Add AltiVec RAID-Z

Implements the RAID-Z function using AltiVec SIMD. This is basically the NEON code translated to AltiVec. Note that the 'fletcher' algorithm requires 64-bits operations, and the initial implementations of AltiVec (PPC74xx a.k.a. G4, PPC970 a.k.a. G5) only has up to 32-bits operations, so no 'fletcher'. Reviewed-by: Brian Behlendorf <[email protected]> Signed-off-by: Romain Dolbeau <[email protected]> Closes openzfs#9539
loli10K · Jan 23, 2020 · 35b0749 · 35b0749
1 parent 1a69856
commit 35b0749
Show file tree

Hide file tree

Showing 12 changed files with 5,200 additions and 1 deletion.
diff --git a/cmd/raidz_test/raidz_test.h b/cmd/raidz_test/raidz_test.h
@@ -38,6 +38,7 @@ static const char *raidz_impl_names[] = {
 	"avx512bw",
 	"aarch64_neon",
 	"aarch64_neonx2",
+	"powerpc_altivec",
 	NULL
 };
 

diff --git a/include/os/linux/kernel/linux/Makefile.am b/include/os/linux/kernel/linux/Makefile.am
@@ -10,6 +10,7 @@ KERNEL_H = \
 	$(top_srcdir)/include/os/linux/kernel/linux/simd.h \
 	$(top_srcdir)/include/os/linux/kernel/linux/simd_x86.h \
 	$(top_srcdir)/include/os/linux/kernel/linux/simd_aarch64.h \
+	$(top_srcdir)/include/os/linux/kernel/linux/simd_powerpc.h \
 	$(top_srcdir)/include/os/linux/kernel/linux/mod_compat.h \
 	$(top_srcdir)/include/os/linux/kernel/linux/page_compat.h \
 	$(top_srcdir)/include/os/linux/kernel/linux/compiler_compat.h

diff --git a/include/os/linux/kernel/linux/simd.h b/include/os/linux/kernel/linux/simd.h
@@ -30,6 +30,9 @@
 
 #elif defined(__aarch64__)
 #include <linux/simd_aarch64.h>
+
+#elif defined(__powerpc__)
+#include <linux/simd_powerpc.h>
 #else
 
 #define	kfpu_allowed()		0

diff --git a/include/os/linux/kernel/linux/simd_powerpc.h b/include/os/linux/kernel/linux/simd_powerpc.h
@@ -0,0 +1,109 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License (the "License").
+ * You may not use this file except in compliance with the License.
+ *
+ * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+ * or http://www.opensolaris.org/os/licensing.
+ * See the License for the specific language governing permissions
+ * and limitations under the License.
+ *
+ * When distributing Covered Code, include this CDDL HEADER in each
+ * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+ * If applicable, add the following below this CDDL HEADER, with the
+ * fields enclosed by brackets "[]" replaced with your own identifying
+ * information: Portions Copyright [yyyy] [name of copyright owner]
+ *
+ * CDDL HEADER END
+ */
+/*
+ * Copyright (C) 2019 Romain Dolbeau
+ *           <[email protected]>
+ */
+
+/*
+ * USER API:
+ *
+ * Kernel fpu methods:
+ *	kfpu_allowed()
+ *	kfpu_begin()
+ *	kfpu_end()
+ *	kfpu_init()
+ *	kfpu_fini()
+ *
+ * SIMD support:
+ *
+ * Following functions should be called to determine whether CPU feature
+ * is supported. All functions are usable in kernel and user space.
+ * If a SIMD algorithm is using more than one instruction set
+ * all relevant feature test functions should be called.
+ *
+ * Supported features:
+ *	zfs_altivec_available()
+ */
+
+#ifndef _LINUX_SIMD_POWERPC_H
+#define	_LINUX_SIMD_POWERPC_H
+
+/* only for __powerpc__ */
+#if defined(__powerpc__)
+
+#include <linux/preempt.h>
+#include <linux/export.h>
+#include <linux/sched.h>
+#include <asm/switch_to.h>
+#include <sys/types.h>
+#include <linux/version.h>
+
+#define	kfpu_allowed()		1
+#define	kfpu_begin()					\
+	{						\
+		preempt_disable();			\
+		enable_kernel_altivec();		\
+	}
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 5, 0)
+#define	kfpu_end()				\
+	{					\
+		disable_kernel_altivec();	\
+		preempt_enable();		\
+	}
+#else
+/* seems that before 4.5 no-one bothered disabling ... */
+#define	kfpu_end()		preempt_enable()
+#endif
+#define	kfpu_init()		0
+#define	kfpu_fini()		((void) 0)
+
+/*
+ * Check if AltiVec instruction set is available
+ */
+static inline boolean_t
+zfs_altivec_available(void)
+{
+	boolean_t res;
+	/* suggested by macallan at netbsd dot org */
+#if defined(__powerpc64__)
+	u64 msr;
+#else
+	u32 msr;
+#endif
+	kfpu_begin();
+	__asm volatile("mfmsr %0" : "=r"(msr));
+	/*
+	 * 64 bits -> need to check bit 38
+	 * Power ISA Version 3.0B
+	 * p944
+	 * 32 bits -> Need to check bit 6
+	 * AltiVec Technology Programming Environments Manual
+	 * p49 (2-9)
+	 * They are the same, as ppc counts 'backward' ...
+	 */
+	res = (msr & 0x2000000) != 0;
+	kfpu_end();
+	return (res);
+}
+#endif /* defined(__powerpc) */
+
+#endif /* _LINUX_SIMD_POWERPC_H */
diff --git a/include/sys/vdev_raidz_impl.h b/include/sys/vdev_raidz_impl.h
@@ -90,7 +90,7 @@ typedef boolean_t	(*will_work_f)(void);
 typedef void		(*init_impl_f)(void);
 typedef void		(*fini_impl_f)(void);
 
-#define	RAIDZ_IMPL_NAME_MAX	(16)
+#define	RAIDZ_IMPL_NAME_MAX	(20)
 
 typedef struct raidz_impl_ops {
 	init_impl_f init;
@@ -152,6 +152,9 @@ extern const raidz_impl_ops_t vdev_raidz_avx512bw_impl;
 extern const raidz_impl_ops_t vdev_raidz_aarch64_neon_impl;
 extern const raidz_impl_ops_t vdev_raidz_aarch64_neonx2_impl;
 #endif
+#if defined(__powerpc__)
+extern const raidz_impl_ops_t vdev_raidz_powerpc_altivec_impl;
+#endif
 
 /*
  * Commonly used raidz_map helpers

diff --git a/lib/libspl/include/sys/simd.h b/lib/libspl/include/sys/simd.h
@@ -437,6 +437,46 @@ zfs_avx512vbmi_available(void)
 #define	kfpu_begin()		do {} while (0)
 #define	kfpu_end()		do {} while (0)
 
+#elif defined(__powerpc__)
+
+#define	kfpu_allowed()		1
+#define	kfpu_initialize(tsk)	do {} while (0)
+#define	kfpu_begin()		do {} while (0)
+#define	kfpu_end()		do {} while (0)
+
+/*
+ * Check if AltiVec instruction set is available
+ * No easy way beyond 'altivec works' :-(
+ */
+#include <signal.h>
+#include <setjmp.h>
+
+#ifdef __ALTIVEC__
+static jmp_buf env;
+static void sigillhandler(int x)
+{
+	longjmp(env, 1);
+}
+#endif
+
+static inline boolean_t
+zfs_altivec_available(void)
+{
+	boolean_t has_altivec = B_FALSE;
+#ifdef __ALTIVEC__
+	sighandler_t savesig;
+	savesig = signal(SIGILL, sigillhandler);
+	if (setjmp(env)) {
+		signal(SIGILL, savesig);
+		has_altivec = B_FALSE;
+	} else {
+		__asm__ __volatile__("vor 0,0,0\n" : : : "v0");
+		signal(SIGILL, savesig);
+		has_altivec = B_TRUE;
+	}
+#endif
+	return (has_altivec);
+}
 #else
 
 #define	kfpu_allowed()		0

diff --git a/lib/libzpool/Makefile.am b/lib/libzpool/Makefile.am
@@ -130,6 +130,7 @@ KERNEL_C = \
 	vdev_raidz_math_scalar.c \
 	vdev_raidz_math_sse2.c \
 	vdev_raidz_math_ssse3.c \
+	vdev_raidz_math_powerpc_altivec.c \
 	vdev_removal.c \
 	vdev_root.c \
 	vdev_trim.c \
@@ -201,3 +202,8 @@ libzpool_la_LIBADD += $(ZLIB) -ldl
 libzpool_la_LDFLAGS = -pthread -version-info 2:0:0
 
 EXTRA_DIST = $(USER_C)
+
+if TARGET_CPU_POWERPC
+vdev_raidz_math_powerpc_altivec.$(OBJEXT): CFLAGS += -maltivec
+vdev_raidz_math_powerpc_altivec.l$(OBJEXT): CFLAGS += -maltivec
+endif
diff --git a/man/man5/zfs-module-parameters.5 b/man/man5/zfs-module-parameters.5
@@ -3374,6 +3374,7 @@ Possible options are:
   avx512bw - implementation using AVX512F & AVX512BW instruction sets (64bit x86 only)
   aarch64_neon - implementation using NEON (Aarch64/64 bit ARMv8 only)
   aarch64_neonx2 - implementation using NEON with more unrolling (Aarch64/64 bit ARMv8 only)
+  powerpc_altivec - implementation using Altivec (PowerPC only)
 .sp
 Default value: \fBfastest\fR.
 .RE

diff --git a/module/zfs/Makefile.in b/module/zfs/Makefile.in
@@ -142,4 +142,11 @@ $(MODULE)-$(CONFIG_X86) += vdev_raidz_math_avx512bw.o
 $(MODULE)-$(CONFIG_ARM64) += vdev_raidz_math_aarch64_neon.o
 $(MODULE)-$(CONFIG_ARM64) += vdev_raidz_math_aarch64_neonx2.o
 
+$(MODULE)-$(CONFIG_PPC) += vdev_raidz_math_powerpc_altivec.o
+$(MODULE)-$(CONFIG_PPC64) += vdev_raidz_math_powerpc_altivec.o
+
+ifeq ($(CONFIG_ALTIVEC),y)
+$(obj)/vdev_raidz_math_powerpc_altivec.o: c_flags += -maltivec
+endif
+
 -include @abs_top_builddir@/module/os/linux/zfs/Makefile
diff --git a/module/zfs/vdev_raidz_math.c b/module/zfs/vdev_raidz_math.c
@@ -67,6 +67,9 @@ const raidz_impl_ops_t *raidz_all_maths[] = {
 	&vdev_raidz_aarch64_neon_impl,
 	&vdev_raidz_aarch64_neonx2_impl,
 #endif
+#if defined(__powerpc__)
+	&vdev_raidz_powerpc_altivec_impl,
+#endif
 };
 
 /* Indicate that benchmark has been completed */