perf bench: Copy kernel files needed to build mem{cpy,set} x86_64 ben…

…chmarks

commit 7d7d1bf1d1dabe435ef50efb051724b8664749cb upstream.

We can't access kernel files directly from tools/, so copy the required
bits, and make sure that we detect when the original files, in the
kernel, gets modified.

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-z7e76274ch5j4nugv048qacb@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tools/arch/x86/include/asm/disabled-features.h

@@ -0,0 +1,65 @@
+#ifndef _ASM_X86_DISABLED_FEATURES_H
+#define _ASM_X86_DISABLED_FEATURES_H
+
+/* These features, although they might be available in a CPU
+ * will not be used because the compile options to support
+ * them are not present.
+ *
+ * This code allows them to be checked and disabled at
+ * compile time without an explicit #ifdef.  Use
+ * cpu_feature_enabled().
+ */
+
+#ifdef CONFIG_X86_INTEL_MPX
+# define DISABLE_MPX	0
+#else
+# define DISABLE_MPX	(1<<(X86_FEATURE_MPX & 31))
+#endif
+
+#ifdef CONFIG_X86_64
+# define DISABLE_VME		(1<<(X86_FEATURE_VME & 31))
+# define DISABLE_K6_MTRR	(1<<(X86_FEATURE_K6_MTRR & 31))
+# define DISABLE_CYRIX_ARR	(1<<(X86_FEATURE_CYRIX_ARR & 31))
+# define DISABLE_CENTAUR_MCR	(1<<(X86_FEATURE_CENTAUR_MCR & 31))
+# define DISABLE_PCID		0
+#else
+# define DISABLE_VME		0
+# define DISABLE_K6_MTRR	0
+# define DISABLE_CYRIX_ARR	0
+# define DISABLE_CENTAUR_MCR	0
+# define DISABLE_PCID		(1<<(X86_FEATURE_PCID & 31))
+#endif /* CONFIG_X86_64 */
+
+#ifdef CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS
+# define DISABLE_PKU		0
+# define DISABLE_OSPKE		0
+#else
+# define DISABLE_PKU		(1<<(X86_FEATURE_PKU & 31))
+# define DISABLE_OSPKE		(1<<(X86_FEATURE_OSPKE & 31))
+#endif /* CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS */
+
+/*
+ * Make sure to add features to the correct mask
+ */
+#define DISABLED_MASK0	(DISABLE_VME)
+#define DISABLED_MASK1	0
+#define DISABLED_MASK2	0
+#define DISABLED_MASK3	(DISABLE_CYRIX_ARR|DISABLE_CENTAUR_MCR|DISABLE_K6_MTRR)
+#define DISABLED_MASK4	(DISABLE_PCID)
+#define DISABLED_MASK5	0
+#define DISABLED_MASK6	0
+#define DISABLED_MASK7	0
+#define DISABLED_MASK8	0
+#define DISABLED_MASK9	(DISABLE_MPX)
+#define DISABLED_MASK10	0
+#define DISABLED_MASK11	0
+#define DISABLED_MASK12	0
+#define DISABLED_MASK13	0
+#define DISABLED_MASK14	0
+#define DISABLED_MASK15	0
+#define DISABLED_MASK16	(DISABLE_PKU|DISABLE_OSPKE)
+#define DISABLED_MASK17	0
+#define DISABLED_MASK18	0
+#define DISABLED_MASK_CHECK BUILD_BUG_ON_ZERO(NCAPINTS != 19)
+
+#endif /* _ASM_X86_DISABLED_FEATURES_H */

tools/arch/x86/include/asm/required-features.h

@@ -0,0 +1,106 @@
+#ifndef _ASM_X86_REQUIRED_FEATURES_H
+#define _ASM_X86_REQUIRED_FEATURES_H
+
+/* Define minimum CPUID feature set for kernel These bits are checked
+   really early to actually display a visible error message before the
+   kernel dies.  Make sure to assign features to the proper mask!
+
+   Some requirements that are not in CPUID yet are also in the
+   CONFIG_X86_MINIMUM_CPU_FAMILY which is checked too.
+
+   The real information is in arch/x86/Kconfig.cpu, this just converts
+   the CONFIGs into a bitmask */
+
+#ifndef CONFIG_MATH_EMULATION
+# define NEED_FPU	(1<<(X86_FEATURE_FPU & 31))
+#else
+# define NEED_FPU	0
+#endif
+
+#if defined(CONFIG_X86_PAE) || defined(CONFIG_X86_64)
+# define NEED_PAE	(1<<(X86_FEATURE_PAE & 31))
+#else
+# define NEED_PAE	0
+#endif
+
+#ifdef CONFIG_X86_CMPXCHG64
+# define NEED_CX8	(1<<(X86_FEATURE_CX8 & 31))
+#else
+# define NEED_CX8	0
+#endif
+
+#if defined(CONFIG_X86_CMOV) || defined(CONFIG_X86_64)
+# define NEED_CMOV	(1<<(X86_FEATURE_CMOV & 31))
+#else
+# define NEED_CMOV	0
+#endif
+
+#ifdef CONFIG_X86_USE_3DNOW
+# define NEED_3DNOW	(1<<(X86_FEATURE_3DNOW & 31))
+#else
+# define NEED_3DNOW	0
+#endif
+
+#if defined(CONFIG_X86_P6_NOP) || defined(CONFIG_X86_64)
+# define NEED_NOPL	(1<<(X86_FEATURE_NOPL & 31))
+#else
+# define NEED_NOPL	0
+#endif
+
+#ifdef CONFIG_MATOM
+# define NEED_MOVBE	(1<<(X86_FEATURE_MOVBE & 31))
+#else
+# define NEED_MOVBE	0
+#endif
+
+#ifdef CONFIG_X86_64
+#ifdef CONFIG_PARAVIRT
+/* Paravirtualized systems may not have PSE or PGE available */
+#define NEED_PSE	0
+#define NEED_PGE	0
+#else
+#define NEED_PSE	(1<<(X86_FEATURE_PSE) & 31)
+#define NEED_PGE	(1<<(X86_FEATURE_PGE) & 31)
+#endif
+#define NEED_MSR	(1<<(X86_FEATURE_MSR & 31))
+#define NEED_FXSR	(1<<(X86_FEATURE_FXSR & 31))
+#define NEED_XMM	(1<<(X86_FEATURE_XMM & 31))
+#define NEED_XMM2	(1<<(X86_FEATURE_XMM2 & 31))
+#define NEED_LM		(1<<(X86_FEATURE_LM & 31))
+#else
+#define NEED_PSE	0
+#define NEED_MSR	0
+#define NEED_PGE	0
+#define NEED_FXSR	0
+#define NEED_XMM	0
+#define NEED_XMM2	0
+#define NEED_LM		0
+#endif
+
+#define REQUIRED_MASK0	(NEED_FPU|NEED_PSE|NEED_MSR|NEED_PAE|\
+			 NEED_CX8|NEED_PGE|NEED_FXSR|NEED_CMOV|\
+			 NEED_XMM|NEED_XMM2)
+#define SSE_MASK	(NEED_XMM|NEED_XMM2)
+
+#define REQUIRED_MASK1	(NEED_LM|NEED_3DNOW)
+
+#define REQUIRED_MASK2	0
+#define REQUIRED_MASK3	(NEED_NOPL)
+#define REQUIRED_MASK4	(NEED_MOVBE)
+#define REQUIRED_MASK5	0
+#define REQUIRED_MASK6	0
+#define REQUIRED_MASK7	0
+#define REQUIRED_MASK8	0
+#define REQUIRED_MASK9	0
+#define REQUIRED_MASK10	0
+#define REQUIRED_MASK11	0
+#define REQUIRED_MASK12	0
+#define REQUIRED_MASK13	0
+#define REQUIRED_MASK14	0
+#define REQUIRED_MASK15	0
+#define REQUIRED_MASK16	0
+#define REQUIRED_MASK17	0
+#define REQUIRED_MASK18	0
+#define REQUIRED_MASK_CHECK BUILD_BUG_ON_ZERO(NCAPINTS != 19)
+
+#endif /* _ASM_X86_REQUIRED_FEATURES_H */

tools/arch/x86/lib/memcpy_64.S

@@ -0,0 +1,179 @@
+/* Copyright 2002 Andi Kleen */
+
+#include <linux/linkage.h>
+#include <asm/cpufeatures.h>
+#include <asm/alternative-asm.h>
+
+/*
+ * We build a jump to memcpy_orig by default which gets NOPped out on
+ * the majority of x86 CPUs which set REP_GOOD. In addition, CPUs which
+ * have the enhanced REP MOVSB/STOSB feature (ERMS), change those NOPs
+ * to a jmp to memcpy_erms which does the REP; MOVSB mem copy.
+ */
+
+.weak memcpy
+
+/*
+ * memcpy - Copy a memory block.
+ *
+ * Input:
+ *  rdi destination
+ *  rsi source
+ *  rdx count
+ *
+ * Output:
+ * rax original destination
+ */
+ENTRY(__memcpy)
+ENTRY(memcpy)
+	ALTERNATIVE_2 "jmp memcpy_orig", "", X86_FEATURE_REP_GOOD, \
+		      "jmp memcpy_erms", X86_FEATURE_ERMS
+
+	movq %rdi, %rax
+	movq %rdx, %rcx
+	shrq $3, %rcx
+	andl $7, %edx
+	rep movsq
+	movl %edx, %ecx
+	rep movsb
+	ret
+ENDPROC(memcpy)
+ENDPROC(__memcpy)
+
+/*
+ * memcpy_erms() - enhanced fast string memcpy. This is faster and
+ * simpler than memcpy. Use memcpy_erms when possible.
+ */
+ENTRY(memcpy_erms)
+	movq %rdi, %rax
+	movq %rdx, %rcx
+	rep movsb
+	ret
+ENDPROC(memcpy_erms)
+
+ENTRY(memcpy_orig)
+	movq %rdi, %rax
+
+	cmpq $0x20, %rdx
+	jb .Lhandle_tail
+
+	/*
+	 * We check whether memory false dependence could occur,
+	 * then jump to corresponding copy mode.
+	 */
+	cmp  %dil, %sil
+	jl .Lcopy_backward
+	subq $0x20, %rdx
+.Lcopy_forward_loop:
+	subq $0x20,	%rdx
+
+	/*
+	 * Move in blocks of 4x8 bytes:
+	 */
+	movq 0*8(%rsi),	%r8
+	movq 1*8(%rsi),	%r9
+	movq 2*8(%rsi),	%r10
+	movq 3*8(%rsi),	%r11
+	leaq 4*8(%rsi),	%rsi
+
+	movq %r8,	0*8(%rdi)
+	movq %r9,	1*8(%rdi)
+	movq %r10,	2*8(%rdi)
+	movq %r11,	3*8(%rdi)
+	leaq 4*8(%rdi),	%rdi
+	jae  .Lcopy_forward_loop
+	addl $0x20,	%edx
+	jmp  .Lhandle_tail
+
+.Lcopy_backward:
+	/*
+	 * Calculate copy position to tail.
+	 */
+	addq %rdx,	%rsi
+	addq %rdx,	%rdi
+	subq $0x20,	%rdx
+	/*
+	 * At most 3 ALU operations in one cycle,
+	 * so append NOPS in the same 16 bytes trunk.
+	 */
+	.p2align 4
+.Lcopy_backward_loop:
+	subq $0x20,	%rdx
+	movq -1*8(%rsi),	%r8
+	movq -2*8(%rsi),	%r9
+	movq -3*8(%rsi),	%r10
+	movq -4*8(%rsi),	%r11
+	leaq -4*8(%rsi),	%rsi
+	movq %r8,		-1*8(%rdi)
+	movq %r9,		-2*8(%rdi)
+	movq %r10,		-3*8(%rdi)
+	movq %r11,		-4*8(%rdi)
+	leaq -4*8(%rdi),	%rdi
+	jae  .Lcopy_backward_loop
+
+	/*
+	 * Calculate copy position to head.
+	 */
+	addl $0x20,	%edx
+	subq %rdx,	%rsi
+	subq %rdx,	%rdi
+.Lhandle_tail:
+	cmpl $16,	%edx
+	jb   .Lless_16bytes
+
+	/*
+	 * Move data from 16 bytes to 31 bytes.
+	 */
+	movq 0*8(%rsi), %r8
+	movq 1*8(%rsi),	%r9
+	movq -2*8(%rsi, %rdx),	%r10
+	movq -1*8(%rsi, %rdx),	%r11
+	movq %r8,	0*8(%rdi)
+	movq %r9,	1*8(%rdi)
+	movq %r10,	-2*8(%rdi, %rdx)
+	movq %r11,	-1*8(%rdi, %rdx)
+	retq
+	.p2align 4
+.Lless_16bytes:
+	cmpl $8,	%edx
+	jb   .Lless_8bytes
+	/*
+	 * Move data from 8 bytes to 15 bytes.
+	 */
+	movq 0*8(%rsi),	%r8
+	movq -1*8(%rsi, %rdx),	%r9
+	movq %r8,	0*8(%rdi)
+	movq %r9,	-1*8(%rdi, %rdx)
+	retq
+	.p2align 4
+.Lless_8bytes:
+	cmpl $4,	%edx
+	jb   .Lless_3bytes
+
+	/*
+	 * Move data from 4 bytes to 7 bytes.
+	 */
+	movl (%rsi), %ecx
+	movl -4(%rsi, %rdx), %r8d
+	movl %ecx, (%rdi)
+	movl %r8d, -4(%rdi, %rdx)
+	retq
+	.p2align 4
+.Lless_3bytes:
+	subl $1, %edx
+	jb .Lend
+	/*
+	 * Move data from 1 bytes to 3 bytes.
+	 */
+	movzbl (%rsi), %ecx
+	jz .Lstore_1byte
+	movzbq 1(%rsi), %r8
+	movzbq (%rsi, %rdx), %r9
+	movb %r8b, 1(%rdi)
+	movb %r9b, (%rdi, %rdx)
+.Lstore_1byte:
+	movb %cl, (%rdi)
+
+.Lend:
+	retq
+ENDPROC(memcpy_orig)