nr) __xor_lsx_##nr
+#include "xor_template.c"
+
+#undef LD
+#undef ST
+#undef XOR
+#undef LD_INOUT_LINE
+#undef LD_AND_XOR_LINE
+#undef ST_LINE
+#undef XOR_FUNC_NAME
+
+#endif /* CONFIG_CPU_HAS_LSX */
+
+#ifdef CONFIG_CPU_HAS_LASX
+
+#define LD(reg, base, offset)	\
+	"xvld $xr" #reg ", %[" #base "], " #offset "\n\t"
+#define ST(reg, base, offset)	\
+	"xvst $xr" #reg ", %[" #base "], " #offset "\n\t"
+#define XOR(dj, k)	"xvxor.v $xr" #dj ", $xr" #dj ", $xr" #k "\n\t"
+
+#define LD_INOUT_LINE(base)	\
+	LD(0, base, 0)		\
+	LD(1, base, 32)
+
+#define LD_AND_XOR_LINE(base)	\
+	LD(2, base, 0)		\
+	LD(3, base, 32)		\
+	XOR(0, 2)		\
+	XOR(1, 3)
+
+#define ST_LINE(base)		\
+	ST(0, base, 0)		\
+	ST(1, base, 32)
+
+#define XOR_FUNC_NAME(nr) __xor_lasx_##nr
+#include "xor_template.c"
+
+#undef LD
+#undef ST
+#undef XOR
+#undef LD_INOUT_LINE
+#undef LD_AND_XOR_LINE
+#undef ST_LINE
+#undef XOR_FUNC_NAME
+
+#endif /* CONFIG_CPU_HAS_LASX */
diff --git a/lib/raid/xor/loongarch/xor_simd.h b/lib/raid/xor/loongarch/xor_simd.h
new file mode 100644
index 000000000000..f50f32514d80
--- /dev/null
+++ b/lib/raid/xor/loongarch/xor_simd.h
@@ -0,0 +1,38 @@
+/* SPDX-License-Identifier: GPL-2.0-or-later */
+/*
+ * Simple interface to link xor_simd.c and xor_simd_glue.c
+ *
+ * Separating these files ensures that no SIMD instructions are run outside of
+ * the kfpu critical section.
+ */
+
+#ifndef __LOONGARCH_LIB_XOR_SIMD_H
+#define __LOONGARCH_LIB_XOR_SIMD_H
+
+#ifdef CONFIG_CPU_HAS_LSX
+void __xor_lsx_2(unsigned long bytes, unsigned long * __restrict p1,
+		 const unsigned long * __restrict p2);
+void __xor_lsx_3(unsigned long bytes, unsigned long * __restrict p1,
+		 const unsigned long * __restrict p2, const unsigned long * __restrict p3);
+void __xor_lsx_4(unsigned long bytes, unsigned long * __restrict p1,
+		 const unsigned long * __restrict p2, const unsigned long * __restrict p3,
+		 const unsigned long * __restrict p4);
+void __xor_lsx_5(unsigned long bytes, unsigned long * __restrict p1,
+		 const unsigned long * __restrict p2, const unsigned long * __restrict p3,
+		 const unsigned long * __restrict p4, const unsigned long * __restrict p5);
+#endif /* CONFIG_CPU_HAS_LSX */
+
+#ifdef CONFIG_CPU_HAS_LASX
+void __xor_lasx_2(unsigned long bytes, unsigned long * __restrict p1,
+		  const unsigned long * __restrict p2);
+void __xor_lasx_3(unsigned long bytes, unsigned long * __restrict p1,
+		  const unsigned long * __restrict p2, const unsigned long * __restrict p3);
+void __xor_lasx_4(unsigned long bytes, unsigned long * __restrict p1,
+		  const unsigned long * __restrict p2, const unsigned long * __restrict p3,
+		  const unsigned long * __restrict p4);
+void __xor_lasx_5(unsigned long bytes, unsigned long * __restrict p1,
+		  const unsigned long * __restrict p2, const unsigned long * __restrict p3,
+		  const unsigned long * __restrict p4, const unsigned long * __restrict p5);
+#endif /* CONFIG_CPU_HAS_LASX */
+
+#endif /* __LOONGARCH_LIB_XOR_SIMD_H */
diff --git a/lib/raid/xor/loongarch/xor_simd_glue.c b/lib/raid/xor/loongarch/xor_simd_glue.c
new file mode 100644
index 000000000000..11fa3b47ba83
--- /dev/null
+++ b/lib/raid/xor/loongarch/xor_simd_glue.c
@@ -0,0 +1,77 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * LoongArch SIMD XOR operations
+ *
+ * Copyright (C) 2023 WANG Xuerui <git@xen0n.name>
+ */
+
+#include <linux/sched.h>
+#include <linux/raid/xor_impl.h>
+#include <asm/fpu.h>
+#include <asm/xor.h>
+#include "xor_simd.h"
+
+#define MAKE_XOR_GLUE_2(flavor)							\
+static void xor_##flavor##_2(unsigned long bytes, unsigned long * __restrict p1,\
+		      const unsigned long * __restrict p2)			\
+{										\
+	kernel_fpu_begin();							\
+	__xor_##flavor##_2(bytes, p1, p2);					\
+	kernel_fpu_end();							\
+}										\
+
+#define MAKE_XOR_GLUE_3(flavor)							\
+static void xor_##flavor##_3(unsigned long bytes, unsigned long * __restrict p1,\
+		      const unsigned long * __restrict p2,			\
+		      const unsigned long * __restrict p3)			\
+{										\
+	kernel_fpu_begin();							\
+	__xor_##flavor##_3(bytes, p1, p2, p3);					\
+	kernel_fpu_end();							\
+}										\
+
+#define MAKE_XOR_GLUE_4(flavor)							\
+static void xor_##flavor##_4(unsigned long bytes, unsigned long * __restrict p1,\
+		      const unsigned long * __restrict p2,			\
+		      const unsigned long * __restrict p3,			\
+		      const unsigned long * __restrict p4)			\
+{										\
+	kernel_fpu_begin();							\
+	__xor_##flavor##_4(bytes, p1, p2, p3, p4);				\
+	kernel_fpu_end();							\
+}										\
+
+#define MAKE_XOR_GLUE_5(flavor)							\
+static void xor_##flavor##_5(unsigned long bytes, unsigned long * __restrict p1,\
+		      const unsigned long * __restrict p2,			\
+		      const unsigned long * __restrict p3,			\
+		      const unsigned long * __restrict p4,			\
+		      const unsigned long * __restrict p5)			\
+{										\
+	kernel_fpu_begin();							\
+	__xor_##flavor##_5(bytes, p1, p2, p3, p4, p5);				\
+	kernel_fpu_end();							\
+}										\
+
+#define MAKE_XOR_GLUES(flavor)				\
+	MAKE_XOR_GLUE_2(flavor);			\
+	MAKE_XOR_GLUE_3(flavor);			\
+	MAKE_XOR_GLUE_4(flavor);			\
+	MAKE_XOR_GLUE_5(flavor);			\
+							\
+struct xor_block_template xor_block_##flavor = {	\
+	.name = __stringify(flavor),			\
+	.do_2 = xor_##flavor##_2,			\
+	.do_3 = xor_##flavor##_3,			\
+	.do_4 = xor_##flavor##_4,			\
+	.do_5 = xor_##flavor##_5,			\
+}
+
+
+#ifdef CONFIG_CPU_HAS_LSX
+MAKE_XOR_GLUES(lsx);
+#endif /* CONFIG_CPU_HAS_LSX */
+
+#ifdef CONFIG_CPU_HAS_LASX
+MAKE_XOR_GLUES(lasx);
+#endif /* CONFIG_CPU_HAS_LASX */
diff --git a/lib/raid/xor/loongarch/xor_template.c b/lib/raid/xor/loongarch/xor_template.c
new file mode 100644
index 000000000000..0358ced7fe33
--- /dev/null
+++ b/lib/raid/xor/loongarch/xor_template.c
@@ -0,0 +1,110 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright (C) 2023 WANG Xuerui <git@xen0n.name>
+ *
+ * Template for XOR operations, instantiated in xor_simd.c.
+ *
+ * Expected preprocessor definitions:
+ *
+ * - LINE_WIDTH
+ * - XOR_FUNC_NAME(nr)
+ * - LD_INOUT_LINE(buf)
+ * - LD_AND_XOR_LINE(buf)
+ * - ST_LINE(buf)
+ */
+
+void XOR_FUNC_NAME(2)(unsigned long bytes,
+		      unsigned long * __restrict v1,
+		      const unsigned long * __restrict v2)
+{
+	unsigned long lines = bytes / LINE_WIDTH;
+
+	do {
+		__asm__ __volatile__ (
+			LD_INOUT_LINE(v1)
+			LD_AND_XOR_LINE(v2)
+			ST_LINE(v1)
+		: : [v1] "r"(v1), [v2] "r"(v2) : "memory"
+		);
+
+		v1 += LINE_WIDTH / sizeof(unsigned long);
+		v2 += LINE_WIDTH / sizeof(unsigned long);
+	} while (--lines > 0);
+}
+
+void XOR_FUNC_NAME(3)(unsigned long bytes,
+		      unsigned long * __restrict v1,
+		      const unsigned long * __restrict v2,
+		      const unsigned long * __restrict v3)
+{
+	unsigned long lines = bytes / LINE_WIDTH;
+
+	do {
+		__asm__ __volatile__ (
+			LD_INOUT_LINE(v1)
+			LD_AND_XOR_LINE(v2)
+			LD_AND_XOR_LINE(v3)
+			ST_LINE(v1)
+		: : [v1] "r"(v1), [v2] "r"(v2), [v3] "r"(v3) : "memory"
+		);
+
+		v1 += LINE_WIDTH / sizeof(unsigned long);
+		v2 += LINE_WIDTH / sizeof(unsigned long);
+		v3 += LINE_WIDTH / sizeof(unsigned long);
+	} while (--lines > 0);
+}
+
+void XOR_FUNC_NAME(4)(unsigned long bytes,
+		      unsigned long * __restrict v1,
+		      const unsigned long * __restrict v2,
+		      const unsigned long * __restrict v3,
+		      const unsigned long * __restrict v4)
+{
+	unsigned long lines = bytes / LINE_WIDTH;
+
+	do {
+		__asm__ __volatile__ (
+			LD_INOUT_LINE(v1)
+			LD_AND_XOR_LINE(v2)
+			LD_AND_XOR_LINE(v3)
+			LD_AND_XOR_LINE(v4)
+			ST_LINE(v1)
+		: : [v1] "r"(v1), [v2] "r"(v2), [v3] "r"(v3), [v4] "r"(v4)
+		: "memory"
+		);
+
+		v1 += LINE_WIDTH / sizeof(unsigned long);
+		v2 += LINE_WIDTH / sizeof(unsigned long);
+		v3 += LINE_WIDTH / sizeof(unsigned long);
+		v4 += LINE_WIDTH / sizeof(unsigned long);
+	} while (--lines > 0);
+}
+
+void XOR_FUNC_NAME(5)(unsigned long bytes,
+		      unsigned long * __restrict v1,
+		      const unsigned long * __restrict v2,
+		      const unsigned long * __restrict v3,
+		      const unsigned long * __restrict v4,
+		      const unsigned long * __restrict v5)
+{
+	unsigned long lines = bytes / LINE_WIDTH;
+
+	do {
+		__asm__ __volatile__ (
+			LD_INOUT_LINE(v1)
+			LD_AND_XOR_LINE(v2)
+			LD_AND_XOR_LINE(v3)
+			LD_AND_XOR_LINE(v4)
+			LD_AND_XOR_LINE(v5)
+			ST_LINE(v1)
+		: : [v1] "r"(v1), [v2] "r"(v2), [v3] "r"(v3), [v4] "r"(v4),
+		    [v5] "r"(v5) : "memory"
+		);
+
+		v1 += LINE_WIDTH / sizeof(unsigned long);
+		v2 += LINE_WIDTH / sizeof(unsigned long);
+		v3 += LINE_WIDTH / sizeof(unsigned long);
+		v4 += LINE_WIDTH / sizeof(unsigned long);
+		v5 += LINE_WIDTH / sizeof(unsigned long);
+	} while (--lines > 0);
+}