This is a simple patch that replaces movnti with movq and clwb.

It is suitable only for benchmarking (it doesn't handle alignment and
page faults well).

Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>

---
 arch/x86/lib/copy_user_64.S |   17 +++++++++--------
 arch/x86/lib/usercopy_64.c  |   25 +++++++++++++++++--------
 2 files changed, 26 insertions(+), 16 deletions(-)

Index: linux-nova/arch/x86/lib/copy_user_64.S
===================================================================
--- linux-nova.orig/arch/x86/lib/copy_user_64.S	2020-03-24 15:17:09.792758258 -0400
+++ linux-nova/arch/x86/lib/copy_user_64.S	2020-03-29 04:51:03.329260149 -0400
@@ -224,18 +224,19 @@ ENTRY(__copy_user_nocache)
 2:	movq 1*8(%rsi),%r9
 3:	movq 2*8(%rsi),%r10
 4:	movq 3*8(%rsi),%r11
-5:	movnti %r8,(%rdi)
-6:	movnti %r9,1*8(%rdi)
-7:	movnti %r10,2*8(%rdi)
-8:	movnti %r11,3*8(%rdi)
+5:	movq %r8,(%rdi)
+6:	movq %r9,1*8(%rdi)
+7:	movq %r10,2*8(%rdi)
+8:	movq %r11,3*8(%rdi)
 9:	movq 4*8(%rsi),%r8
 10:	movq 5*8(%rsi),%r9
 11:	movq 6*8(%rsi),%r10
 12:	movq 7*8(%rsi),%r11
-13:	movnti %r8,4*8(%rdi)
-14:	movnti %r9,5*8(%rdi)
-15:	movnti %r10,6*8(%rdi)
-16:	movnti %r11,7*8(%rdi)
+13:	movq %r8,4*8(%rdi)
+14:	movq %r9,5*8(%rdi)
+15:	movq %r10,6*8(%rdi)
+16:	movq %r11,7*8(%rdi)
+	clwb (%rdi)
 	leaq 64(%rsi),%rsi
 	leaq 64(%rdi),%rdi
 	decl %ecx
Index: linux-nova/arch/x86/lib/usercopy_64.c
===================================================================
--- linux-nova.orig/arch/x86/lib/usercopy_64.c	2020-03-24 15:17:09.794758318 -0400
+++ linux-nova/arch/x86/lib/usercopy_64.c	2020-03-29 04:51:16.841669119 -0400
@@ -172,20 +172,29 @@ void __memcpy_flushcache(void *_dst, con
 	}
 
 	/* 4x8 movnti loop */
-	while (size >= 32) {
+	while (size >= 64) {
 		asm("movq    (%0), %%r8\n"
 		    "movq   8(%0), %%r9\n"
 		    "movq  16(%0), %%r10\n"
 		    "movq  24(%0), %%r11\n"
-		    "movnti  %%r8,   (%1)\n"
-		    "movnti  %%r9,  8(%1)\n"
-		    "movnti %%r10, 16(%1)\n"
-		    "movnti %%r11, 24(%1)\n"
+		    "movq    %%r8,   (%1)\n"
+		    "movq    %%r9,  8(%1)\n"
+		    "movq   %%r10, 16(%1)\n"
+		    "movq   %%r11, 24(%1)\n"
+		    "movq  32(%0), %%r8\n"
+		    "movq  40(%0), %%r9\n"
+		    "movq  48(%0), %%r10\n"
+		    "movq  56(%0), %%r11\n"
+		    "movq    %%r8, 32(%1)\n"
+		    "movq    %%r9, 40(%1)\n"
+		    "movq   %%r10, 48(%1)\n"
+		    "movq   %%r11, 56(%1)\n"
+		    "clwb    (%1)\n"
 		    :: "r" (source), "r" (dest)
 		    : "memory", "r8", "r9", "r10", "r11");
-		dest += 32;
-		source += 32;
-		size -= 32;
+		dest += 64;
+		source += 64;
+		size -= 64;
 	}
 
 	/* 1x8 movnti loop */